Two Way-Constructions in Dutch: Motion Along a

TWO WAY-CONSTRUCTIONS IN DUTCH: MOTION ALONG A PATH AND TRANSITION TO A LOCATION A thesis submitted in partial fulfilment of the requirements for the Degree of Master of Arts in Linguistics in the University of Canterbury by Marie-Elaine van Egmond University of Canterbury July 2006 ii CONTENTS Acknowledgements ................................................................................................ vii Abstract ................................................................................................................... viii Chapter 1 Introduction ...................................................................................................... 1 1.1 Outline of the thesis ........................................................................................... 2 1.2 The way-construction in English ....................................................................... 4 1.2.1 Semantics of the way-construction ..................................................... 5 1.2.2 The verb .............................................................................................. 6 1.2.3 The way NP ......................................................................................... 9 1.2.4 The PP ................................................................................................. 11 1.2.5 The structure of the way-construction ................................................ 12 1.2.6 The conceptual structure of the way-construction .............................. 12 1.2.7 Two constructional analyses ............................................................... 14 1.2.8 Decompositional analysis ................................................................... 16 1.2.9 Relation to the ‘fake’ object resultative construction ......................... 16 1.2.10 Summary ........................................................................................... 17 1.3 The weg-construction in Dutch .......................................................................... 18 1.3.1 History of the weg-construction .......................................................... 20 1.3.2 Relation to the ‘fake’ object resultative .............................................. 22 1.3.3 The ‘zich-verplaatsings-constructie’ in Dutch .................................... 22 1.4 Terminology ....................................................................................................... 23 Chapter 2 Data collection .................................................................................................. 27 Chapter 3 The weg-construction in Dutch ....................................................................... 33 3.1 Introduction ........................................................................................................ 33 3.2 The verb ............................................................................................................. 35 3.2.1 Constraints on the verb ....................................................................... 42 iii 3.2.2 Unaccusative verbs ............................................................................. 49 3.2.3 Subjective motion ............................................................................... 53 3.2.4 Means and manner interpretation ........................................................ 55 3.2.5 Summary ............................................................................................. 56 3.3 The reflexive element ........................................................................................ 57 3.3.1 Characteristics of the reflexive ........................................................... 57 3.3.2 The reflexive is not a semantic argument of the verb ......................... 58 3.3.3 The ditransitive pattern in Dutch ........................................................ 61 3.3.4 Summary ............................................................................................. 63 3.4 The weg NP ........................................................................................................ 63 3.4.1 Characteristics of the weg NP ............................................................. 64 3.4.2 The weg NP is not a semantic argument of the verb .......................... 64 3.4.3 Adjectival modifiers on the weg NP ................................................... 65 3.4.4 Summary ............................................................................................. 68 3.5 The PP ................................................................................................................ 68 3.5.1 Path prepositions ................................................................................. 69 3.6 Relation to ‘fake’ object resultative construction .............................................. 71 3.7 The structure of the weg-construction ................................................................ 72 3.7.1 A Minimalist approach ....................................................................... 75 3.7.2 Conceptual structure of the weg-construction ..................................... 80 3.8 The weg-construction as a constructional idiom ................................................ 81 3.9 Some of Verhagen’s weg-construction verbs .................................................... 83 3.9.1 Zoeken ‘search’ and vinden ‘find’ ....................................................... 83 3.9.2 Verschaffen ‘provide’ .......................................................................... 86 3.10 Summary .......................................................................................................... 88 Chapter 4 The Transition to Location Construction in Dutch ...................................... 90 4.1 Introduction ........................................................................................................ 90 4.1.1 The verb .............................................................................................. 91 4.1.2 The reflexive element ......................................................................... 96 4.1.3 The PP ................................................................................................. 99 4.1.4 Summary ............................................................................................. 99 4.2 Semantics of the TLC ........................................................................................ 100 iv 4.2.1 Temporal relation between the subevents ........................................... 102 4.2.2 Gradualness ......................................................................................... 103 4.2.3 Telicity ................................................................................................ 104 4.2.4 Direct predication of the PP of the subject ......................................... 104 4.2.5 Means and manner interpretation ........................................................ 107 4.2.6 Manner of motion verbs ...................................................................... 108 4.2.7 Conclusions ......................................................................................... 109 4.3 An event structure account ................................................................................. 109 4.3.1 The event structure of the English resultative construction ................ 110 4.3.2 Simple and complex events ................................................................ 111 4.3.3 Relation to ‘fake’ reflexive resultative construction ........................... 114 4.3.4 Conceptual structure of TLC .............................................................. 116 4.3.5 Summary ............................................................................................. 118 4.4 The structure of the TLC .................................................................................... 119 4.4.1 A Minimalist approach ....................................................................... 120 4.5 Discussion .......................................................................................................... 124 4.5.1 TLCs that imply motion ...................................................................... 124 4.5.2 Reflexive for pragmatic reasons ......................................................... 128 4.5.3 Verbs that can take a weak reflexive as a thematic object .................. 131 4.5.4 Verhagen’s reflexive verbs ................................................................. 135 4.6 Summary ............................................................................................................ 137 Chapter 5 Discussion and conclusions ............................................................................. 139 5.1 Overview of the weg-construction and the TLC ................................................ 139 5.2 Apparent overlap in meaning of the weg-construction and the TLC ................. 143 5.2.1 TLCs with a PP that can be interpreted as the end of a path .............. 144 5.2.2 TLCs with manner of motion verbs .................................................... 145 5.3 The English way-construction is ambiguous ..................................................... 148 5.3.1 Temporal relation between subevents ................................................. 148 5.3.2 Repeated action ................................................................................... 150 5.3.3 Telicity ................................................................................................ 151 5.3.4 Direct predication of the PP over the subject ...................................... 152 5.3.5 Means and manner interpretation ........................................................ 153 v 5.3.6 The two meanings of the English way-construction ........................... 154 5.3.7 Similarities between the way-construction and the ‘fake’ object resultative ............................................................................................ 155 5.3.8 Differences between the way-construction and the ‘fake’ object resultative ............................................................................................ 157 5.3.9 A Minimalist approach ....................................................................... 157 5.3.10 Constructional idiom or decompositional analysis? ......................... 159 5.3.11 Summary ........................................................................................... 160 5.5 Way-constructions in other Germanic languages ............................................... 161 5.6 Conclusions ........................................................................................................ 163 5.7 Issues for further research .................................................................................. 165 References ............................................................................................................... 166 Appendices .............................................................................................................. 170 Appendix 1 First Dutch questionnaire ............................................................... 170 Appendix 2 Second Dutch questionnaire ........................................................... 173 Appendix 3 Third Dutch questionnaire .............................................................. 177 Appendix 4 English questionnaire ..................................................................... 179 Appendix 5 Verbs found in the weg-construction with Google ........................ 181 Appendix 6 Verbs found in the TLC with Google ............................................ 184 vi LIST OF TABLES 2.1 Search algorithms used to search for weg-constructions in the INL corpus ...... 28 2.2 Search algorithms used to search for TLCs in the INL corpus .......................... 28 2.3 Search algorithms used to search for weg-constructions with one to six variables intervening between the fixed elements in the INL corpus .......... 29 2.4 Search algorithms used to search for TLCs with 1 to 3 variables intervening at a given place in the string ........................................................... 30 3.1 Verbs found in the weg-construction (INL corpus) ........................................... 35 3.2 Prepositions heading the PP of the weg-construction (INL corpus) .................. 69 4.1 Prepositions heading the PP of the TLC (INL corpus) ...................................... 99 5.1 Semantic differences between the weg-construction and the TLC .................... 140 vii ACKNOWLEDGEMENTS I am happy to have the opportunity here to thank a number of people that have been of great help in completing this thesis. First and foremost, I want to thank Ida Toivonen and Kate Kearns, my supervisors. Ida’s critical and challenging comments, and Kate’s helpful and optimistic comments have been invaluable. I want to thank no less my friends, family and all the other people who filled in those numerous questionnaires (and who have been doing so for years!). Without them this research would not have been possible. Also thanks to those who supplied me with data that they picked up from the Dutch media. I especially want to express my gratitude to Heidi Quinn, for giving me a crash course on Minimalism, for her continuous availability, and for her infectious enthusiasm and passion for syntax. I have had a fantastic time in New Zealand, which, besides studying Linguistics at the University of Canterbury, is largely due to the wonderful people I met here. So cheers to you all! However, none of this would have been so fantastic without Henrik. I thank him for his inexhaustible support, optimism, encouragement and …. TLC. viii ABSTRACT This thesis introduces a Dutch construction, called the Transition to Location Construction (TLC). The TLC is parallel to the way-construction in English. A second Dutch equivalent of the way-construction, called the weg-construction, is investigated in this thesis as well. The two Dutch constructions have a different meaning and syntax: the weg-construction is ditransitive and denotes motion along a path, whereas the TLC is transitive and denotes a transition to a location, which does not involve the traversal of a path. This thesis gives a detailed description of both Dutch constructions and demonstrates that they represent a mismatch in the syntax-semantics mapping: the verb has two syntactic complements, but these are not semantic argument of the verb. Moreover, the syntactic head of the sentence is not the semantic head, because the main verb is subordinate to a GO or CAUSE function. Both constructions are very productive and should therefore be taken seriously by any theory of syntax. The Minimalist Program currently does not incorporate constructions. In the Minimalist account offered here, several additional assumptions are made to account for the fact that it is not the verb that determines the complement configuration of the wegconstruction and the TLC. In the literature, the English way-construction is considered to denote motion along a path. Based on the two Dutch constructions investigated here, the wayconstruction will be shown to be in fact ambiguous between a motion along a path reading and a transition to a location reading. Furthermore, this path/transition distinction is present in other Germanic languages as well, which has previously not been recognised. 1 Chapter 1 Introduction This thesis investigates two Dutch equivalents of the English way-construction. One of these constructions is the weg-construction, which has previously been described by Verhagen (2002, 2003, 2004). The second Dutch way-construction is what I call the Transition to Location Construction, or TLC. The way-construction in English has received considerable attention in the literature (Jackendoff 1990; Marantz 1992; Goldberg 1995, Levin & Rappaport Hovav 1995), because it is a, as Jackendoff puts it, “fairly outrageous example of mismatch between syntax and semantics” (1990:218). That is, the way-construction violates the argument structure of the verb: it has two syntactic complements, a way NP and a prepositional phrase (PP), but neither of these is a semantic argument of the verb. Furthermore, the way-construction denotes motion along a path, but the verb does not have to be a motion verb. Thus, although the verb is the syntactic head of the construction, it does not seem to be the semantic head. The same peculiarities hold for the two Dutch constructions investigated here: both constructions contain two syntactic complements that are not selected by the verb. For the weg-construction, these are a reflexive NP and een weg ‘a way’ NP, which includes a PP. The weg-construction thus instantiates the double object construction, but the ditransitive pattern is generally unproductive in Dutch. The TLC has a reflexive NP and a PP complement, which are also not subcategorised for by the verb. Like the English way-construction, the Dutch constructions represent a mismatch in the syntax-semantics mapping: the weg-construction and the TLC denote that the subject respectively goes and gets to be somewhere by means of the action denoted by the verb, but the verb does not have to be a motion verb. In this thesis, I will investigate the two Dutch constructions in detail. I will show that they both are very productive: a large variety of verbs can be used to denote the means by which a path is travelled or a location is reached, and their PPs can be headed by a range of spatial prepositions. However, I will argue that, although both Dutch constructions are translated with a way-construction in English, they in fact have a very 2 different meaning: the weg-construction denotes motion along a path, whereas the TLC denotes a transition to a location, which does not involve the traversal of a path. Consequently, the two constructions have a different event structure: the wegconstruction describes a simple event, which consists of two subevents that are temporally co-extensive and which take place at the same location. By contrast, the TLC represents a complex event, consisting of two distinct subevents that are not necessarily co-extensive, and which may also take place at different locations. Both constructions contain a weak reflexive, but this reflexive is present for different reasons. In the TLC, the reflexive is licensed because each distinct subevent in the event structure requires a separate argument XP in the syntax (cf. the Argument-per-Subevent Condition, Rappaport Hovav & Levin 1998 and the references therein). Since the weg-construction has a simple event structure, the reflexive is not present to provide a syntactic argument of a distinct subevent. Instead, the reflexive is an indirect object, which is a part of the construction. The two Dutch constructions also differ in their degree of compositionality. I will argue that the weg-construction is a constructional idiom, with a meaning that cannot be derived from its individual parts, and with a syntax that is highly unusual. The TLC, on the other hand, has a greater degree of compositionality: the PP of the TLC is a resultative predicate on the reflexive NP, so this construction is an instance of the more general resultative construction. These differences in meaning and compositionality of the two Dutch constructions may shed light on some of the controversial analyses that exist in the literature about the English way-construction. I propose that the English construction can be ambiguous between a motion along a path reading and a transition to a location reading. Furthermore, these two constructions also seem to exist in other Germanic languages, which previously have been collapsed into one construction in the literature. 1.1 Outline of the thesis This thesis is organised as follows. The remainder of this chapter first reviews the literature on the way-construction in English, where the similarities between the different analyses and the controversial issues will be emphasised. Then Verhagen’s (2002, 2003, 3 2004) account of the weg-construction in Dutch will be summarised, followed by a definition of the terms that are used in this thesis. Chapter 2 describes the method of data collection. Data were collected by means of corpus searches and questionnaires. Chapter 3 investigates the Dutch weg-construction. This construction entails motion along a path, by means of the action denoted by the verb. The constraints on the verb that can be found in the weg-construction will be investigated, and evidence will be provided that all elements of the construction are obligatorily present, but none of them besides the subject is a semantic argument of the verb. I will offer a Minimalist account of the weg-construction, where several additional assumptions had to be made in order to account for the fact that the verb in the construction has two NP complements that are not its own thematic arguments. Chapter 4 introduces the TLC. The similarities between the weg-construction and the TLC will be discussed first. The constraints on the verb that can be found in both constructions are the same, and the PPs are headed by the same range of spatial prepositions. I will provide evidence that neither of the two syntactic complements of the verb in the TLC is a semantic argument of the verb. Next, the difference in meaning between the two constructions will be discussed in detail. I will argue that the meaning of the TLC differs from the meaning of the weg-construction because it does not entail motion along a path. Instead, the TLC describes a stative location that is reached, which does not involve the traversal of a path, and consequently no motion either. The superficially identical prepositions of the two constructions belong to conceptually distinct categories, and the event structure of the constructions is different as well. The semantic difference between the two constructions is reflected in their syntax: apart from the absence of the een weg phrase, the reflexive NP and the PP occur in structurally different positions in the TLC than in the weg-construction. Chapter 5 discusses some apparent overlaps in the meaning of the wegconstruction, the TLC and simple motion along a path sentences. Subsequently, the English way-construction will be argued to have both the meaning of the wegconstruction and the TLC. The English construction can therefore be ambiguous between a motion along a path reading and a transition to a location reading. Furthermore, these 4 two different meanings also seem to exist in other Germanic languages, which have previously not been recognised. 1.2 The way-construction in English This section reviews and discusses the previous accounts of the way-construction in English (Jackendoff 1990, 2002; Marantz 1992; Goldberg 1995, 1996; Levin & Rappaport Hovav 1995). The English way-construction consists of a verb, a possessive way NP and a PP, as exemplified in (1). (1) a. Frank dug his way out of the prison (Goldberg 1995:199, 1) b. We ate our way across the US. (Jackendoff 1990:212, 8a) c. Bill belched his way out of the restaurant. (Jackendoff 1990:211, 1a) d. Sam joked his way into the meeting. (Jackendoff 1990:211, 1c) The examples in (1) denote that the subject referent moves along the path designated by the PP, by means of the action denoted by the verb. In (1a), Frank moved out of the prison along the path that he created by means of digging. In the example in (1b), the subject referents are moving across the US by means of (or maybe ‘while’) eating, and so on. However, none of the elements in the way-construction entails motion, as shown in (2) from Goldberg. The sentence in (2a) involves the way-construction in (1a), whereas in (2b) the noun way has been replaced by escape route. (2) a. *Frank dug his way out of prison, but he hasn’t gone yet. (1995:200, 2') b. Frank dug his escape route out of prison, but he hasn’t gone yet. (1995:199, 4') The unacceptability of the sentence in (2a) supports the hypothesis that the wayconstruction denotes motion. When way is replaced by another noun denoting a path, the sentence no longer denotes motion, as shown in (2b). Jackendoff (1990) observes that the verbs in (1) cannot normally appear with a PP complement, as shown in (3) for belch and joke. 5 (3) a. *Bill belched out of the restaurant. (1990:211, 2a) b. *Sam joked into the meeting. (1990:211, 2c) Furthermore, any thematic object of the verb is not allowed in the way-construction, as can be seen in (4). (4) a. *Frank dug an escape route his way out of prison. b. *We ate hot dogs our way across the US. (Jackendoff 1990:212, 9a) Jackendoff and Goldberg conclude that, although the way NP and the PP are obligatorily present in the way-construction, they are not semantic arguments of the verb. 1.2.1 Semantics of the way-construction Jackendoff suggests that there are two distinct paraphrases of the English wayconstruction: one in which the verb designates the means of motion and one in which the verb denotes an action that accompanies the motion. The examples in (5) give the two paraphrases of the way-construction in (1d) above (1990:214). (5) a. Sam got into the meeting by joking. b. Sam went into the meeting while joking. Goldberg refers to these different paraphrases as the means interpretation and the manner interpretation, respectively. She argues that the means interpretation is the primary one and finds the manner interpretation marginal, because not all speakers accept the manner interpretation in (5b). The means interpretation of the way-construction entails that a path is created by the subject referent, by means of removing obstacles or overcoming other barriers, as in the examples in (1) above. Consider also the following example. (6) Sally made her way into the ballroom. (Goldberg 1995:204, 13) 6 Goldberg argues that this sentence can only be understood that Sally moved through some obstacles, like a crowd. It cannot mean that she simply walked into an empty ballroom. “In fact, the most common interpretation of the way-construction involves motion through a crowd, mass, obstacle, or other difficulty - that is, there is some reason why a path needs to be created” (1995:204). When the path and the motion are metaphorical, the way-construction also implies difficulty or a metaphorical barrier, as in (7) from Goldberg. (7) a. ??Sally drank her way through a glass of lemonade. (1995:204, 14a) b. Sally drank her way through a case of vodka. (1995:204, 14b) Goldberg points out that the way-construction in (7a) is odd because there is not much difficulty to drinking a glass of lemonade. Drinking a case of vodka, on the other hand, does imply some difficulty and is therefore much better. The manner interpretation does not imply the creation of a path and consequently no external difficulty either. Consider the example in (8). (8) I knitted my way across the Atlantic. (Goldberg 1995:213, 57) The manner interpretation entails that the subject referent simply moves along a preestablished path, where the verb denotes the accompanying manner (1995:209). The next section discusses the individual elements of the way-construction. 1.2.2 The verb Jackendoff proposes that the verb in the way-construction is constrained by both semantic and syntactic considerations. A syntactic constraint is that the verb has to be used intransitively, as was shown in (4) above, where the presence of a thematic object of the verb is unacceptable. A semantic constraint is that the verb in the way-construction has to be “capable of being construed as a process”, which means that it must be a process verb, or else describe a repeatable bounded event (1990:213). This process constraint rules out 7 stative verbs, as also noted by Levin & Rappaport Hovav, henceforth L&RH, (1995). Consider the following examples. (9) a. *Sylvia is knowing her way to a first prize. (L&RH 1995:150, 38) b. *Jill remained her way to a ticket to the show. (L&RH 1995:150, 39) c. *Bill slept/blushed his way to New York. (Jackendoff 1990:213, 12c) The verbs in these examples are semantically incompatible with the way-construction, as they do not have any internal structure and hence cannot be construed as a process. The process constraint also rules out verbs that denote a bounded event that cannot be repeated, as shown in the following examples. (10) a. *The window opened/broke its way into the room. (Jackendoff 1990:213, 12) b. *The apples fell their way into the crates. (L&RH 1995:148, 32b) These examples are infelicitous because open, break and fall are bounded events that are not repeatable. Jackendoff points out that when the verb denotes a bounded event that is repeatable, the way-construction strongly implies a repetition of that event. Consequently, the following two examples have a slightly different meaning (1990:224). (11) a. Willy jumped into Harriet’s arms. b. Willy jumped his way into Harriet’s arms. Whereas the sentence in (11a) implies a single jump, the way-construction in (11b) strongly implies several jumps (see also L&RH 1995:200). Goldberg notes that the verb in the way-construction has to be conceivable as the means by which a path is created and/or travelled, by removing obstacles or overcoming other barriers. Therefore, the motion takes place despite external difficulty, which can account for the unacceptability of plain motion verbs in the construction, such as the following. 8 (12) *She went/walked/ran/moved/stepped her way to New York. (1995:205) Goldberg argues that these plain motion verbs do not imply that there is any difficulty involved, which makes them semantically unacceptable in the construction. Other semantic constraints to the verb that can occur in the way-construction are that the motion must be self-propelled and directed (Goldberg 1995:212-4). This can account for the ungrammatical examples in respectively (13a) and (13b). (13) a. *The butter melted its way off the turkey. (1995:212, 48) b. *She wandered her way over the field. (1995:214, 58) The example in (13a) is unacceptable because the motion is not self-propelled, while (13b) is odd because wander is aimless and not directed. L&RH (1995) and Marantz (1992) argue for a different explanation for the unacceptability of the examples in (10) and (13a) above. They note that the verbs in these examples are unaccusative and propose that these verbs are incompatible with the wayconstruction. This is because unaccusative verbs are not able to assign Case to unsubcategorised objects like the way NP (cf. Burzio’s generalization, Burzio 1986) (1995:137). Even if an unaccusative verb satisfies the repeatable action or selfpropelledness constraint, it is still unacceptable in the way-construction, as illustrated in (14) from L&RH. (14) a. *Andrea appeared her way to fame. (1995:150, 40a) b. *She arrived her way to the front of the line. (1995:148, 32c) Goldberg (1995), on the other hand, argues that the constraints of the verb in the wayconstruction are not related to unaccusativity, because unaccusative verbs are attested in the data as well. She provides the following examples. (15) a. … Bull’s strategy of trying to grow its way out of its extensive computermarketing problems. (1995:213, 49) 9 b. The bank-debt restructuring is the centrepiece of Lomas Financial’s monthslong effort to shrink its way back to profitability … (1995:213, 50) Goldberg notes that the unaccusative verbs in these examples have an “agentive interpretation”, but she still assumes them to be unaccusative.1 She concludes that the relevant constraints are semantic, that is, the motion must be self-propelled (1995:213). According to Goldberg, this self-propelledness constraint is not related to unaccusativity. In sum, the choice of verb in the way-construction is constrained by both semantic and syntactic considerations. A syntactic constraint is that the verb has to be used intransitively. A semantic constraint is that the verb has to be a process verb, which excludes stative verbs. Jackendoff and Goldberg propose further semantic constraints, namely that the verb must denote a repeated action and the motion must be directed and self-propelled. L&RH account for the latter constraint by arguing that unaccusative verbs are unacceptable in the way-construction. 1.2.3 The way NP Jackendoff argues that the way NP in the way-construction is not a semantic argument of the verb, but merely a phrase that “happens to be in object position” (1990:216). This observation is supported by the fact that the construction cannot be passivized. Jackendoff gives the following ungrammatical passive of the way-construction in (1c) above. (16) *His way was belched out of the restaurant by Bill. (1990:216, 17) The fact that the direct object cannot be moved into subject position indicates that it is not a semantic argument of the verb.2 Rather, the way NP parallels the non-object the bucket in kick the bucket, which likewise cannot be passivized (1990:216). Jackendoff concludes that the way NP is a “meaningless syntactic marker”, which drops out of the 1 In chapter 3, I will argue that these verbs are in fact polysemous in English, that is, they can be both unaccusative and unergative. This means that their subject can be either an Agent or a Theme. 2 Jackendoff notes that another reason why the way-construction cannot be passivized is because it contains a bound pronoun, which cannot be passivized. 10 interpretation of the sentence (see also section 1.2.6 below). He notes that adjectival modifiers on way either turn up as manner adverbs to the superordinate going event, or as absolutives. The way-construction in (17) then has the two paraphrases given in (18) (1990:217). (17) Sam joked his insidious way into the meeting. (18) a. Sam insidiously got into the meeting by joking. b. Sam, insidious (as ever), got into the meeting by joking. Jackendoff suggests that, although the adjective modifying way is in a syntactically subordinate position, it corresponds conceptually to a modifier of the superordinate going event. That is, the adjective does not modify the path. Goldberg, however, claims that adjectival modifiers on way do modify the path, based on the following examples (1995:206, 20). (19) a. ... the goats wending their familiar way across the graveyard. b. ... that he could make his own way to school. In these examples, it is the path that is familiar or own, not the going event or the subject.3 Goldberg argues that the way NP is a meaningful element in the construction, which denotes the path that is created (see also section 1.2.8 below). Marantz (1992) and L&RH (1995) argue that the way NP is parallel to the fake reflexive and the inalienably possessed body parts in ‘fake’ object resultative constructions like She yelled herself hoarse and He cried his eyes blind, respectively. Marantz argues that the way NP names a path, which “is the person named by the possessor of way extended through space” (1992:185). L&RH propose that the noun way 3 Note that Goldberg’s examples both involve verbs that occur idiomatically in the way-construction: wend in (19a) can only occur in the way-construction and make in (19b) is a special case because it was used in the way-construction almost three centuries before the construction was extended to include other verbs, and is the most frequent verb in the construction today (1995:206). Furthermore, wend and make do not denote the means by which a path is created (cf. *The goats went across the graveyard by wending and *He went to school by making). This will be discussed further in chapter 3. 11 is an inalienably possessed head, which is bound to the subject by the possessive pronoun (1995:198). Like Goldberg, Marantz argues that an adjectival modifier on way modifies the path and not the going event, which he illustrates with the following example. (20) He belched his silly way home. (1992:185, 12a) Marantz claims that silly modifies the path named by way, it does not modify the going: “silly describes the path of he, spread out spatially from some understood starting point to ‘home’ - he was silly while belching on his way home” (1992:185).4 To summarise, the various accounts differ as to whether the way NP is a meaningful element that plays a role in the semantics of the way-construction or not. Goldberg argues that way denotes the path that is created, so adjectival modifiers on the way NP modify the path. By contrast, Jackendoff assumes the way NP to be meaningless, so adjectival modifiers do not modify the path, but turn up as manner adverbs to the going event, or as absolutives. Marantz and L&RH assume the way NP to be exactly parallel to the fake reflexive in the resultative construction. According to Marantz, adjectival modification on this reflexive way NP modify the path, which is the person named by the possessor. 1.2.4 The PP The PP in the way-construction is necessarily a directional phrase. Jackendoff and Goldberg suggest that it denotes the path that is travelled, which can be both literal and metaphorical. Marantz and L&RH propose that the PP describes a goal that is reached, which is “typically an attained location” (L&RH 1995:198), or it may be a location that is completely traversed by the entity named by the way path (Marantz 1992:180). Marantz and L&RH both assume that the PP of the way-construction serves as a secondary resultative predicate on the way NP. 4 Since Marantz argues that the subject actually is the path, his paraphrase of (20) that “he was silly while belching on his way home” corresponds to the absolutive paraphrase proposed by Jackendoff (cf. He, silly, belched on his way home). 12 1.2.5 The structure of the way-construction All authors agree that the PP is a sister of the verb, rather than for instance a modifier of the way NP. Jackendoff states that the way NP and the PP do not form a constituent, based on the fact that an adverb can be inserted between them, as in (21a). The example in (21b) shows that an adverb cannot be inserted between the verb and the way NP, indicating that these do form a constituent. (21) a. Bill belched his way noisily out of the restaurant. (1990:212, 5a) b. *Bill belched noisily his way out of the restaurant. (1990:212, 6a) Jackendoff assigns the following structure to the way-construction, where the way NP occupies the direct object position of the verb, and the PP is a separate constituent in the VP (1990:212). (22) [SUBJi [VP V [NP POSSi way] [PP path]]] L&RH also assume the PP to be a sister of the verb, as they take the way-construction to be parallel to the fake object resultative. They assume resultative predicates to be VPinternal and attached to the lowest bar level. They do not take a stand, however, on whether the fake object in the resultative construction occurs in direct object position, or whether it is the subject of a small clause (1995:49). 1.2.6 The conceptual structure of the way-construction Jackendoff (1983) introduces an explicit theory of meaning, or conceptual structure, called Conceptual Semantics. Jackendoff (1990) uses the machinery of this theory to represent the meaning of the way-construction. I will use the same tools to communicate the meanings of the constructions that will be discussed in this thesis, so this machinery will be briefly described here. Jackendoff uses a repertoire of major conceptual categories or “semantic parts of speech”, which include amongst others Thing, Event, State, Place and Path. Consider the 13 correspondence between the syntactic structure in (23a) and the conceptual structure in (23b) (1990:45). (23) a. [S [NP John ][VP ran [PP into [NP the room]]]] b. [Event GO ([Thing JOHN], [Path TO ([Place IN ([Thing ROOM])])])] The conceptual categories Event, Thing, Place and Path can be elaborated into a function and its arguments; parentheses are used to indicate arguments of a function. For instance, the conceptual category Path can be elaborated as one of several functions that map a reference Thing or Place into a related trajectory. Some of these functions are: TO, FROM, TOWARD, AWAY-FROM and VIA. The room in (23) is an argument of the Place function IN, which in turn is an argument of the Path function TO. Jackendoff assumes theta-roles to be part of the level of conceptual structure, where they “are nothing but a particular structural configuration”. They are not part of the syntax (1990:47). For example, the theta-role Goal in (23) is a structural argument of the Path function TO, a Source would be a structural argument of the Path function FROM, and so on. The Event function GO in (23) indicates that the sentence expresses motion. The subject of the verb corresponds to the first argument of GO and the Path PP corresponds to the second argument. The subject and the PP are required by the lexical entry of the verb run (where the PP is optional). The way-construction represents a mismatch in the syntax-semantics mapping: the way NP and the PP complements are not required by the lexical entries of the verb. Moreover, the main verb is subordinate to a GO or GET function. Jackendoff (1990) proposes that the way-construction is a ‘constructional idiom’, that is, a specialized syntactic form with an idiomatic meaning. The meaning of this constructional idiom is roughly ‘traverse a path PP by/while doing V’. The two conceptual structures of the wayconstruction in (1d) above Sam joked his way into the meeting can be represented as follows. (24) a. [Event GET ([Thing SAM], [Path TO ([Place IN ([Thing MEETING])])]) [BY [Event JOKE ([Thing SAM])]]] 14 b. [Event GO ([Thing SAM], [Path TO ([Place IN ([Thing MEETING])])]) [WITH [Event JOKE ([Thing SAM])]]] The conceptual structure (CS) in (24a) represents the means interpretation in (5a) above: Sam got into the meeting by means of joking. The CS in (24b) gives the manner interpretation in (5b): Sam went into the meeting while joking. GET and GO are the main Events in the CS, which have two arguments: a Thing (the subject) and a Path. Furthermore, the way NP is not present in the interpretation of the way-construction, as Jackendoff assumes it to be meaningless. The main verb joke is demoted to a subordinate means or manner modifier in the CS of the sentence. 1.2.7 Two constructional analyses The way-construction as a constructional idiom can be represented as follows (Jackendoff 1990:221). (25) a. [VP Vh [NP POSSj way] PP k ] may correspond to b. GO ([α]j, [Path ]k) WITH/BY ACT ([X]αe) -BOUNDED h Event In this notation, ACT[X] means that the verb is an action verb. The -BOUNDED notation indicates that the verb has to be a process verb. The α is a notation for binding and indicates that the Agent of the subordinate Event ACT[X] and the Theme of the superordinate GO Event are the same individual. That is, the subject is both going and performing the action encoded in the verb. The Theme of the superordinate GO event is also coindexed j with the possessor in (25a), to ensure that the reflexive is bound to the subject. This rules out ungrammatical examples like *Bill belched Harry’s way into the room (1990:215). 15 Jackendoff proposes that (25) is a kind of lexical item (1990:222), since the form and the meaning of the VP are fixed and it also has a meaning of its own. That is, the VP consists of a verb, POSS way and a PP, which roughly has the meaning of ‘create and travel the PP path by means of doing V’ and which has to be stored in the lexicon as such. This VP takes four conceptual arguments: e (the external argument), j (the possessive way NP), k (the PP), and h (the main verb of the sentence). Goldberg (1995) advocates a slightly different meaning of the way-construction. She assumes that the way NP is a distinct and meaningful argument of the construction, which denotes the path that is created. The PP expresses the path that is travelled, as shown in Goldberg’s Construction Grammar analysis in (26).5 (26) Semantics: creator, | Syntax: [SUBJi create-move, created-way, | means/manner [V | path | [POSSi way] OBL]] The way NP and the PP, which are not semantic arguments of the verb, are assumed to be contributed by the construction. In this analysis, the way NP does not disappear from the interpretation of the sentence, as it is assumed to represent the created path. The main verb is not demoted to a subordinate means or manner modifier, but gets the function ‘create-move’. Goldberg argues that the hypothesis that way is an affected entity motivates the syntactic form of the construction, because the way NP is a direct object and effected entities are generally direct objects (1995:208). Given the semantics of the means interpretation, Goldberg hypothesises that the way-construction can be regarded as a “conventionalized amalgam” that combines the syntax and semantics of creation expressions like (27a), with intransitive motion expressions like (27b). (27) a. Sally made a path. (1995:207, 28) b. Sally moved into the ballroom. (1995:207, 29) 5 OBL stands for ‘oblique phrase’, which is the PP. 16 The resulting amalgam is a structure with three complements: the creator-theme (the subject), the created-way (the way phrase) and the path (the PP). 1.2.8 Decompositional analysis Marantz (1992) and L&RH (1995) propose that the meaning of the way-construction may be compositionally derived from its individual parts and the rules of syntax. This is because they assume that the way-construction is parallel to the fake object resultative construction, so the PP of the way-construction serves as a secondary resultative predicate on the reflexive. L&RH argue that the meaning of the resultative construction can be compositionally derived from the meaning of the verb plus the meaning of the XP. The only additional aspect of meaning that is not explicitly represented in the syntax is the causal relationship between the action described by the verb and the state expressed by the resultative XP. L&RH account for this causal relationship by assuming that the resultative XP is added at the lowest bar level within the VP. As a result, the XP is integrated into the core eventuality named by the verb. “The only type of eventuality with a state following any kind of process is an accomplishment” (1995:54). Since accomplishments always describe causative changes of state, they argue, an XP denoting a state that follows an activity verb can only be interpreted as denoting the result state of an accomplishment. The causal relation between the activity and the change of state thus “follows from the interpretation of the eventuality as an accomplishment” (1995:55). However, L&RH do not discuss how their analysis of the resultative construction can be extended to account for the meaning of the way-construction, which they presume denotes a change of location, and not a change of state. 1.2.9 Relation to the ‘fake’ object resultative construction As mentioned above, Marantz and L&RH assume the way-construction to be parallel to the fake object resultative. Goldberg recognises that the way-construction bears some similarities to this construction, for example that its syntactic complements are not semantic arguments of the verb. However, she points out several differences between the two constructions (1996:48-50). First, the way-construction can be used with a wide variety of verbs, whereas the verb in the resultative construction in general, and in fake 17 object resultatives in particular, is highly restricted. Second, the way-construction at least marginally allows a manner interpretation, whereas the resultative does not. Third, the way-construction implies the creation of a path by removing obstacles or involving other external difficulty, but the resultative does not have such implications. Finally, Dutch is a language that has fake object resultatives but does not have a way-construction, an observation she credits to Annie Zaenen. Goldberg concludes that because of these differences, the way-construction cannot be assimilated to the fake object resultative construction. Jackendoff (2002) assumes the way-construction to belong to the same family of constructional idioms as the resultative construction. For all members of this family, it is not the verb that determines the complement configuration, but rather the construction itself. The members of this family of constructional idioms have the syntactic structure [VP V NP PP/AP/particle] and vary as to which parts of the VP are open positions.6 In the way-construction, the verb and the PP are free variables and the NP is fixed (i.e. POSS way). The resultative construction is composed entirely of free variables (2002:175). However, the constructions have a different meaning: the way-construction means ‘traverse the path PP by/while doing V’, whereas the resultative means ‘cause NP to become AP by V-ing (with (it))’ (2002:176). Thus, the two constructions are not the same construction. 1.2.10 Summary The above review revealed that the different accounts of the English way-construction vary considerably. The authors disagree about whether the way NP is meaningful element in the construction that denotes the path that is created, or whether it is a meaningless syntactic marker. Alternatively, the way NP has been argued to be a reflexive path NP that is bound to the subject. Furthermore, it is disputed whether the way-construction is a construction, whose meaning cannot be determined from its individual parts, or whether its interpretation can be derived compositionally. The authors also have different opinions about the necessity to posit several different semantic constraints on the verb that can 6 Other members of this family are: V one’s head off, V one’s heart out, V up a storm and the ‘time’-awayconstruction V NP away (e.g. we’re twisting the night away) (Jackendoff 2002 chapter 6). 18 occur in the way-construction, or that these constraints can also be captured by a ban on unaccusative verbs. Finally, the analyses differ about the degree of similarity between the way-construction and the resultative construction. The way-construction has been argued to be reducible to the fake reflexive resultative construction, whereas others have identified the need to posit a distinct construction. None of the authors addresses the telicity of the way-construction. However, by arguing that the way-construction is a resultative, L&RH (1995) and Marantz (1992) implicitly state that the construction is necessarily telic. 1.3 The weg-construction in Dutch Verhagen (2002, 2003, 2004) describes the weg-construction in Dutch, in a reaction to Goldberg’s claim mentioned above that Dutch does not have a way-construction. The weg-construction consists of a verb, a reflexive, een weg ‘a way’ and a PP, as illustrated in (28) from Verhagen (2002). (28) a. Zo blufte zij zich een weg uit Auschwitz. (2002:410, 12) like.that bluffed she REFL a way out A. ‘That was how she bluffed her way out of Auschwitz.’ b. Twee bussen boren zich een weg naar het hart van Istanbul. (2002:410, 13) two buses drill REFL a way to the heart of I. ‘Two buses are drilling their way to the heart of Istanbul.’ c. De priesters wurmen zich een weg door de gelovigen. (2002:411, 14) the priests squeeze REFL a way through the faithful ‘The priests are squeezing their way through the faithful.’ Verhagen observes that the similarities with the English way-construction “are obvious”: there is a constant lexical element weg or way, a variety of verbs indicating the means by which a path is created, and a PP that specifies the path that is travelled (2002:411). Furthermore, the meaning of the weg-construction also entails motion despite difficulty, where a path is created by removing obstacles. Verhagen does not discuss the individual elements of the construction, but he focuses on the superficial similarities and differences 19 between the English and the Dutch construction. He also discusses the diachronic development of the constructions in both languages. Verhagen found 18 different verbs in the weg-construction in his corpus. These verbs denote the means by which a path is created and/or travelled and belong to the conceptual domain of ‘force-dynamics’ (2002:419). The verb banen is the most common verb, accounting for about half of the findings. Banen does not have a meaning of its own and only occurs idiomatically in the weg-construction (see section 1.3.1 below). Verhagen notes that the Dutch weg-construction is ditransitive, with the reflexive as the indirect, benefactive object and the een weg phrase as the direct object (2002:414). The Dutch and the English construction thus have a different syntax: the relationship between the subject and the created way is marked by a possessive determiner in English, whereas it is marked with a reflexive in indirect object position in Dutch. The Dutch construction furthermore differs from the English one in that the manner interpretation “does not exist at all” for the weg-construction (2002:416). The following example can only mean that the subject referent created a path by means of whistling, which is strange. (29) ??Hij zich een weg naar de voordeur. (2002:416, 21) floot he whistled REFL a way to the front.door ‘He whistled his way to the front door.’ Following Goldberg, Verhagen gives the following constructional analysis of the wegconstruction (2002:411, 16).7 (30) Sem: creator, | Syn: [SUBJi 7 create-move, | means [V for-self, | [REFLi created-way, path | | [een weg] OBL]] This is an exact copy from Verhagen. Note that one square bracket is missing at the end of the structure. Furthermore, the reflexive, een weg and the PP form a constituent in this structure. 20 In this analysis, the weg NP is a distinct meaningful element in the weg-construction, which denotes the path that is created. The PP denotes the path that is travelled, so according to Verhagen the construction in fact assigns four arguments to the verb: a creator, a beneficiary, a created path and a travelled path. Verhagen does not provide any evidence for the hypothesis that the weg NP is a distinct meaningful element of the construction, nor does he address the issue of how the verb can have three internal arguments. Interestingly, the syntax of the weg-construction is ditransitive, but this pattern is unproductive in modern Dutch (2002:415). The pattern can only be used with verbs of transfer like geven ‘give’, sturen ‘send’ and zenden ‘mail’, as in (31a). The ditransitive pattern is ungrammatical with a direct object that is created, as shown in (31b). Instead, the beneficiary has to be expressed in a PP, as in (31c). (31) a. Jan gaf haar een boterham. J. gave her a sandwich ‘Jan gave her a sandwich.’ b. *Jan maakte haar een boterham. (2002:415, 17) J. made her a sandwich ‘Jan made her a sandwich.’ c. Jan maakte een boterham voor haar. (2002:415, 19) J. made a sandwich for her ‘Jan made a sandwich for her.’ Verhagen concludes that the weg-construction is a productive instance of a generally unproductive pattern. He suggests an answer for the unusual syntax of the wegconstruction by looking at the diachronic development of the construction. 1.3.1 History of the weg-construction Verhagen (2002:422) notes that the ditransitive pattern was fully productive in Dutch in the 17th and 18th century. The verb banen was very common in the ditransitive pattern. This verb does not have a meaning of its own in modern Dutch and can only occur in the 21 weg-construction, but it used to mean ‘level out, flatten’. The combination of banen with weg as a direct object was especially frequent, which meant ‘to level out a road’. Banen + een weg in turn frequently occurred with a beneficiary, who could also be somebody else than the subject, as in the following examples from Verhagen (2002:422). (32) a. Turcken en Arabianen Sullen noyt goe weghen banen Voor den Christen. Turks and Arabs shall never good ways flatten for the Christian ‘Turks and Arabs will never pave good ways for the Christian.’ b. Koomt gy my een weg tot grooter droefheid baanen? cometh thou me a way to greater sorrow smooth ‘Are you coming to pave me a way to greater sorrow?’ c. … gy en moet u selven dare toe den wegh niet banen … thou NEG must you self.DAT there to the.ACC way not pave ‘… you must not pave yourself the way towards it …’ In the examples in (32a-b) the beneficiary is somebody other than the subject, but in (32c) the subject is the beneficiary. Verhagen hypothesises that reflexive expressions like (32c) were extended to also convey motion of the subject referent. He argues that this is a logical extension, because if you level out a path for yourself, you normally also travel it. Hence, something that was initially a pragmatic inference eventually came to be an institutionalised meaning (2004:343). The banen + reflexive + een weg pattern was subsequently extended to also include other verbs that could denote the means of levelling out a road, as well as to include a metaphorical path. In sum, Verhagen proposes that the weg-construction with its unusual ditransitive syntax originated in the 17th and 18th century from the verb banen in the ditransitive pattern, which at that time was fully productive. Banen is the most common verb in the weg-construction today, but it has lost its meaning in modern Dutch. This verb can only occur idiomatically in the weg-construction. 22 1.3.2 Relation to the ‘fake’ object resultative Verhagen argues that the weg-construction is “certainly not an instance of the resultative construction”, because it has ditransitive syntax, whereas the syntax of the resultative is transitive (2002:414). He claims that the weg-construction is not an instance of the ditransitive benefactive pattern either, because this pattern is not productive when the direct object is created. Verhagen concludes that the weg-construction constitutes a “kind of island in the whole of the grammar” (2002:415). Even though the weg-construction provides counterevidence to Goldberg’s claim that Dutch does not have a wayconstruction, Verhagen states that the Dutch construction in fact supports her claim that the English way-construction cannot be assimilated to the resultative. This is because the weg-construction differs significantly from the resultative construction in Dutch. This difference supports the idea that the two constructions cannot be reduced to the same pattern, and thus confirms the independent status of the way-construction in English. 1.3.3 The ‘zich-verplaatsings-constructie’ in Dutch Verhagen devotes two paragraphs to a distinct but related construction to the wegconstruction, which he calls the zich-verplaatsings-constructie ‘REFL-displacementconstruction’ (2004:341). He gives the following three examples of this construction (the paper is in Dutch, so the glosses are mine).8 (33) a. Hij worstelde zich door tal van wetenschappelijke werken. he wrestled REFL through number of scientific works ‘He wrestled his way through a number of scientific works.’ b. De jongens slepen zich the boys door de dode uren. drag REFL through the dead hours ‘The boys are dragging themselves through the dead hours.’ 8 I did not gloss all Verhagen’s examples with a way-construction in English, because I will argue in chapter 4 that the reflexive is used as a semantic argument of the verb in (33b), and the reflexive in (33c) is present because the verb bewegen is inherently reflexive. 23 c. Hij bewoog zich door een geluidssluis naar de belendende tent. he moved REFL through a sound.gate to the neighbouring tent ‘He moved through the sound gate to the neighbouring tent.’ Verhagen notes that since the zich-verplaatsings-constructie is syntactically distinct from the weg-construction, because it is not ditransitive. However, the two constructions are semantically related. Verhagen notes that a semantic difference between the wegconstruction and the zich-verplaatsings-constructie is that the latter does not necessarily imply overcoming obstacles, as in (33b) and (33c). Some verbs can be found in both constructions, whereas others can only be found in the zich-verplaatsings-constructie but not in the weg-construction, and vice versa (2004:341). For example, bewegen in (33c) can only occur in the zich-verplaatsings-constructie and banen can only be found in the weg-construction. He concludes that although the two Dutch constructions are similar, they cannot be reduced to one pattern. 1.4 Terminology This section defines some of the terms that are used in this thesis. • ‘Strong’ and ‘weak’ reflexives A strong reflexive in Dutch is one that is marked with zelf ‘self’, and a weak reflexive is a bare reflexive without zelf. Reflexives that are semantic arguments of the verb are normally strong, as in (34). The reflexive in (34a) is a direct object, whereas the reflexive in (34b) is an indirect object. Weak reflexives normally occur with for instance reflexive verbs, as in (35). (34) a. Willem bewondert zich*(zelf). (Reinhart & Reuland 1993:690, 69) W. admires REFL.self ‘Willem admires himself.’ b. Willem geeft zich*(zelf) een boek. W. gives REFL self a book ‘Willem gives himself a book.’ 24 (35) Jan schaamt zich(*zelf). J. shames REFL.self ‘Jan is ashamed.’ • Agent and Theme In this thesis, an Agent is assumed to be an animate instigator or initiator of an event, who performs a volitional and controlled action. An Agent is included in the macro-role Actor, which is an initiator, instigator or causer of an event or state, and which can be animate or inanimate. A Theme is an entity that is in motion or that undergoes a change of state (cf. e.g. Gruber 1965; Jackendoff 1990; Dowty 1991). The subject of unergative verbs is typically an Agent/Actor, whereas the subject of unaccusative verbs is typically a Theme. • Path, goal and location Following Jackendoff (1990), I define a path as a trajectory that is travelled in physical or in metaphorical space, which may or may not have an endpoint. A path is dynamic. A location is a place in physical or metaphorical space, which is stative. A location can be predicated of the subject (e.g. John is in the kitchen), whereas a path cannot (cf. *John is into the kitchen). A location can also refer to the place where an action takes place, in which case it is a frame locative (e.g. John did his homework in the kitchen). A location that is the endpoint of a path is a Goal, and a location that is the beginning of a path is a Source. • The ditransitive pattern Following Verhagen, I will assume that the Dutch weg-construction is ditransitive. I define the ditransitive pattern as involving two NP complements, as illustrated in (36a). This definition excludes verbs that take complements other than an NP, such as zetten ‘put’ in (36b), which takes an NP and a PP complement. 25 (36) a. Jan geeft Marijke een kopje thee. J. gives M. a cup.DIM tea ‘Jan gives Marijke a cup of tea.’ b. Marijke zet het kopje op tafel. M. puts the cup.DIM on table ‘Marijke puts the cup on the table.’ In this thesis, only the double object construction in (36a) instantiates the ditransitive pattern. Instances like (36b), which contain only one NP complement, are considered to be transitive. • Unaccusativity in Dutch There is a variety of different accounts of unaccusativity in Dutch, which include semantic accounts (e.g. Zaenen 1993; Lieber & Baayen 1997), as well as syntactic accounts (e.g. Hoekstra 1988). What all accounts agree on, however, is that auxiliary selection is a diagnostic for unaccusativity in Dutch: unergative verbs select unaccusative verbs select BE. HAVE, and Therefore, I will use auxiliary selection as a test to determine whether a verb is unaccusative or unergative in this thesis. Some verbs have an unergative and an unaccusative variant. For instance, manner of motion verbs are normally unergative and select HAVE, as in (37a). However, when combined with a directed path phrase, a manner of motion verb is unaccusative and selects BE, as in (37b). (37) a. Paul heeft (in het bos) gerend. P. has in the forest run ‘Paul ran (in the forest). b. Paul is het bos in gerend. P. is the forest in run ‘Paul ran into the forest.’ 26 These examples show that a directed path is realised as a postpositional phrase in Dutch. The prepositional phrase in (37a) is a frame locative that denotes the location of the running, which can thus also be omitted. This sentence means that Paul was running around inside the forest, which is atelic. The example in (37b) with postpositional in means that Paul ran along the path that leads into the forest, where the endpoint of the path was reached. Therefore, this sentence is telic. The postpositional phrase in (37b) is a complement of the verb, whereas the prepositional phrase in (37a) is an adjunct. 27 Chapter 2 Data collection Data were collected in this thesis by means of corpus searches, internet searches and questionnaires. The corpus that was used is the Instituut voor Nederlandse Lexicologie (INL) corpus (www.inl.nl), which consists of 38 million words. This corpus allows search algorithms where a specific number of variables can be inserted at any place in the string. For example, one can search for a “zich [1-6 words] een weg” string, where up to six words can intervene between zich and een weg in the weg-construction. However, whereas a search for instances of the weg-construction is relatively straightforward because of the fixed element een weg, searching for instances of the TLC is not so simple. There are many V-REFL-PP strings in Dutch that are unrelated to the TLC, so a search for this string in the corpus results in many hits, a large part of which do not instantiate this construction. The internet searches were performed using Google (www.google.com). Google does not have the option to search for strings where a number of variables can intervene, which makes it difficult to construct searches for the weg-construction, apart from searching for simple connected strings such as “zich een weg”. Furthermore, since the TLC does not have such a fixed element, it is only possible to search for specific verbs and reflexives with Google (such as “schopt zich” ‘kicks REFL’, “rent zich” ‘runs REFL’, and so on). Table 2.1 displays the strings that were used to search for instances of the wegconstruction in the INL corpus. The table also includes the exact algorithms used. Note that the first person singular reflexive form can be both me and mij; the second person singular reflexive is the same as the second person plural (je) and the third person singular and plural reflexives are also identical (zich). 28 Table 2.1 Search algorithms used to search for weg-constructions in the INL corpus pers, num string search algorithm 1st sg “me een weg” [lemma=‘me’][lemma=‘een’][lemma=‘weg’] 1st sg “mij een weg” [lemma=‘mij’][lemma=‘een’][lemma=‘weg’] 2nd sg/pl “je een weg” [lemma=‘je’][lemma=‘een’][lemma=‘weg’] 3rd sg/pl “zich een weg” [lemma=‘zich’][lemma=‘een’][lemma=‘weg’] 1st pl “ons een weg” [lemma=‘ons’][lemma=‘een’][lemma=‘weg’] Table 2.2 displays the strings that were used to search for instances of the TLC in the INL corpus, as well as the corresponding algorithms. To investigate whether the PP is necessarily realised as a prepositional phrase, I also included the option for a postpositional phrase. Table 2.2 Search algorithms used to search for TLCs in the INL corpus String Search algorithm verb-REFL-preposition-NP [ws=‘w’][lemma=‘zich’][ws=‘v’]<‘NP’> verb-REFL-NP-postposition [ws=‘w’][lemma=‘zich’]<‘NP’>[ws=‘v’] These search algorithms yielded numerous findings, only some of which instantiate the TLC; the majority are unrelated to this construction. Consider the following examples from the INL corpus. (1) a. Gemke speelde zich in de dubbele cijfers … G. played REFL in the double numbers ‘Gemke played her way into the double numbers.’ (i.e. into debts) b. … die instelling vertaalde zich zondag in een uitstekende wedstrijdmentaliteit. that attitude translated REFL Sunday in a perfect contest.mentality ‘That attitude translated itself into a perfect contest mentality on Sunday.’ 29 c. Zo stortten de inwoners van Echten zich zaterdag in de dorpsloop. like.that threw the inhabitants of E. REFL Saturday in the village.run ‘That was how the inhabitants of Echten threw themselves into the village run on Saturday.’ Only the example in (2a) instantiates the TLC, because the reflexive in this sentence is not a semantic argument of the verb. By contrast, the reflexive in (2b) belongs to the reflexive verb REFL vertalen ‘translate, transform’, and the reflexive in (2c) is a semantic argument of the verb. For the weg-construction, the possibility of a specific number of variables to intervene at a given place in the string was included as well. Table 2.3 shows the search algorithms that were used to search for weg-constructions with one to six variables between een and weg, or between the reflexive and een weg. This is exemplified in table 2.3 for the third person sg/pl only. Table 2.3 Search algorithms used to search for weg-constructions with one to six variables intervening between the fixed elements in the INL corpus pers, num 3rd sg/pl string search algorithm “zich een…weg” [lemma=‘zich’][lemma=‘een’][?/1..6][lemma=‘weg’] “zich …een weg” [lemma=‘zich’][?/1..6][lemma=‘een’][lemma=‘weg’] For the TLC, the possibility of one to three variables to intervene at a given place in the string was also included. Table 2.4 shows the search algorithms that were used to search for TLCs with one to three variables between the reflexive and the PP, or between verb and the reflexive.1 1 Since I did not find any instances of the TLC with a postpositional phrase, I only searched for TLCs involving a prepositional phrase here. 30 Table 2.4 Search algorithms used to search for TLCs with 1 to 3 variables intervening at a given place in the string String Search algorithm verb-REFL-[…]-preposition-NP [ws=‘w’][lemma=‘zich’][?/1..3][ws=‘v’]<‘NP’> verb-[…]-REFL-preposition-NP [ws=‘w’][?/1..3][lemma=‘zich’][ws=‘v’]<‘NP’> However, a corpus search alone is not sufficient for an investigation of the kind of verb that can occur in a construction. This is because a corpus is always limited and moreover, it does not reveal what is not grammatical in a language. To circumvent this problem, speakers’ judgements were obtained as well. This was done by means of three different questionnaires. The first questionnaire was designed to test the acceptability of unaccusative verbs in the weg-construction and the TLC. The following examples of a wegconstruction and a TLC with unaccusative smelten ‘melt’ were included in the first questionnaire. (2) a. De boter smolt zich een weg van de hete kalkoen. the butter melted REFL a way from the hot turkey ‘The butter melted its way off the hot turkey.’ b. De boter smolt zich van de hete kalkoen. the butter melted REFL off the hot turkey ‘The butter melted its way off the hot turkey.’ The first questionnaire also aimed at investigating the acceptability of weg-constructions and TLCs that only have a manner interpretation, instead of denoting the means by which the motion takes place or a location is achieved. The following examples were included in the first questionnaire. (3) a. Marianne floot M. zich een weg door de tunnel. whistled REFL a way through the tunnel ‘Marianne whistled her way through the tunnel.’ 31 b. Marianne floot M. zich door de tunnel. whistled REFL through the tunnel ‘Marianne whistled her way through the tunnel.’ The second questionnaire was designed to investigate the telicity of the weg-construction and the TLC. The participants were asked for their judgements regarding pairs of wegconstructions or TLCs, one of which included a durative phrase (e.g. minutenlang ‘for minutes’), and the other a non-durative phrase (e.g. in twee minuten ‘in two minutes’). This is illustrated with in (4) with a pair of TLCs from the second questionnaire. (4) a. Pieter van den Hoogenband heeft zich minutenlang in de finale gezwommen. P. van den H. has REFL minutes.long in the final swum ‘Pieter van den Hoogenband swam his way into the final for minutes.’ b. Pieter van den Hoogenband heeft zich in twee minuten in de finale gezwommen. P. van den H. has REFL in two minutes in the final swum ‘Pieter van den Hoogenband swam his way into the final in two minutes.’ The second questionnaire also aimed at testing the acceptability of manner of motion verbs in the weg-construction, which cannot refer to the means by which obstacles are removed, but instead denote the actual manner of moving. Consequently, the reflexive and the weg NP can also be omitted, which results in a slightly different meaning. The following examples were included in the second questionnaire. (5) a. De aanvaller slalomt zich behendig een weg langs zijn tegenstanders. the attacker slaloms REFL nimbly a way past his opponents ‘The attacker nimbly slaloms his way past his opponents.’ b. De aanvaller slalomt behendig langs zijn tegenstanders. the attacker slaloms nimbly past his opponents ‘The attacker nimbly slaloms past his opponents.’ 32 The participants were asked to give the difference in meaning between pairs like (5). The second questionnaire also investigated the meaning that is contributed by the reflexive in the TLC. Speakers were asked for the difference in meaning between members of several pairs of sentences, such as the following. (6) a. De kever knaagt zich door de bast. the beetle gnaws REFL through the bark ‘The beetle gnaws its way through the bark.’ b. De tor knaagt door de bast. the beetle gnaws through the bark ‘The beetle gnaws through the bark.’ Finally, a third questionnaire was designed to test some of the hypotheses that were developed in this thesis, regarding adjectival modification on the weg NP in the wegconstruction and the telicity of the TLC. The three questionnaires were especially designed to explore the semantic differences between the weg-construction and the TLC, which will be discussed in detail in chapter 4. In addition, an English questionnaire was designed to investigate the difference in meaning between the way-construction and the fake reflexive resultative in English, as well as the telicity of the way-construction. The instructions of the questionnaires were simple: the participants were asked to judge a number of sentences as ‘+’ (good), ‘-’ (bad) or ‘?’ (don’t know). The first questionnaire consisted of 27 sentences and was filled in by 22 speakers. The questionnaire and the results are given in appendix 1. The second questionnaire consisted of 32 sentences and was filled in by 32 speakers. The results are given in appendix 2. The third questionnaire and the results are given in appendix 3. This questionnaire consisted of 12 sentences and was filled in by 30 speakers. Appendix 4 contains the English questionnaire with its results. This questionnaire consisted of 30 sentences and was filled in by 31 speakers. 33 Chapter 3 The weg-construction in Dutch 3.1 Introduction This chapter describes and discusses the weg-construction in Dutch. This construction consists of a verb, a reflexive, een weg ‘a way’ and a prepositional phrase, as illustrated in (1).1 (1) a. Jan schopt zich een weg door de menigte. J. kicks REFL a way through the crowd ‘Jan is kicking his way through the crowd.’ b. De crimineel graaft zich een weg uit de gevangenis. the criminal digs REFL a way out the prison ‘The criminal digs his way out of the prison.’ These weg-construction examples denote motion: in (1a), Jan is moving through the crowd by means of kicking, and in (1b) the criminal is moving out of the prison by means of digging. However, the verb is not a motion verb, and a PP combined with a verb can normally only be interpreted as a frame locative, that is, it cannot describe a path. A directed motion path is normally realised as a postpositional phrase in Dutch (see chapter 1). Instead, the verb in the weg-construction denotes the means by which the subject referent creates a path, by removing obstacles or overcoming other barriers. For instance, in (1a), Jan is creating a path for himself through the crowd by kicking the people that are in his way, and he travels this path by continuously kicking. In the weg-construction in (1b), the criminal creates a path out of prison by removing soil by means of digging. Since obstacles have to be removed in order to move, the weg-construction implies motion despite external difficulty, which requires effort. The weg-construction violates the argument structure of the verb: it has two syntactic NP complements, namely a reflexive in the weg-construction in indirect 1 Although the noun weg only has the literal meaning of ‘road’ or ‘path’ in Dutch and does not have the metaphorical meaning that way has in English, I will gloss it with way in the weg-construction examples. This is to show the similarities between the Dutch and the English construction. 34 object position and a weg NP in direct object position. However, neither of these is a semantic argument of the verb. Moreover, the weg-construction is an instance of the double object construction, but this construction is very unproductive in Dutch. All the elements in the weg-construction are obligatorily present, and they are fixed. That is, the weg NP has to be indefinite and weg cannot be replaced by another noun denoting a path. The reflexive is obligatorily weak and the phrase following the weg NP has to be a prepositional phrase, it cannot be a postpositional phrase. The Dutch weg-construction is very similar to the English way-construction: both constructions denote motion along a path that is created and travelled by means of the action denoted by the verb, even though the verb does not have to be a motion verb. Both constructions have two syntactic complements that are not selected by the verb, one of which includes the fixed element weg or way. Moreover, both constructions are very productive: they allow a large variety of verbs, which have to obey certain constraints. In this chapter I will argue that the constraints are the same for the constructions in both languages. However, there are some differences between the Dutch and the English construction, which mainly are syntactic. The Dutch wegconstruction is syntactically ditransitive, whereas the English way-construction is transitive. Moreover, I will argue below that the PP in the weg-construction is an adjunct of the weg NP, whereas the PP of the way-construction is assumed to be a sister of the verb. In what follows, the individual elements of the weg-construction will be discussed, starting with the verb. Section 3.2 shows that a large variety of activity verbs can be found in the weg-construction, which are nevertheless subject to several constraints. Section 3.3 provides evidence that the reflexive in the weg-construction is not a semantic argument of the verb. Furthermore, the reflexive will be argued to have lost its meaning of beneficiary in the present-day weg-construction. Section 3.4 shows that the weg NP is not a semantic argument of the verb either. Based on adjectival modification of the weg NP, I will conclude that the weg NP is non-referential. Section 3.5 discusses the PP, which will be argued to specify the path, and to be headed by a Path preposition. The relation of the weg-construction to the ‘fake’ object resultative construction will be discussed in section 3.6, where I will conclude that these two constructions are not related. Section 3.7 then offers a Minimalist analysis of the structure of the weg-construction. Contra Verhagen (2002, 2003, 2004), I will argue that the PP is an adjunct to the weg NP. This constituent describes the path that 35 is created and travelled. I will propose that one argument is sufficient to denote both the path that is created and the path that is travelled, because these two events are in fact the same event. Section 3.8 proposes that the weg-construction is a ‘constructional idiom’, whose meaning cannot be compositionally derived from its individual parts or from rules of syntax. Therefore, this constructional idiom has to be learnt by speakers of Dutch. Finally, section 3.9 discusses some verbs that Verhagen includes in the set of verbs that can occur in the weg-construction, but I will argue that these patterns do not instantiate this construction. 3.2 The verb Only 12 different verbs were found in the weg-construction in the INL corpus using the search algorithms described in chapter 2. These verbs are listed in table 3.1. Table 3.1 Verbs found in the weg-construction (INL corpus) Verb Number of % of total / hits / verb verb banen ‘?’ 46 78.0 vechten ‘fight’, worstelen ‘wrestle’ 2 3.4 1 1.7 slaan ‘hit’, fluisteren ‘whisper’, rommelen ‘mess around’, boren ‘drill’, forceren ‘force’, dwingen ‘compel’, hakken ‘chop’, ploegen ‘plough’, puzzelen ‘solve puzzles’ Total number of hits 59 This result is comparable to that of Verhagen (2002), who found 18 different verbs in his corpus. Of the 12 verbs in table 3.1, five are unergative and six are normally transitive. The verb banen cannot be classified, because it only occurs idiomatically in the weg-construction. This verb is the most common verb in the weg-construction, but it does not have a meaning of its own. The following example gives a wegconstruction with the verb banen. 36 (2) Handenschuddend baande de Sint zich een weg naar het podium op het plein hands.shaking door baande de S. REFL a way to the stage on the square de dichte drommen van zijn jonge fans. (INL) through the dense crowds of his young fans ‘Shaking hands, Saint Nicolas made his way to the stage on the square through the dense crowds of his young fans.’ As mentioned in chapter 1, Verhagen (2002) shows that banen used to have a meaning of its own in old Dutch. He proposes that the weg-construction as it exists today may have originated from old Dutch with the verb banen in the ditransitive benefactive pattern (see also section 3.3.2 below). An internet search with Google provided numerous other verbs that can occur in the weg-construction. A search for the “zich een weg” string without any form of the verb banen resulted in about 15,500 hits. Altogether about 180 different verbs were found, which are listed in appendix 5. They are subdivided according to their valence: intransitive verbs (which are subdivided into unaccusative and unergative), verbs that can be both transitive and intransitive, and transitive verbs. It is important to keep in mind that this list is probably not an exhaustive one because a corpus search does not necessarily reveal all possible verbs in a particular construction. Moreover, since it is not possible with Google to include any number of variables in the “zich een weg” string, more verbs might have been found if one could search for strings like “zich […] een weg” or “zich een […] weg”. Verbs of any valence are admitted into the weg-construction. The verbs in the examples in (3) are unergative and those in (4) can normally be both transitive and intransitive. (3) a. Ook al brult en krijst Cobain zich een weg also even roars and screams C. REFL a door de chaos in zijn way through the chaos in his hersenpan … brainpan ‘Even if Cobain roars and screams his way through the chaos in his brainpan …’ www.goddeau.com/content/view/72 37 b. Samen met een klas van twintig 4 VWO’ers redeneert Keune zich een weg together with a class of twenty 4 VWO’ers reasons K. REFL a way naar de oplossing. to the solution ‘Together with a class of twenty 4th graders Keune reasons his way to the solution.’ www.ru.nl/.../vox-online/archief/zoeken_op_nummer c. … hoe de rest van de familie zich een weg kluift door de taaie kalkoen.2 how the rest of the family REFL a way gnaws through the tough turkey ‘… how the rest of the family is gnawing their way through the tough turkey.’ www.beatrijs.com/kerstviering.htm (4) a. Met volharding leest hij zich een weg door de paperassen van de with perseverance reads he REFL a way through the paperwork of the Utrechtse hoogleraar… U. professor ‘With perseverance, he reads his way through the paperwork of the professor from Utrecht …’ www.library.uu.nl/nieuws/archief/mededelingen/leeszaalbezoeker/29977_288.html b. Bruce Lee […] slaat, hakt, schopt en snijdt zich een weg door een eindeloze B. L. hits slashes kicks and cuts REFL a way through an endless stroom tegenstanders. stream opponents ‘Bruce Lee hits, slashes, kicks and cuts his way through an endless stream of opponents.’ www.kungfufilms.nl/Queen_boxer_recensie.htm The weg-construction can also contain verbs that are normally transitive, as illustrated in (5). 2 Although gnaw can be transitive in English, kluiven is intransitive in Dutch. 38 (5) a. Arabieren kopen zich een weg door Tanzania. Arabs buy REFL a way through T. ‘Arabs are buying their way through Tanzania.’ www.wereld-delen.nl/afrikanieuws/Tanzania.htm b. Op de Afsluitdijk was verkeer praktisch onmogelijk. Auto’s ploegden zich een on the A. weg was traffic practically impossible cars ploughed REFL a door een dertig centimeter dikke sneeuwlaag. (INL) way through a thirty centimetre thick snow.layer ‘Traffic was practically impossible on the Afsluitdijk. Cars were ploughing their way through a thirty centimetre thick layer of snow.’ These examples indicate that the verb in the weg-construction is used intransitively, because the Arabs in (5a) are not buying a road, nor are they buying themselves. Likewise, the cars in (5b) are not ploughing a road or themselves. The thematic object of these verbs has to be inferred from the context. The intransitivity constraint will be discussed in more detail in section 3.2.1 below. In all of the above examples, the subject referent is moving by means of performing the action denoted by the verb. The verb in the weg-construction often designates the means by which obstacles are removed in order to create a path, like the impact verbs slaan ‘hit’, hakken ‘slash’, schoppen ‘kick’ and snijden ‘cut’ in (4b) above and kluiven ‘gnaw’ in (3c). This path is then travelled by continuously removing the obstacles by means of the action denoted by the verb. In the wegconstruction in (3c), the people are progressing through the turkey by gnawing the meat away, and Bruce Lee in (4b) is moving through a stream of opponents by getting them out of his way by injuring them by hitting, slashing, and so on. The verb in the weg-construction can also denote the means by which barriers are overcome. For example, in (4a) above, the subject referent is overcoming the mental barrier put up by the paperwork by reading it. Some more examples of wegconstructions involving barriers that are overcome are given in (6). 39 (6) a. … de vrachtwagen vecht zich verbeten een weg over een grindspoor vol the truck fights REFL grimly a way over a gravel.track full onverwachte kuilen. (INL) unexpected pits ‘The truck grimly fights its way over a gravel track full of unexpected pits.’ b. In hoofdstuk 1 laat ik zien hoe iedereen zich een weg kan bluffen in de in chapter 1 let I see how everybody REFL a way can bluff in the wondere wereld van het beleggen. wondrous world of the investment ‘In chapter 1 I show how everybody can bluff his way in the wondrous world of investment.’ web.inter.nl.net/users/ii/intro_BewBel.htm c. Hij bluft, kliert, liegt en grapt zich een weg door televisieland. he bluffs nags lies and jokes REFL a way through television.land ‘He bluffs, nags, lies and jokes his way through television land.’ www.maestros.nl/scans/overigen/eddy/eddy-tvz.php All these examples imply effort: in (6a), the pits in the track constitute barriers that the truck tries to overcome by means of fighting. The examples in (6b-c) refer to social barriers that need to be overcome: in (6b), the subject referent teaches his students how to move around in the investment world by violating social constraints by means of bluffing. In the example in (6c), the subject referent overcomes the social barriers that he encounters in the television world by means of bluffing, nagging, and so on. The latter two examples show that the motion and the path can also be metaphorical, as the subjects are not literally moving. They could also progress along this metaphorical path without physically moving from their spots. In sum, the verb in the weg-construction designates the means by which a path is created and travelled. The path can be both literal and metaphorical. Since some obstacles need to be removed or barriers need to be overcome in order to move, the weg-construction implies motion despite external difficulty, which requires effort. Interestingly, in all of the above examples, the verb is not a motion verb. When the reflexive and the weg NP are omitted the sentences become ungrammatical, as shown in (7) for two of the examples above. 40 (7) a. *Bruce Lee slaat, hakt, schopt en snijdt door de tegenstanders. (cf. 4b) B. L. hits slashes kicks and cuts through the opponents b. *De vrachtwagen vecht over een grindspoor vol kuilen. (cf. 6c) the truck fights over a gravel.track full pits A PP combined with a verb can normally only be interpreted as a frame locative in Dutch. These examples are ungrammatical because the prepositional phrases are headed by the Path prepositions door ‘through’ and over ‘over’, which are incompatible with the non-motion verbs. Nevertheless, manner of motion verbs can also be found in the wegconstruction. These verbs do not denote the means by which a path is created, but only the means by which a path is travelled. Consider the following examples. (8) a. … kleine gangetjes waar het sperma zich een weg doorheen moet zwemmen. small tubes where the sperm REFL a way through.PRT must swim ‘… small tubes that the sperm has to swim its way through.’ home.tiscali.nl/kinderwens/baarmoederslijmvlies.html b. Ecodock roeit zich een weg naar de finish. E. rows REFL a way to the finish ‘Ecodock is rowing its way to the finish.’ www.stichtingtop.org/nl/infobase_foto.php?link_id=94 c. Chris de Witte […] slalomde zich een weg door C. de W. de Bredase defensie. slalomed REFL a way through the B. defence ‘Chris de Witte slalomed his way through the Breda defence.’ www.fctwente.nl/wedstrijden/resultaten/20012002/20012002-nac-twe.stm The subject referents in (8) are moving in the manner designated by the verb. In (8a), the sperm is moving by swimming, Ecodock in (8b) is moving by rowing, and so on. Since these verbs are motion verbs that denote the actual manner of moving, the reflexive and the weg NP can also be omitted. However, this results in a slightly different meaning. Compare the weg-constructions in (8) with the bare motion sentences in (9). 41 (9) a. … kleine gangetjes waar het sperma doorheen moet zwemmen. small tubes where the sperm through.PRT must swim ‘… small tubes where the sperm has to swim through.’ b. Ecodock roeit naar de finish. E. rows to the finish ‘Ecodock is rowing to the finish.’ c. Chris de Witte slalomde C. de W. door de Bredase defensie. slalomed through the B. defence ‘Chris de Witte slalomed through the Breda defence.’ Speakers interpret the weg-constructions in (8) to imply more difficulty than the bare motion instances in (9). For instance, the weg-construction in (8a) emphasises that the fallopian tubes that the sperm has to swim through are very long and narrow, which requires more effort than in (9a). Similarly, the water seems to be rougher and the path seems to be longer in the weg-construction in (8b) than in the bare motion sentence in (9b). Two sentences very similar to (8c) and (9c) were included in the third questionnaire and the participants were asked for the difference in meaning (see appendix 3). All but two of the 30 participants said that the weg-construction implies obstacles that have to be overcome, which requires more effort and possibly a longer path than the bare motion sentence. I conclude a weg-construction with a manner of motion verb implies more difficulty and a longer path than its bare motion counterpart. The two participants that did not accept the weg-construction similar to (8c) said that the construction is only acceptable with a verb that can be interpreted as the means by which obstacles are removed or barriers are overcome. That is, they only accept weg-constructions with a verb that denotes the means by which a path is created. They find manner of motion verbs unacceptable in the weg-construction because then the meaning does not involve the creation of a path. No instance of a weg-construction with a manner of motion verb was found in the INL corpus in table 3.1 (assuming that banen is not a motion verb), and these verbs are also less common in the Google findings in appendix 5: about 20% of the verbs found with Google are manner of motion verbs designating the actual manner of motion. In sum, when the verb in the weg-construction is a manner of motion verb, it denotes the means by which the path is travelled, which implies difficulty and effort. 42 In the majority of the cases, however, the verb designates the means by which a path is created and travelled, by removing obstacles or overcoming other barriers. 3.2.1 Constraints on the verb The previous section showed that a large variety of activity verbs can be found in the weg-construction, which denote the means by which is path is created and/or travelled. A test to determine whether a verb is an activity verb is to frame it in What X did was… (Jackendoff 1990). The verbs in all of the above examples fit in this frame (cf. What Bill did was kick/lie/row/hit/slash/bluff/…). As in the English wayconstruction, the verb in the weg-construction has to be “capable of being construed as a process” (Jackendoff 1990:213). This means that the verb has to have some internal structure: it has to be an activity verb, or else denote a repeatable bounded event. This process constraint rules out stative verbs. I did not find any instance of a weg-construction with a stative verb in the data, and a weg-construction with such a verb also sounds odd, as in shown in (10). (10) a. *John weet / lijkt / is zich een weg naar de top. J. knows seems is REFL a way to the top b. *Peter stinkt zich een weg uit het huis. P. stinks REFL a way out the house c. *Marieke bloost / slaapt zich een weg door de M. les. blushes sleeps REFL a way through the lecture The stative verbs in these examples are not conceivable as the means by which a path is created and/or travelled, so they are semantically odd in the weg-construction. Like stative verbs, plain motion verbs are not conceivable as the means by which a path is created, or as involving any difficulty to the motion. These verbs are therefore infelicitous in the weg-construction, as shown in (11a). The example in (11b) shows that a weg-construction that does not involve any difficulty in the motion is unacceptable as well. (11) a. *De man gaat zich een weg door het bos. the man goes REFL a way through the forest 43 b. *De bejaarden wandelden zich een weg the elderly door het park. strolled REFL a way through the park The sentence in (11b) was present in the third questionnaire and approved of by only one of the 30 speakers (see appendix 3). The motion verbs in (11) do not denote any means of or difficulty in the motion, and are therefore semantically odd. The process constraint also rules out verbs that denote a bounded event that cannot be repeated, as shown by the unacceptability of the following examples. (12) a. *De zanger stierf zich een weg naar roem. the singer died REFL a way to fame b. *Het raam breekt / opent zich een weg in de kamer. the window breaks opens REFL a way in the room Sterven ‘die’, breken ‘break’ and openen ‘open’ are bounded events that are not repeatable, which can account for their unacceptability in the weg-construction. As in the English way-construction, when the verb in the weg-construction denotes a repeatable bounded event, the construction strongly implies a repetition of that event. Consider the following examples. (13) a. Binnenkort zal Amerika zich een weg hebben gebombardeerd naar een soon will A. geheel nieuwe vrijehandelszone. completely new REFL a way have bombed to a free.trade.zone ‘Soon the Americans will have bombed their way to a completely new free trade zone.’ www.nologo.org/newsite/detaild.php?ID=201 b. … kan plotseling iedere ongeletterde pummel zich een weg klikken langs can suddenly every illiterate moron REFL a way click along duizenden webpagina's met tekst en beeld. thousands web.pages with text and image ‘… every illiterate moron is suddenly able to click his way along thousands of web pages with text and image.’ www.computable.nl/artikels/archief4/d05rr4yv.htm 44 Speakers interpret these examples to imply several bombs in (13a) and several clicks in (13b), respectively. Since the path is travelled by means of the action denoted by the verb, the weg-construction strongly implies a repetition of the event denoted by the verb. The semantic constraints posited by Goldberg that the verb in the wayconstruction has to denote a volitional and self-propelled action can also account for the unacceptability of the following weg-construction examples. (14) a. ??De athlete viel zich een weg naar de finish. the athlete.FEM fell REFL a way to the finish b. *De sneeuw smolt zich een weg door de lente. the snow melted REFL a way through the spring c. *De soep koelde zich een weg naar kamertemperatuur. the soup cooled REFL a way to room.temperature Although vallen ‘fall’ in (14a) is repeatable, this example is unacceptable because vallen is interpreted as not volitional. Similarly, although smelten ‘melt’ in (14b) and koelen ‘cool’ in (14c) are process verbs with some internal structure, they are unacceptable because they are not volitional or self-propelled. Finally, the motion must be directed and cannot be aimless, as also observed by Goldberg for the English construction. This semantic constraint can account for the unacceptability of the following weg-construction example. (15) *Het meisje dwaalde zich een weg over het veld. the girl wandered REFL a way over the field This sentence is infelicitous because dwalen ‘wander’ is aimless and thus not directed. The semantic constraints that the verb in the weg-construction needs to denote a volitional, self-propelled and directed action indicate that the subject must be an Agent. That is, it has to be an animate entity that performs a volitional and controlled action. When the subject is inanimate, it is personified and interpreted as an Agent, as in the following examples. 45 (16) a. … de vrachtwagen vecht zich verbeten een weg over een grindspoor vol the truck fights REFL grimly a way over a gravel.track full onverwachte kuilen. (= 6a) unexpected pits ‘The truck grimly fights its way over a gravel track full of unexpected pits.’ b. Een laatste kreun klauwde zich een weg a last door de rokerige kamer. groan clawed REFL a way through the smoky room ‘A last groan clawed its way through the smoky room.’ www.lectrr.be/column.php?ID=139&naam=La%20Vache%20Folle The subject referents in these examples are both inanimate, but they are both portrayed as Agents who perform a volitional and controlled action. The truck in (16a) is fighting and the groan in (16b) is clawing, which are activities that only animate entities can perform. The presence of the adverb verbeten ‘grimly’ in (16a) enhances the idea that the truck is portrayed as an animate entity. Another constraint on the verb in the weg-construction is that it has to be used intransitively. The presence of a semantic argument of the verb in direct object position is unacceptable, as shown by the following examples. (17) a. *De crimineel graaft zich een gang een weg uit de gevangenis. (cf. 1b) the criminal digs REFL a tunnel a way out the prison b. *Arabieren kopen zich bedrijven een weg door Tanzania. (cf. 5a) Arabs buy REFL businesses a way through T. c. *Amerika bombardeert zich Baghdad een weg naar een nieuwe A. bombs REFL B. a way to a new vrijehandelszone. (cf. 13a) free.trade.zone The presence of a semantic argument in direct object position is ungrammatical because the direct object position is occupied by the weg NP. However, the criminal is not digging a road in the weg-construction (1b), the Arabs are not buying a road in (5a) and America is not bombing a road in (13a). Therefore, the weg NP is not a semantic argument of the verb. The reflexive in indirect object position is not a semantic argument of the verb either, because the criminal is not digging himself, the 46 Arabs are not buying themselves and America is not bombing itself. This will be discussed in more detail in respectively section 3.4.2 and 3.3.2. The intransitivity constraint is confirmed by the fact that verbs that do not normally take NP and PP complements also occur in the weg-construction, as shown in (18a). The sentence in (18b) shows that the verb in (18a) is normally intransitive. (18) a. *Hij grapt (vele grappen) door televisieland. (cf. 6c) he jokes many jokes through television.land b. Hij grapt. he jokes ‘He is joking.’ The fact that intransitive verbs also occur in the weg-construction confirms that reflexive, the weg NP and the PP are not semantic arguments of the verb. The intransitivity requirement can account for the unacceptability of strictly (di)transitive verbs in the weg-construction. These verbs strictly require the presence of one or two thematic objects as their complements, as shown in (19a) for the strictly transitive verb verwoesten ‘destroy’ and in (19b) for the strictly ditransitive verb geven ‘give’. (19) a. Hij verslindt *(een berg boterhammen). he devours a pile sandwiches ‘He devours a pile of sandwiches.’ b. Hij geeft *(zichzelf) *(een boek). he gives REFL.self a book ‘He gives himself a book.’ Although strictly (di)transitive verbs can denote a volitional, self-propelled and repeatable action, the corpus searches did not reveal any instance of the wegconstruction with such a verb. A weg-construction with a strictly transitive or ditransitive verb also sounds strange, as shown in respectively (20a) and (20b). (20) a. *Hij verslindt zich een weg door de berg boterhammen. he devours REFL a way through the pile sandwiches 47 b. *Olivier geeft zich een weg naar de harten van de lezeressen. O. gives REFL a way to the hearts of the readers.FEM The example in (20a) is semantically odd because the only possible interpretation is one in which een weg is a semantic argument of the verb, so this sentence means that the subject is devouring een weg. Moreover, this sentence is ungrammatical because it is ditransitive, which, except for verbs of transfer and the weg-construction, is unproductive in Dutch. The weak reflexive and the weg NP in (20b) will also be interpreted as semantic arguments of the verb, so the only possible reading is that Olivier is giving himself a road. This is not only semantically odd, but also ungrammatical, because the reflexive indirect object of verbs of transfer needs to be marked with zelf ‘self’ (see chapter 1). I conclude that strictly transitive and strictly ditransitive verbs are incompatible with the weg-construction because they cannot be used intransitively. The indirect object and direct object of these verbs cannot be mapped onto the syntactic NP complements in the weg-construction, as these are not semantic arguments of the verb. The intransitivity constraint can also account for the unacceptability of reflexive verbs in the weg-construction. Verbs that inherently require a reflexive can be considered to be syntactically strictly transitive. The corpus searches revealed no instance of a weg-construction with a reflexive verb.3 A weg-construction with such a verb is unacceptable, as shown in (21). (21) a. *De dieven verstopten zich een weg naar de uitgang. the thieves hid REFL a way b. *Piet vergiste zich een weg door P. to the exit het lesuur. erred REFL a way through the class.hour However, these verbs are semantically compatible with the weg-construction, as they denote a self-propelled, repeatable and possibly volitional action. Moreover, one 3 I did find a couple of instances with a reflexive motion verb like REFL begeven ‘go’ and REFL verspreiden ‘spread’. However, these examples sound very odd: two weg-construction examples with zich begeven and zich verspreiden were included in the third questionnaire (see appendix 3). Of the 30 speakers, respectively 90% and 83% thought these sentences were ungrammatical. 48 could imagine a context in which the subject moves by hiding or making mistakes, as in the following attested English way-construction examples. (22) a. … we'd have to run, fight, and hide our way to the exit all the way!! www.sir-toby.com/extend-a-story/story-1/code/read.php?episode=42025 b. No matter how much I blunder and mistake my way through life, as long as I "believe" this and that doctrine then I'm assured of a place in paradise. www.figu.org/cgi-local/forum/us/board-profile.cgi?action=rate&topic=13&page=3454&post In the way-construction in (22a), the subject referents are moving towards the exit by repeatedly hiding, and in (22b) the subject referent is going through life by (or while) repeatedly making mistakes. These way-constructions show that, given the right context, somebody can move by means of (or ‘while’) hiding or making mistakes, suggesting that the Dutch verbs are semantically compatible with the wegconstruction. I propose that the Dutch reflexive verbs are unacceptable because they are strictly transitive, so they cannot fulfil the requirement of being used intransitively. To summarise, the verb in the weg-construction is used intransitively, which rules out verbs that strictly require the presence of one or two thematic objects, as well as verbs that are inherently reflexive. Semantic constraints on the verb in the weg-construction include that the verb has to be able to be construed as a process, which means that it has to be an activity verb or else denote a repeatable bounded event. The weg-construction then strongly implies a repetition of that event. Stative verbs and plain motion verbs are unacceptable because they cannot be conceived as the means by which a path is created and/or travelled. Furthermore, the action denoted by the verb has to be volitional, self-propelled and directed - in other words, the subject has to be an Agent. Jackendoff calls the intransitivity requirement on the verb in the wayconstruction a “syntactic requirement” (1990:212). However, I think this wording is rather confusing because the verb in the weg-construction is syntactically ditransitive, as it has two NP complements. Instead, the weg-construction rather represents a mismatch in the syntax-semantics mapping, because the semantic arguments of the verb (if any) cannot be mapped onto the syntactic arguments of the verb in the construction. This mismatch is illustrated by the unacceptability of strictly 49 (di)transitive verbs and reflexive verbs: the thematic object(s) of the strictly (di)transitive verbs cannot be mapped onto the NP complements of the wegconstruction. Likewise, the reflexive of a reflexive verb cannot be mapped onto the reflexive contributed by the weg-construction. 3.2.2 Unaccusative verbs This section investigates the acceptability of unaccusative verbs in the wegconstruction. As discussed in chapter 1, there is some debate in the literature about the acceptability of unaccusative verbs in the way-construction in English. For instance, Goldberg argues that the constraints on the verb involve the semantic constraints discussed in the previous section, rather than a having to do with unaccusativity, because she found unaccusative verbs in the data as well (1995:213). L&RH (1995), on the other hand, propose that the way-construction is a diagnostic for unergative verbs, so the construction does not allow unaccusative verbs. They assume that unaccusative verbs are incompatible with the way-construction because they cannot assign Case to unsubcategorised objects like the way NP (1995:137). The verbs in the ungrammatical examples in (12) and (14) above, with sterven ‘die’, smelten ‘melt’, breken ‘break’ and so on are all unaccusative (i.e., they select the BE-auxiliary). I found no instance of the weg-construction with an unaccusative verb in the data. To investigate the acceptability of unaccusative verbs in the construction, the following two sentences with unaccusative smelten ‘melt’ and groeien ‘grow’ were included in the first questionnaire. (23) a. De boter smolt zich een weg van de hete kalkoen. the butter melted REFL a way from the hot turkey ‘The butter is melting its way off the hot turkey.’ b. De zonnebloemen groeien zich een weg naar het licht. the sunflowers grow REFL a way to the light ‘The sunflowers are growing their way towards the light.’ Only one of the 22 participants approved of the weg-construction in (23a) with unaccusative smelten, but almost a third of the speakers accepted (23b) with groeien (see appendix 1). These results suggest that smelten is not, but groeien may be (marginally) acceptable in the weg-construction. 50 Interestingly, the evidence that Goldberg provides for her claim that unaccusative verbs can occur in the way-construction also involves the verb grow (as well as shrink, see section 1.2.2). However, she notes that the verbs in these examples have an “agentive interpretation”, yet she assumes them to be unaccusative. Grow is polysemous in English, that is, it can be both unaccusative and unergative (Kate Kearns, p.c.). This suggests that grow is unergative in Goldberg’s example, not unaccusative, as it is used agentively. Therefore, her example does not provide sufficient evidence for the claim that unaccusative verbs are allowed in the wayconstruction. The Dutch verb groeien ‘grow’ could then be polysemous too. This verb normally selects the BE-auxiliary and is unaccusative, as in (24a). However, it can also have an unergative reading, in which case it selects HAVE, as in (24b). (24) a. Een bevolking die naar verwachting eind dit jaar zal zijn gegroeid tot 935 a population that to expectation end this year will be grown to 935 miljoen mensen. million people ‘A population that is expected to have grown to 935 million people by the end of this year.’ www.goedzo.com/index.php/2004/11/08/ b. … de breedte van de ring geeft aan hoe snel de boom heeft gegroeid. the width of the ring indicates how fast the tree has grown ‘The width of the ring indicates how fast the tree has grown.’ www.hetweer.org/HetWonderlijkeWeer/klimaats3.htm The example in (24b) shows that groeien has an unergative variant, which suggests that it is polysemous in Dutch as well. Hence, it could be the unergative version in the weg-construction in (23b), which has an agentive subject. This hypothesis is confirmed by the possibility to insert an agentive adverb in this weg-construction, and by the selection of the HAVE-auxiliary, as shown in (25). (25) De zonnebloemen hebben / *?zijn zich the sunflowers have koppig een weg naar het licht gegroeid. be REFL stubborn a way to the light grown ‘The sunflowers have stubbornly grown their way towards the light.’ 51 This example suggests that groeien is unergative here, because its subject is an Agent rather than a Theme. I conclude that the acceptability for some speakers of groeien in the wegconstruction does not provide evidence for the claim that unaccusative verbs are allowed in the weg-construction, because this verb is polysemous. When this verb occurs in the weg-construction, it is unergative. Further evidence for the claim that unaccusative verbs are incompatible with the weg-construction is provided by unaccusative verbs that have a transitive alternant, such as breken ‘break’ and smelten ‘melt’. The subject of the transitive alternant is an Agent, so the intransitive version of this verb is unergative. This unergative version is acceptable in the weg-construction, as illustrated by the following attested examples. (26) a. Deze magma-bel breekt zich een weg door de aardkorst en vormt aan this magma bubble breaks REFL a way through the earth.crust and forms on het aardoppervlak een vulkaan. the earth.surface a volcano ‘The magma breaks its way through the earth’s crust and forms a volcano at the surface.’ www.geoclopedie.nl/Alfabet/Begrip-H.htm b. Hun koperen top kan tot circa 195 graden Celsius heet worden, zodat de their copper top can till circa 195 degrees C cryobots zich een weg door hot become so.that the het ijs smelten. cryobots REFL a way through the ice melt ‘Their copper top can heat up to 195 degrees C, so the cryobots can melt their way through the ice.’ www.hetlaatstecontinent.be/continent/lake_vostok.html We can tell that the verbs in these examples are not unaccusative, because their subjects are Agents, not Themes. That is, the magma in (26a) is the Agent of the breaking, and the cryobots in (26b) are the Agents of the melting. The unaccusative alternants of these verbs are not acceptable in the weg-construction, as was shown in respectively (12b) and (14b) above. 52 The auxiliary that is selected by the weg-construction provides another piece of evidence for the hypothesis that the verb in the weg-construction is unergative. Of the alternating verbs discussed above, the unaccusative variant selects BE and the transitive variant takes HAVE, as shown in (27) for smelten. (27) a. Het ijs is gesmolten. the ice is melted ‘The ice has melted.’ b. De zon heeft het ijs gesmolten. the sun has the ice melted ‘The sun melted the ice.’ In the weg-construction, these alternating verbs select HAVE, as shown in (28). (28) a. Het magma heeft / *is zich een weg the magma has door de aardkorst gebroken. is REFL a way through the earth.crust broken ‘The magma has broken its way through the earth’s crust.’ b. De cryobots hebben / *zijn zich een weg door the cryobots have het ijs gesmolten. are REFL a way through the ice melted ‘The cryobots have melted their way through the ice.’ The selection of HAVE rather than BE confirms that the verb in the weg-construction is agentive: the magma in (28a) and the cryobots in (28b) are the Agents of respectively the breaking and the melting, not the Themes. Manner of motion verbs provide another piece of evidence for the hypothesis that the verb that enters the weg-construction is unergative. As mentioned in chapter 1, manner of motion verbs are normally unergative and select the HAVE-auxiliary. However, when combined with a directional phrase they are unaccusative and take BE. Although it denotes directed motion, the weg-construction selects HAVE with a manner of motion verb as well, as shown in (29) for the example in (8c) above. (29) Chris de Witte heeft / *is zich een weg door C. de W. has de Bredase defensie geslalomd. is REFL a way through the B. defence slalomed ‘Chris de Witte slalomed his way through the Breda defence.’ 53 Although the motion is directed, the weg-construction selects the HAVE-auxiliary, which confirms that the verb in the weg-construction is unergative.4 I conclude that the verb in the weg-construction is unergative. Unaccusative verbs are semantically incompatible with the weg-construction because their subject is a Theme, which cannot perform a volitional and self-propelled action. 3.2.3 Subjective motion Manner of motion verbs are not very frequent in the weg-construction, and they are also not approved of by all speakers. However, two manner of motion verbs are quite common in the weg-construction, which are kronkelen and slingeren, both meaning ‘wind’. These verbs denote the actual manner of motion, so the reflexive and the weg NP can also be omitted, which results in a reading with a shorter path, which requires less effort to travel. The subjects of these verbs are usually inanimate and they refer to anything that is long and windy, such as rivers, roads, queues, and so on, as exemplified in (30). Note that weg-constructions with these verbs do not denote motion because nothing is moving. (30) a. … de moddersporen kronkelen zich een weg door het bos. the mud.tracks wind REFL a way through the forest ‘… the mud tracks wind their way through the forest.’ www.autozine.nl/text/291.pdf b. … slingert de Orke zich een weg naar zijn eindbestemming de Eder. winds the O. REFL a way to his destination the E. ‘… the Orke winds its way to its destination the Eder.’ www.devluchtendevisser.nl/html/landen/salmonidenduitsland.htm The subject in (30a) is a mud track and in (30b) it is the river Orke, but mud tracks and rivers do not move. That is, the path of a river is a pre-established channel which it does not move out of (or only very slowly, which is not the intended meaning in this example). In fact, the mud track and the river themselves constitute the path, which therefore is pre-established. Since the mud track and the river are already there, 4 Zaenen (1993) argues that auxiliary selection in Dutch is related to telicity: atelic predicates select and telic ones select BE. I will argue in section 3.6 that the weg-construction is not necessarily telic, which agrees with the selection of HAVE rather than BE. HAVE 54 they do not move through the landscape ‘by means of’ winding, so these examples differ form the general meaning of the weg-construction. Langacker (1990:157) refers to this kind of perception of motion with inanimate subjects in stable situations where nothing is actually moving as subjective motion. He proposes that our conception of motion in such instances is related to the fact that the subjects themselves are elongated, path-like objects, which have a spatial configuration with some internal complexity. Our perception that these objects are moving through space is built-up incrementally, which requires some processing time; “rather than springing instantaneously into full-blown existence, the conception might be built-up incrementally, with all facets of it being active only at the conclusion of this ‘build-up phase’” (1990:158). Hence, Langacker attributes the motion we perceive to the order in which the various facets of the spatial configuration are activated. Since the subject NPs themselves are elongated path-like objects, we perceive them as building up incrementally, from some starting point to eventually reaching their full extension. So although nothing is actually moving, these examples still instantiate the weg-construction. This is because the perceived motion is an abstract subjective motion. By contrast, when the subject is animate, a weg-construction with slingeren or kronkelen does denote real motion along a path, by means of (or while) V-ing. In this case they are like any other weg-construction with a manner of motion verb, as illustrated in (31). (31) We slingeren ons een weg naar de guesthouse van het Nationaal Park… we wind REFL a way to the guesthouse of the N. P. ‘We are winding our way to the guesthouse of the National Park.’ www.geocities.com/freija_jeroen/verslagen_oost_afrika_Uganda.html The subject referents in this example are people sitting in a bus that is driving down a mountain on a windy road. The verb thus expresses the actual manner of motion and not the means by which a path is created, because the path is pre-established. The reason why they are winding is because the road they are on is winding. Since the verb denotes the actual manner of motion, the reflexive + een weg can also be omitted, again yielding a slightly different meaning: the path in the weg-construction seems to be longer and require more effort to travel. 55 3.2.4 Means and manner interpretation In all of the above examples, the verb denotes the means by which a path is created and/or travelled, by removing obstacles or overcoming other barriers. As a result, there is some external difficulty to the motion of the subject. However, Jackendoff (1990) and Goldberg (1995) note that for some speakers the verb in the English wayconstruction can also denote an activity accompanying the motion. Goldberg refers to this reading as the manner interpretation, which does not involve any external difficulty, as it does not entail the creation of a path. Verhagen (2003:38) claims that the manner interpretation is not at all available in Dutch. However, I found a few instances in which the verb can only be interpreted as denoting an action that accompanies the motion of the subject, instead of the means by which a path is created, such as the following. (32) a. Verhuurbootjes fluisteren zich een weg door de Dorpsgracht. (INL) rental.boats.DIM whisper REFL a way through the village.canal ‘Rental boats are whispering their way across the Dorpsgracht.’ b. … ruzieden we ons een weg over de Laan van Meerdervoort heen … quarrelled we REFL a way over the L. van M. across ‘We quarrelled our way across the Laan van Meerdervoort.’ www.infiniti.web-log.nl/index.log?ID=441644 These examples sound odd to me, because (32a) implies that the boats are moving by means of whispering, and (32b) means that the subject referents are moving by means of quarrelling. To test the availability of the manner interpretation in Dutch, the wegconstruction in (33) was included in the third questionnaire, and the participants were asked to indicate to indicate the meaning of this sentence, where they could choose between the two meanings in (34). (33) Jan boert zich een weg uit het restaurant. J. belches REFL a way out the restaurant ‘Jan belches his way out of the restaurant.’ 56 (34) a. Jan walks out the restaurant while belching. b. Jan uses the belching as a means to get out of the restaurant, for example to frighten the guests in such a way that they are going out of his way. Appendix 3 shows that 37% of the 30 speakers indicated the manner interpretation in (34a) as the most appropriate paraphrase, whereas 63% opted for the means interpretation in (34b). These results indicate that in Dutch too, there are some dialects where a manner interpretation is available. For the majority of speakers, however, the only interpretation is a means interpretation. One could argue that manner of motion verbs in the weg-construction also instantiate the manner interpretation, as these verbs do not denote the means by which a path is created (e.g. Ze roeien zich een weg naar de finish ‘They are rowing their way to the finish’). However, manner of motion verbs do not really denote an activity that accompanies the motion, because when somebody rows to the finish, he does not move while rowing, but rather by means of rowing. I conclude that manner of motion verbs do not denote the means by which a path is created, but the means by which a path is travelled. Therefore, they do not instantiate a true manner interpretation of the weg-construction. 3.2.4 Summary The weg-construction denotes motion of the subject referent along a self-created path, which is created and/or travelled by means of the action denoted by the verb. The verb in the weg-construction refers to the means by which obstacles are removed or barriers are overcome in order to move. Consequently, the motion takes place with some effort and external difficulty. For some speakers, the verb can also denote a manner accompanying the motion. A manner of motion verb in the weg-construction denotes the actual manner of motion, so the reflexive and the weg NP can also be omitted. However, the weg-construction implies more effort and a longer path than its bare motion counterpart. Since a manner of motion verb in the weg-construction does not denote the means of creating a path, not all speakers accept the construction with a manner of motion verb. A large variety of activity verbs can be found in the weg-construction. Constraints on the verb in the weg-construction include that the verb must be able to be used unergatively, which rules out strictly (di)transitive verbs, inherently reflexive 57 verbs and unaccusative verbs. Furthermore, the verb must be conceivable as the means by which a path is created and/or travelled. This semantic constraint rules out stative verbs and plain motion verbs. Since the motion in the weg-construction takes place by means of continuously executing the action described by the verb, the verb has to be a process verb or else denote a repeatable bounded event. In the latter case, the weg-construction strongly implies a repetition of that event. The weg-construction has two syntactic NP complements that are not semantic arguments of the verb. The construction thus represents a mismatch in the mapping between syntax and semantics: the thematic objects of the verb (if any) cannot be mapped onto the syntactic arguments of the verb in the weg-construction. 3.3 The reflexive element The first postverbal NP in the weg-construction is a reflexive, which occurs in indirect object position of the verb. This section provides evidence for the claim that the reflexive is not a semantic argument of the verb in the weg-construction. Furthermore, I will argue that the reflexive in the weg-construction has lost the meaning of beneficiary. 3.3.1 Characteristics of the reflexive The reflexive has to agree with the subject in both person and number, as illustrated in (34). (34) a. *Tarzan hakte T. mij een weg door de jungle. slashed REFL.1.SG a way through the jungle b. Ik hakte mij een weg door de jungle. I slashed REFL.1.SG a way through the jungle ‘I slashed my way through the jungle.’ The example in (34a) is unacceptable because the subject and the reflexive do not agree in person: mij is a first person singular reflexive but the subject Tarzan is third person. The sentence in (34b) shows a grammatical example with the first person singular reflexive. I will assume throughout this paper that reflexives agree with the subject in both person and number. The reflexive in the weg-construction is obligatorily weak, as shown in (35). 58 (35) Tarzan hakt T. zich(*zelf) een weg door de jungle. shlashes REFL.self a way through the jungle ‘Tarzan slashes his way through the jungle.’ This example shows that a strong reflexive marked with zelf ‘self’ is unacceptable in the weg-construction. Moreover, the reflexive is obligatorily present in the weg-construction. The following examples show that omission of the reflexive does not result in ungrammaticality, but it changes the meaning of the sentence. Compare the wegconstruction in (36a) with the sentence in (36b). (36) a. #Tarzan heeft zich een weg door T. de jungle gehakt, maar hij is nog niet has REFL a way through the jungle slashed but he is yet not gegaan. gone ‘Tarzan slashed his way through the jungle, but he has not gone yet.’ b. Tarzan heeft een weg T. has door de jungle gehakt, maar hij is nog niet gegaan. a path through the jungle slashed but he is yet not gone ‘Tarzan slashed a path through the jungle, but he has not gone yet.’ The sentence in (36a) with the reflexive is non-sensical because it instantiates the weg-construction, which denotes motion. The sentence in (36b) without the reflexive, on the other hand, is fine, because it does not express motion. This sentence can only mean that Tarzan slashed a path through the jungle, where weg is a semantic argument of the verb, but it does not imply that he also travels the path. Therefore, the reflexive is obligatorily present in the weg-construction, as the sentence has a different meaning in the absence of the reflexive. 3.3.2 The reflexive is not a semantic argument of the verb This section provides evidence for the claim that the reflexive in the weg-construction is not a semantic argument of the verb. First of all, as shown in several examples above, intransitive verbs can also occur in the construction, which do not normally take any semantic arguments besides a subject. Secondly, if the reflexive were a 59 thematic object of the verb, we would expect it to be exchangeable for another object. However, this is not possible, as shown in (37). (37) *Tarzan hakt T. Jane een weg door slashes J. de jungle. a way through the jungle ‘Tarzan slashes Jane a way through the jungle.’ In this example, the reflexive zich has been replaced with the full NP Jane in indirect object position and the sentence in ungrammatical.5 Third, verbs that normally require a strong reflexive as their thematic object cannot take a strong reflexive in the weg-construction. This is illustrated (38). (38) a. Jan schopt zich*(zelf). J. kicks REFL.self ‘Jan kicks himself.’ b. Jan schopt zich(*zelf) een weg J. kicks REFL.self door de menigte. a way through the crowd ‘Jan kicks his way through the crowd.’ The reflexive direct object needs to be marked with zelf in (38a), but the reflexive object in the weg-construction in (38b) may not be marked with zelf. The ‘sloppy identity’ test may provide a fourth piece of evidence that the reflexive in the weg-construction is not a thematic object of the verb. Sells, Zaenen & Zec (1987) use this test to distinguish between semantically transitive and semantically intransitive predicates in English, with a reflexive occurring in direct object position. They discuss the following sentence. (39) John defends himself better than Peter. (1987:175, 19) If this were a semantically intransitive construction, they argue, one would expect this sentence to only have one reading, which is given in (40a). However, this sentence has the two additional readings given in (40b-c) (1987:175, 20). 5 This sentence would be acceptable with the beneficiary Jane expressed in a prepositional phrase. However, this sentence would not denote motion. 60 (40) a. John defends himself better than Peter defends himself. (‘sloppy identity’) b. Johni defends himself better than Peter defends himi. (‘strict identity’) c. Johni defends himself better than hei defends Peter. (‘object comparison’) The ambiguity of (39) arises because this example can be semantically transitive and intransitive. The sloppy identity reading in (40a) arises from the construction being semantically intransitive, because the reflexive is bound to the subject. Sells, Zaenen & Zec propose that the readings in (40b-c) come from the construction being semantically transitive. The reflexives are free NPs that are not bound to the subject and are interpreted as a thematic object of the verb. Sells, Zaenen & Zec’s example is syntactically transitive, whereas the wegconstruction is syntactically ditransitive. However, I propose that the sloppy identity test can still be used to show that the reflexive in the weg-construction is not a semantic argument of the verb. This is because a Dutch ditransitive sentence has the same three readings as the transitive sentence discussed by Sells, Zaenen & Zec. The sentence in (41) has the three readings given in (42). (41) Johan geeft zichzelf eerder een kans dan Stefan. J. gives REFL.self earlier a chance than S. ‘Johan gives himself earlier a chance than Stefan.’ (42) a. Johan gives himself a chance earlier than Stefan gives himself a chance. (sloppy identity) b. Johani gives himself a chance earlier than Stefan gives himi a chance. (strict identity) c. Johani gives a chance to Johan earlier than hei gives a chance to Stefan. (object comparison) Only in (42b) and (42c) is the pronoun a free NP, which is used as a semantic argument of the verb. Compare the ditransitive sentence in (41) to the wegconstruction in (43), which only has the sloppy identity reading. 61 (43) Johan werkte zich eerder een weg naar de top dan Stefan. J. worked REFL faster a way to the top than S. ‘Johan worked his way to top earlier than Stefan.’ This weg-construction example can only mean that Johan worked his way to the top earlier than Stefan did, which is the sloppy identity reading. This suggests that the verb is used intransitively. These pieces of evidence indicate that, although the reflexive occupies the indirect object position of the verb in the weg-construction, it is not a semantic argument of the verb. 3.3.3 The ditransitive pattern in Dutch As discussed in chapter 1, Verhagen notes that the weg-construction is a productive instance of a generally unproductive pattern: the construction is ditransitive, but this pattern can only be used in Dutch when the direct object is not created (e.g. with verbs of transfer, 2004:414). Verhagen nevertheless assumes that the reflexive in the weg-construction is a beneficiary, based on the history of the construction. However, I will argue that the reflexive has lost its meaning of beneficiary in the modern wegconstruction. Verhagen shows that the ditransitive pattern used to be productive in Dutch in th the 17 and 18th century. As discussed in chapter 1, he proposes that the weg- construction may have originated from the verb banen with a beneficiary NP and een weg, which meant ‘to level out a path for NP’. The beneficiary could also be somebody other than the subject. The expressions with a reflexive beneficiary were then extended to also convey motion of the subject referent. These patterns were subsequently extended to also include other verbs that denote the means by which a road was levelled out, as well as metaphorical paths. Verhagen’s historical observations suggest that the weg-construction with its odd ditransitive syntax is a remnant from old Dutch. The ditransitive double object construction seems to have lost most of its productivity in modern Dutch, as shown by the following examples. (44) a. *Peter bakt P. zich(zelf) / Marijke een taart. bakes REFL.self M. a cake 62 b. Peter bakt een taart voor zich*(zelf) / Marijke. P. bakes a cake for REFL.self M. ‘Peter bakes a cake for himself/Marijke.’ The double object construction in (44a) is ungrammatical; the sentence is only grammatical when the indirect object is expressed in a prepositional phrase. The double object construction is only grammatical in Dutch with verbs of transfer, such as geven ‘give’ in (19b) and (41) above. Interestingly, the ditransitive pattern seems to have retained somewhat more of its productivity in Flemish, the Belgian dialect of Dutch. The ditransitive patterns that I found in the data with a direct object that is created mainly came from Belgian sites, such as the following.6 (45) a. Dan knabbelt hij zich een tunnel doorheen de dichte begroeiing van het then nibbles he REFL a tunnel through.PRT the dense growing of the tropische woud. tropical forest ‘Then he nibbles himself a tunnel through the dense vegetation of the tropical forest.’ www.wwf.be b. Bouwt zichzelf een slaapplaats, builds REFL.self a sleep.place ‘Builds himself a place to sleep.’ www.melvine.be It is possible that the ditransitive construction has retained (some of) its productivity in Flemish. This hypothesis is confirmed by the results of the second questionnaire, which contained the following double object construction (see appendix 2). 6 Although the reflexive in (45a) is an argument of the verb, it is a weak reflexive. I claimed above that reflexives that are semantic arguments of the verb can be marked with zelf, as can also be seen in (45b). The distribution of strong and weak reflexives is a complex issue in Dutch, which amongst others has to do with focus (cf. Reinhart & Reulandt 1993; Veraart 1996). I will show in the next chapter that some weak reflexives can also be used as a semantic argument of the verb. 63 (46) De mol graaft zich een gang onder de grond. the mole digs REFL a tunnel under de ground ‘The mole digs himself a tunnel under the ground.’ This sentence was approved of by all three Belgian participants and by 10 of the 39 Dutch speakers (together 31% of the participants). Two of the Dutch participants who approved of (46) thought that it was “old-fashioned”. I conclude that the ditransitive pattern has lost most of its productivity in modern Dutch for most speakers, but perhaps not in Flemish. Therefore, speakers of standard Dutch do not interpret the reflexive that occurs in indirect object position in the weg-construction as a semantic beneficiary. Rather, it is a fixed element in the weg-construction that has lost its meaning. 3.3.4 Summary The first NP complement in the weg-construction is a reflexive, which is obligatorily weak. The reflexive occurs in indirect object position, but it is not a semantic argument of the verb. Moreover, the double object construction is unproductive in modern Dutch, so the weg-construction is a productive instance of an unproductive pattern. I follow Verhagen, who hypothesises that the weg-construction with its odd syntax may have originated in the 17th or 18th century Dutch, when the double object construction was fully productive. However, I propose that the reflexive has lost its meaning of beneficiary in the modern weg-construction. 3.4 The weg NP The second NP complement in the weg-construction is the een weg phrase. This section provides evidence for the claim that the weg NP is not a semantic argument of the verb, and also investigates the status of the weg NP in the weg-construction. As discussed in chapter 1, the status of the way NP in the English way-construction is disputed. Goldberg (1995) argues that the way phrase is a meaningful element that denotes the path that is created, whereas Jackendoff suggests it is a “meaningless syntactic marker” (1995:215). Marantz (1992) and L&RH (1995) propose that the way NP is a reflexive path, parallel to the fake reflexive in the resultative construction. Verhagen does not discuss the status of the weg NP in the Dutch construction. However, he follows Goldberg in assuming the weg NP and the PP are 64 separate arguments: the weg NP denotes the path that is created and the PP the path that is travelled (see section 1.3). 3.4.1 Characteristics of the weg NP The weg NP is a fixed element in the weg-construction. The weg noun cannot be replaced by another noun denoting a path, and the article is obligatorily indefinite, as shown in (47a) and (47b), respectively. (47) a. *Jan schopt zich een pad door J. kicks REFL a path through the crowd b. *Jan schopt zich de weg J. de menigte. door de menigte. kicks REFL the way through the crowd Furthermore, the weg NP cannot be omitted, as shown in (48).7 (48) ??Jan schopt zich J. door de menigte. kicks REFL through the crowd In sum, the een weg phrase is a fixed element in the weg-construction that occurs in the direct object position and which cannot be omitted. 3.4.2 The weg NP is not a semantic argument of the verb The fact that een weg is a fixed element provides a first piece of evidence that it is not a semantic argument of the verb; if it were a thematic object of the verb, we would expect it to be replaceable with another noun denoting a path. A second piece of evidence comes from the fact that intransitive verbs occur in the weg-construction as well, as was shown in several examples above. Intransitive verbs do not normally take any semantic arguments besides a subject, which confirms that een weg cannot be a semantic argument of the verb. The incompatibility of the weg-construction with the passive provides a third piece of evidence that the weg NP is not a semantic argument of the verb. The ungrammatical example in (49) shows that the weg-construction cannot be passivized. 7 This sentence may be marginally acceptable as a different construction in Dutch, which describes a transition to a location. This construction is the topic of the next chapter. 65 (49) *Een weg werd (zich) door de menigte geschopt (door Jan/zich). a way was REFL through the crowd kicked by J. REFL The fact that the weg NP cannot be moved into subject position to form a passive sentence confirms that it is not a semantic argument of the verb. That is, it parallels the non-object the bucket in kick the bucket, which likewise cannot be passivized (Jackendoff 1990:216). These pieces of evidence indicate that, although the weg NP occurs in direct object position, it is not a semantic argument of the verb. 3.4.3 Adjectival modifiers on the weg NP Goldberg argues that the way NP is a meaningful element in the way-construction in English, which represent the path that is created. Her claim is based on her observation that adjectival modifiers on way modify the path. For instance, in he made his familiar way home, the path is familiar (1995:215). Jackendoff, on the other hand, states that adjectival modifiers on way do not modify the path. Instead, they are either transferred as adverbial modifiers to the going event, or they turn up as absolutives (1990:217). Thus, Johnny belched his silly way home is either interpreted as Johnny going home in a silly manner, or that he himself was silly. It does not mean that his path was silly. Marantz (1992) argues that adjectives modify the path, but he assumes that the path is the person referred to by the possessive pronoun. The adjective in Johnny belched his silly way home modifies the path (1992:185).8 The Dutch noun weg literally means ‘path’ or ‘road’, it does not have the metaphorical reading that way also has in English.9 The fact that weg only has a literal meaning could mean that it is a meaningful element in the weg-construction denoting the path that is created. The meaning of adjectival modifiers on the weg NP may reveal whether it is a meaningful element in the construction or not. Adjectival 8 Since Marantz assumes that the path is in fact the person referred to by the possessive pronoun on way extended through space, the adjective in fact turns up as an absolutive (cf. Johnny, silly, belched his way home). This suggests that adjectival modifiers on way in fact have the meaning proposed by Jackendoff. 9 Recall that I gloss weg in the examples with way in order to illustrate the similarities of the Dutch weg-construction with the English way-construction. 66 modifiers on the weg NP are rare, but they do occur.10 Most findings involve the verb banen, such as (50), which is the title from Verhagen (2004). (50) Hoe het Nederlands zich een eigen weg baant. how the Dutch.language REFL an own way baant ‘How the Dutch language is making its own way.’ The adjective eigen ‘own’ clearly modifies the path, because only the path can be eigen, not the going event or the subject. However, an adjectival modifier on weg can also turn up as manner adverb to the going event or as an absolutive, as shown in respectively (51a) and (51b). (51) a. … en zich een moeizame weg door een oneindige zandvlakte gebaand. and REFL a laborious way through an endless sand.plain gebaand ‘… and made his laborious way through an endless sand plain.’ www.fellowship.nl/forum/showthread.php?s=dd4a7eb6a1563fac0f3234c1593cf934&postid b. … baande Merckx zich een eenzame weg baande M. REFL a door de gutsende regen. lonely way through the pouring rain ‘Merckx made his lonely way through the pouring rain.’ www.pinguinproductions.be/02_blikvangers/pdf/rvv_vb03.pdf In the weg-construction in (51a), it is the going that is moeizaam ‘laborious’, not the path, because a path cannot be laborious. In (51b) it is the subject who is eenzaam ‘lonely’, not the path or the going, so the adjective is interpreted as an absolutive. These data suggest that an adjective that modifies the weg noun in a weg-construction with banen can either modify the path itself, or it can turn up as a manner adverb to the going event or as an absolutive. The corpus searches revealed few instances of a weg-construction with an adjectival modifier to the weg NP with a different verb than banen, some of which are given in (52). 10 In chapter 2 I claimed that it is not possible to search for strings like “zich een […] weg” with Google. However, since one can search for literal strings, one can search for “zich een mooie weg” ‘REFL a beautiful way’, “zich een snelle weg” ‘REFL a quick way’, and so on. 67 (52) a. Hij heeft zich een lange weg teruggevochten. he has REFL a long way back.fought ‘He fought his long way back,’ www.let.rug.nl/vannoord/trees/h_suite/320.xml b. ... waar het merendeel van de mensheid where the most zich een moeizame weg door het of the humankind REFL a laborious way through the bestaan ploeterde. existence plodded ‘Where most of humankind plodded its laborious way through its existence.’ verhalensite.com/info2.php?s=st&ss=rpr&id=8197&PHPSESSID=1c8a0c71d6a1a0bacfdb c. … zich een moeizame weg naar de top moet zien te werken. REFL a laborious way to the top must see to work ‘… has to work his laborious way to the top.’ www.pimfortuyn.nu/pimpamflet/augustus/13%20augustus%202002%201_34.htm In the example in (52a), it is not clear whether the adjective lang ‘long’ modifies the path or the going event. A path can of course be long, but so can a going event (that is, it can take a long time). The path in this example is metaphorical, and it seems to me that lang is modifying the going event, because the sentence implies that it took the subject a long time to fight his way back (for instance, to functioning properly in society again). The adjectives in (52b-c), on the other hand, clearly modify the going. Since a path cannot be laborious, the adjectives turn up as manner adverbs to the going event. That is, the subject referents are moving laboriously in both examples, which requires effort. A sentence very similar to (52b) was included in the third questionnaire, which was approved of by 63% of the 30 speakers (see appendix 3). These results indicate that, although weg-constructions with an adjectival modifier on the weg NP are rare in the corpora, they are acceptable to many speakers. This adjective does not modify the path but (uncharacteristically) turns up as a modifier to the going event. The fact that adjectival modification on the weg noun does not modify the path suggests that the weg noun is non-referential in the wegconstruction. The observation that adjectives on weg can only modify the path in wegconstructions with the verb banen may be explained by banen being a special verb in the weg-construction, which only occurs idiomatically in the weg-construction. The weg-construction may have originated from the combination of banen with a literal 68 path, which may be why the weg noun can still be interpreted as a real path in wegconstructions with banen. By contrast, in weg-constructions with other verbs than banen, the weg noun may have lost its meaning. Interestingly, the evidence that Goldberg provides to argue that the way noun is a meaningful element in the way-construction also only involves verbs that occur idiomatically in it. She claims that an adjective on way modifies the path, based on way-constructions with make and wend (e.g. he made his familiar way home). However, in way-constructions with different verbs, an adjectival modifier on the way noun does not modify the path. Speakers do not approve of way-constructions with other verbs where way is modified by an adjective that can only be interpreted as modifying the path (e.g. *Bill belched his familiar way out of the restaurant, *Sam joked his own way into the meeting). Furthermore, a Google search for “his/her/their familiar way” only resulted in way-constructions with make and find. This may suggest that in English too, adjectives can only modify the path in the special cases with a verb that occurs idiomatically in the construction (e.g. make and wend). I conclude that adjectival modifiers on the weg NP can only modify the path in weg-constructions with the special verb banen, which means that the weg noun may only be referential in weg-constructions with banen. In weg-constructions with other verbs, the adjective either turns up as a manner adverb or as an absolutive, suggesting that the weg noun is meaningless. Section 3.7.2 shows that the weg phrase does not appear in the conceptual structure of the weg-construction. 3.4.4 Summary The weg NP in the weg-construction is a fixed element that is obligatorily present. Although the weg NP occurs in direct object position, it is not a semantic argument of the verb. Adjectives modifying the weg noun do not modify the path, suggesting that the weg NP in the weg-construction is not referential. 3.5 The PP This section investigates the different kinds of prepositions that can be found in the weg-construction. Table 3.2 gives the prepositions that were found heading the PP of the weg-construction in the INL corpus. 69 Table 3.2 Prepositions heading the PP of the weg-construction (INL corpus) Preposition Number of hits / P % of total / P Door ‘through’ 30 50.8 Naar ‘to’ 14 23.7 over ‘over, across’ 5 8.5 - 3 (with banen only) 5.1 in ‘in’, tussen ‘between’, langs ‘along’ 2 3.4 recht vooruit ‘straight ahead’ 1 1.7 Total number of hits 59 Altogether seven different prepositions were found in the weg-construction in the INL corpus. The two most common prepositions are door ‘through’ and naar ‘to’. These results are comparable to those of Verhagen, who found the same prepositions in similar proportions, plus the prepositions terug ‘back’, tot ‘until’ and uit ‘out’. The same prepositions were also found with Google, with door and naar being the most common ones. Other prepositions include binnen ‘inside’, langs ‘past, along’, omhoog ‘up’, onder ‘under’, op ‘on’, van ‘from’, via ‘via’ and voorbij ‘past’. The preposition in the weg-construction can be intransitive as well, as prepositions like terug ‘back’ and recht vooruit ‘straight ahead’ do not take an NP complement. The only weg-construction examples that lack a PP involve the verb banen. 3.5.1 Path prepositions Besides a few exceptions such as in ‘in’ and binnen ‘inside’, the prepositions that occur in the weg-construction belong to Jackendoff’s conceptual category Path. Place prepositions like in or binnen are very rare in the weg-construction, which suggests that the PP in the weg-construction is headed by a Path preposition. As discussed in chapter 1, a path modifying the VP is realised as a postpositional phrase in Dutch, as in (53a). By contrast, a path that modifies a noun is realised as a prepositional phrase, as shown in (53b). 70 (53) a. Jan rent het bos J. door. runs the forest through ‘Jan runs through the forest.’ (telic) b. een paadje door het bos / *het bos door a path.DIM through the forest the forest through ‘a path through the forest.’ The example in (53b) illustrates that a prepositional phrase that modifies a noun can describe a path. The conceptual structure (CS) of this NP is given in (54). (54) [Thing PAADJE [Path VIA ([Place IN ([Thing BOS])])]] This CS describes the meaning of the NP in (53b), where the PP describes the trajectory of the path denoted by the noun. Jackendoff represents the Path preposition door ‘through’ as two functions: the Path function VIA, which has the Place function IN as its argument. The prepositions that were found in the weg-construction can be expressed with one of Jackendoff’s Path functions, which include TO, FROM, TOWARD, AWAY-FROM, DOWN and VIA. These Path functions map a reference Thing or Place into a related trajectory. The following examples illustrate some of the other Path functions. (55) a. een weg naar de winkel ‘a road to the shop’ b. [Thing WEG [Path TO ([Thing WINKEL])]] (56) a. een tunnel uit de stad ‘a tunnel out of the city’ b. [Thing TUNNEL [Path FROM ([Thing STAD])]] The PPs in these examples describe the trajectory of the path expressed by the noun. The goal in (55) is an argument of the Path function TO and the source in (56) is an argument of the Path function FROM. By contrast, prepositions like in ‘in’ and achter ‘behind’ cannot specify the trajectory of a path. These prepositions belong to the conceptual category Place and they can only describe the location of a path, as illustrated in the following examples. 71 (57) a. een pad in het bos ‘a path in the forest’ b. [Thing PAD [Place IN ([Thing BOS])]] (58) a. de weg achter de berg ‘the road behind the mountain’ b. [Thing WEG [Place BEHIND ([Thing BERG])]] Place prepositions such as those in the above examples are very rare in the wegconstruction. The vast majority of the prepositions heading the PP of the wegconstruction are Path prepositions that can specify the trajectory of a path. I will argue in section 3.8 below that the PP and the weg NP form a constituent, where the PP specifies the trajectory of the weg noun, comparable to the NPs in (54) to (56). 3.6 Relation to the ‘fake’ object resultative construction The relation of the English way-construction to the ‘fake’ object resultative is another controversial point in the literature. Goldberg (1995) and Jackendoff (2002) argue that the way-construction is a distinct construction, but Marantz (1992) and L&RH (1995) propose that the two constructions are exactly parallel. Verhagen states that the two constructions are different constructions in Dutch, because the weg-construction is ditransitive, whereas the resultative is transitive. I agree with Verhagen that the two constructions are different constructions in Dutch, first of all because they are syntactically distinct, but also because they differ in telicity. The resultative construction is telic per definition, whereas the wegconstruction can also be atelic. The standard test to determine the telicity of an expression is to combine it with a durative and a non-durative time adverbial, such as for X time and in X time, respectively. The former can only be combined with an atelic expression, whereas the latter is only acceptable with a telic expression. To investigate the telicity of the weg-construction, the following two sentences were included in the second questionnaire. (59) a. Tarzan heeft zich dagenlang een weg door T. has REFL days.long de jungle gehakt. a way through the jungle slashed ‘Tarzan slashed his way through the jungle for days.’ 72 b. Tarzan heeft zich in twee dagen een weg T. has REFL in two days door de jungle gehakt. a way through the jungle slashed ‘Tarzan slashed his way through the jungle in two days.’ Of the 42 participants, 62% approved of the atelic reading in (59a), whereas about 79% approved of the telic reading in (59b); 57% of the speakers approved of both (see appendix 2). These results indicate that the weg-construction can have either a telic or an atelic reading. Since the weg-construction also has an atelic reading, this construction cannot be assimilated to the resultative construction.11 3.7 The structure of the weg-construction Verhagen assumes that the weg NP and the PP in the weg-construction are separate constituents, which denote the path that is created and the path that is travelled, respectively. The analysis proposed here differs in this regard, because I propose that the weg NP and the PP form a constituent. That is, the PP is an adjunct to the weg noun and specifies the trajectory of the path. This constituent then denotes both the path that is created and the path that is travelled. I will argue that one argument suffices to denote both the path that is created and the path that is travelled, because these two subevents are in fact the same event. There are several reasons that support the hypothesis that the entire weg NP denotes the path in the weg-construction. First of all, as concluded in section 3.4, the een weg phrase is non-referential, so it cannot denote the path that is created. Secondly, if it were the weg noun that denotes the path that is created and the PP denotes the path that is travelled (as suggested by Verhagen), the weg-construction would contain three internal arguments: the reflexive, the weg NP (created path) and the PP (travelled path). This is clearly undesirable, because verbs do not normally take more than two internal arguments. This also means that the creation of the path and the travelling of the path are not necessarily co-extensive, as they are expressed by separate arguments. However, the creation and the travelling of the path are always co-extensive in the weg-construction, because the path is travelled by creating it; 11 Goldberg & Jackendoff (2004:543) argue that the view that resultatives are invariably telic is false. They discuss “atelic path resultatives” and propose that the telicity of path resultatives depends on the boundedness of the resultative phrase. In their view, atelic weg-constructions may therefore still be resultatives. However, I propose that the fact that the weg-construction also has an atelic reading distinguishes it from the fake object resultative construction, which is invariably telic. 73 when the subject stops creating the path, he will stop traversing the path. Consequently, there should only be one argument that expresses the path. The expression of the path of a directed motion verb provides a third piece of evidence for why the PP by itself cannot express the path. As discussed in chapter 1, the path of a directed motion verb is realised as a postpositional phrase; a prepositional phrase with a motion verb is a frame locative. The weg-construction takes a prepositional phrase, as shown in (60a). A weg-construction with a postpositional phrase would be ungrammatical, as shown in (60b). (60) a. Tarzan hakte zich een weg T. door de jungle. slashed REFL a way through the jungle ‘Tarzan slashed his way through the jungle.’ b. *Tarzan hakte T. zich een weg de jungle door. slashed REFL a way the jungle through The fact that the weg-construction takes a prepositional phrase means that this cannot be a directed path that is a complement to the verb, as such a path is normally realised as a postpositional phrase. The prepositional phrase could still be a frame locative that is an adjunct to the verb and that denotes the location of the slashing. However, it is not possible to insert an adverb between een weg and the PP, suggesting that they form a constituent, as shown in (61a). The adverb has to be placed between the reflexive and the weg NP instead, as in (61b). (61) a. *Tarzan hakte zich een weg langzaam door T. slashed REFL a way slowly through the jungle b. Tarzan hakte zich langzaam een weg door T. de jungle. slashed REFL slowly de jungle. a way through the jungle ‘Tarzan slowly slashed his way through the jungle.’ The impossibility of inserting an adverb between een weg and the PP constitutes a fourth piece of evidence for the hypothesis that the entire weg NP denotes the path: the weg NP and the PP form a constituent. Therefore, the PP is not a sister of the verb, but an adjunct to the weg noun. The ungrammaticality of (61a) cannot be due to some rule that prohibits the insertion of an adverb between an NP and a PP in Dutch, 74 because this is very well possible. An adverb can intervene between an NP and a PP complement, as in (62a), as well as between an NP complement and a PP adjunct, as in (62b). (62) a. Hij zette de vaas voorzichtig op de plank. he put the vase carefully on the shelf ‘He put the vase carefully on the shelf.’ b. Zij vermoordde de man langzaam in het park. she murdered the man slowly in the park ‘She slowly murdered the man in the park.’ The fact that an adverb can be inserted between an NP complement of the verb and a PP in both examples indicates a constituent break. The PP op de plank in (62a) is the PP complement of the verb zetten ‘put’, whereas the PP in het park in (62b) is an adjunct to the verb vermoorden ‘murder’. No adverb can be inserted between the weg noun and the PP in the weg-construction, because the PP modifies the weg noun, not the verb. These pieces of evidence suggest that the een weg phrase and the PP form a constituent, which denotes both the path that is created and the path that is travelled. The weg-construction only has two NP complements: the reflexive NP and the weg NP, and thus instantiates the double object construction. Following Jackendoff and Goldberg’s representation of the way-construction, the weg-construction can be assigned the following skeletal structure (Dutch has SOV word order). (63) [SUBJ [VP [REFL][NP een weg PP] V]] This structure shows that the weg-construction has two NP complements, which are both sisters of the verb. This is represented in (63) as a flat, ternary branching structure. However, this is not an accepted structure according to many current syntactic theories (cf. Adger 2003, Radford 2004). The next section offers a Minimalist account of the weg-construction, which is strictly binary branching. 75 3.7.1 A Minimalist approach This section discusses the syntactic structure of the weg-construction. I will propose a Minimalist analysis of the weg-construction, based on the feature checking approaches outlined in Chomsky (2000, 2001) and Adger (2003). However, the Minimalist approach is basically lexicalist and consequently faces several problems in an analysis of the weg-construction. First of all, there is no generally accepted analysis of the double object construction. Secondly, the weg-construction is a construction, which has two complements that are not subcategorised for by the verb. Therefore, I make several assumptions in the analysis presented here, to account for the fact that the weg-construction contains two syntactic arguments whose case features need to be valued and checked, but which are not arguments of the verb. I do, however, not intend to solve the problem of the problematic issue of the double object construction. My sole aim is to present a syntactic structure for the weg-construction. In Minimalism, every NP has an uninterpretable case feature, which needs to be checked in order for the derivation to converge (cf. Chomsky 2000, 2001; Adger 2003 and the references therein).12 I will follow Adger (2003) in assuming that the verb phrase consists of a vP and a VP layer, which accommodate the external argument and internal argument(s), respectively. In Adger’s approach, case features have to be checked under c-command. The c-selectional properties of a verb are represented by uninterpretable category features on V and v (e.g. [uN], [uP]), which need to be checked through Merge with a phrase of the appropriate category (e.g. NP, PP). Transitive verbs are characterised by a v that has a strong uninterpretable N feature [uN], as well as the ability to check and value the accusative case feature of an NP complement. The strong [uN] feature of v triggers the Merge of an NP in the specifier of v. This NP is interpreted as an Agent or causer. The phrase structure in (64) shows the start of the derivation of the transitive sentence Anna kisses John.13 12 Most Minimalist analyses assume that noun phrases are DPs. However, I will call them NPs, in order to be consistent with the rest of this thesis, where I call the weg phrase an NP. Nothing hinges on this labelling, because the analysis holds regardless of whether we assume an NP or a DP analysis of noun phrases. 13 Since the checking of tense features and φ-features has no bearing on my analysis of the wegconstruction, I will only show case and c-selectional features in the phrase structure diagrams. 76 (64) TP 3 T′ 3 T vP [case:nom] 3 [uN] NP v′ Anna 3 [ucase: ] v VP [case:acc] 3 [uN] V NP kiss [uN] John [ucase: ] In (64), the transitive verb kiss has an uninterpretable N feature, so it has to Merge with an NP complement. The NP John checks the [uN] on the transitive verb kiss. The uninterpretable case feature [ucase: ] of the NP John is valued and checked by v, so John gets accusative case. The NP Anna first checks the uninterpretable N feature [uN] on v and subsequently moves up to TP to check the strong [uN] feature on T.14 The [ucase: ] of the NP Anna gets checked and valued by T. Adger adopts the Uniformity of Theta-Assignment Hypothesis (UTAH) (Baker 1988), which states that “identical thematic relationships between predicates and their arguments are represented syntactically by identical structural relationships when items are Merged” (2003:138). The UTAH motivates a different analysis for unergative and for unaccusative verbs: the subject of an unergative verb is an Agent, which is the daughter of vP, whereas the subject of an unaccusative verb is a Theme, which is the daughter of VP. The v of unaccusative verbs has neither a [uN] feature, nor the ability to check and value the accusative case feature of an NP. The phrase structure in (65) shows the start of the derivation for the sentence The snow melts with the unaccusative verb melt. 14 The [uN] feature of T is a purely syntactic EPP feature (Adger 2003:215), which is not associated with any thematic roles. When this feature is strong, as in English and Dutch, it must be checked under sisterhood. 77 (65) TP 3 T′ 3 T vP [case:nom] 3 [uN] v VP 3 V NP melt [uN] the snow [ucase: ] The uninterpretable case feature of the NP the snow cannot be valued and checked by v, as v does not have any case checking abilities. Therefore, the case feature of the snow can only be valued and checked by T, so this NP gets nominative case. Furthermore, since there is no external argument Merged in vP, the internal argument moves up to TP, to check the strong [uN] feature on T. Burzio (1986) noted that a verb that lacks an external argument fails to assign accusative case, and a verb which fails to assign accusative case fails to theta-mark the external argument (Burzio’s generalization, Burzio 1986:178-184). In Minimalism, Burzio’s generalization corresponds to the assumption that the presence of an accusative case feature on v is linked to the presence of a [uN] feature that triggers the Merge with an external argument. In (65), v does not have a [uN] feature or an accusative case feature. Adger and Chomsky (1995) assume that the case features on v and T are uninterpretable, i.e. they must always be checked, although Adger notes that “various linguists have found this idea unintuitive” (2003:212). However, the assumption that the accusative case feature on v is uninterpretable runs into problems with unergative verbs, as these verbs do not have an NP complement that can check the accusative case feature on v. If case features of v are uninterpretable, the derivation of an unergative sentence like Anna laughs should crash, because the accusative case feature of v cannot be checked. However, this sentence is well-formed; (65) shows the start of the derivation of Anna laughs. 78 (66) TP 3 T′ 3 T vP [case:nom] 3 [uN] NP v′ Anna 3 [ucase: ] v VP [case:acc] g [uN] V laugh In (66), the NP Anna first checks the [uN] feature on v and subsequently moves up to TP to check the [uN] on T. The uninterpretable case feature on the NP Anna gets checked and valued by T. Chomsky (2000, 2001) only discusses uninterpretable φ-features on T and v. He does not mention the uninterpretable case features on T and v anymore. Both Chomsky (2000) and Radford (2004) state that there is a correlation between valuedness and interpretability: uninterpretable features enter the derivation unvalued (Radford 2004:288). This can be interpreted to mean that, since the case features on T and v are valued, they must be interpretable. In order to account for the fact that Anna laughs is well-formed, I will assume that the case features on T and v are interpretable - that is, they can but they do not need to be checked. For example, the accusative case feature on the unergative v in (66) is not checked. However, the case feature is available to check and value a case feature of any NP it c-commands. I will argue below that for unergative verbs in the weg-construction, the accusative case feature of v is checked by the reflexive NP. The weg-construction is an instance of the double object construction. There exist many different analysis of the double object construction (cf. e.g. Oehrle 1976 for a ternary branching analysis, Chomsky 1981; Kayne 1984 for binary branching analyses and Larson 1988 for a binary shelled VP analysis). Adger (2003) does not commit himself to a particular analysis of the double object construction, but the analysis I suggest here is in keeping with his assumptions and terminology. The double object construction is problematic for Minimalism, because this lexicalist approach assumes that c-selectional features are subcategorised for by the verb. However, the verbs that enter the weg-construction are unergative and thus do not have any features. Therefore, I assume that the [uN] features on the verb in the weg-construction are contributed by the construction; they do not belong to the verb. 79 To account for the fact that the weg-construction has two NP complements, whose case features need to be checked and valued, I assume an additional VP shell, which has case checking abilities. Therefore, it shares some properties with the vP (the lexical head normally lacks structural case checking abilities). I will call the higher VP projection VP, to show its similarities with vP. I will assume that weg NP is merged as a complement of the lower V and the reflexive indirect object is Merged as the specifier of VP. The phrase structure of the weg-construction Tarzan hakt zich een weg door de jungle ‘Tarzan slashes his way through the jungle’ then looks as follows. (67) TP qp T′ qp vP T [case:nom] qp NP v′ [uN] Tarzan qp [ucase: ] VP v qp [case:acc] NP V [uN] zich qp [ucase: ] VP V qp [case:acc] NP V [uN] een weg door de jungle [ucase: ] [uN] hakt In this structure, the verb has two [uN] features, one on the lexical head and one on the additional VP shell. I assume that these features do not belong to the verb itself, but that they are contributed by the construction. These features check and value the case features of two NP complements of the construction, which are not arguments of the verb. The contribution of these features to the verb reflects the hypothesis made throughout this thesis: the weg-construction contains two NP complements that are not subcategorised for by the verb. In this analysis, V values and checks the accusative case on the NP een weg door de jungle, as it is the closest c-commanding case checker and valuer. v values and checks accusative case on the NP zich, and T values and checks nominative case on the NP Tarzan.15 The NP Tarzan will move up to TP to check the [uN] on T. To 15 In this analysis, both NP complements get valued accusative case. This is motivated by the fact that direct objects and indirect objects are homomorphic in Dutch. As in English, this can only be seen for 80 check its tense features (not shown), the main verb hakken ‘slash’ moves up via V to v to T, and ends up in C (not shown). In the Minimalist approach adopted here, unaccusative verbs do not have the ability to project a specifier or to check and value accusative case (i.e. their v does not have a [case:acc] and [uN] feature). I assume that any verb that enters the wegconstruction gets two [uN] features from the construction, which are checked by the weg NP and the reflexive. This would mean that the case feature of the reflexive NP remain unchecked and unvalued, causing the derivation to crash. However, there is no way that we can exclude the possibility that v of unaccusative verbs receives a [case:acc] and a [uN] feature from the construction. Consequently, in this approach, a structural account of the unacceptability of unaccusative verbs is not plausible. Rather, unaccusative verbs are semantically incompatible with the weg-construction because their subject is a Theme, not an Agent. 3.7.2 Conceptual structure of the weg-construction It was concluded in section 3.3 and 3.4 that the reflexive and the weg noun in the wegconstruction are non-referential. Consequently, they both do not appear in the CS of the weg-construction. The CS of the example in (67) can be represented as follows, where (68a) represents the atelic reading and (68b) the telic reading. (68) a. [Event GO ([Thing TARZAN], [Path VIA ([Place IN ([Thing JUNGLE])])]) [BY [Event HAKKEN ([Thing TARZAN])]]] b. [Event GO ([Thing TARZAN], [Path TO ([Place AT-END-OF ([Path VIA ([Place IN ([Thing JUNGLE])])])])]) [BY [Event HAKKEN ([Thing TARZAN])]]] The CSs in (68) represent the paraphrases of the weg-construction, namely that Tarzan is travelling through the jungle by means of slashing. GO is thus the superordinate event and the main verb is demoted to a subordinate clause. The first argument of GO is the subject Tarzan and the second argument is the Path. The lexical element een weg does not appear in the CS, as it is realised as the Path function. Likewise, the reflexive does not appear in the CS because it no longer has the meaning of beneficiary. The Path function both denotes the path that is created (by pronouns, as only pronouns have morphological case (cf. Ik zie hem ‘I see him’ and Ik geef hem het boek ‘I give him the book’). 81 slashing) and travelled (by slashing), because the creating and the moving are coextensive. The path is unbounded in the conceptual structure in (68a), so it does not have an endpoint. In (68b), the path is bounded and the goal of being at the end of the jungle is reached. This goal is represented by the Place function AT-END-OF, which is an argument of the Path function TO. The CS of a weg-construction that entails subjective motion (see section 3.2.3) does not involve the function GO, because nothing is moving. Instead, Jackendoff uses the State function EXT, which specifies the spatial extension of linear objects along a path (1990:44). The meaning of the weg-construction in (69a) can be represented as in (69b). (69) a. De rivier slingert zich een weg door het dal. ‘The river winds its way through the valley.’ b. [State EXT ([Thing RIVIER], [Path VIA ([Place IN ([Thing DAL])])]) [BY [Event SLINGEREN ([Thing RIVIER])]]] Although the conceptual structure of a subjective motion weg-construction does not involve the function GO, but rather the state function EXT, we still perceive it as denoting motion. As discussed in section 3.2.3, this is a result of our incrementally built-up perception. In sum, the main verb in the weg-construction is demoted to a subordinate means modifier in the CS of the construction. At the same time, a superordinate conceptual function GO is imposed on the CS of the verb. 3.8 The weg-construction as a constructional idiom In the preceding sections I argued that the Dutch weg-construction represents a mismatch between syntax and semantics: the verb has two fixed syntactic NP complements, but these are not semantic arguments of the verb. Moreover, the main verb shows up in a subordinate clause in the construction’s paraphrase. In addition, the ditransitive syntax of the weg-construction is unproductive: the double object construction cannot normally be used in Dutch when the direct object is created. Thus, although the verb is the syntactic head of the sentence, it does not determine the argument structure of the sentence. 82 These findings suggest that the meaning and the form of the weg-construction cannot be compositionally derived from its individual parts or from any existing rules in the Dutch language. Therefore, I propose that the weg-construction is a ‘constructional idiom’, that is, a specialized syntactic form with an idiomatic meaning (Jackendoff 1990). The Dutch weg-construction has the form in (70a), which has the meaning in (70b). Following Jackendoff (1990), optional functions are underlined with a dashed line. (70) a. TP qp T′ qp vP T qp [case:nom] NP v′ [uN] qp VP v qp NP REFL qp NP een weg PP [ucase: ] [case:acc] V′ [uN] qp VP V V [case:acc] [uN] [uN] b. [Event GO ([Thing ], [Path ([Place AT-END-OF ([Path ([Thing -----------------------------[BY/WITH [Event ([Thing ])]]] ])])])]) ---- The optionality of the Place function AT-END-OF in (70b) reflects the hypothesis that the path in the weg-construction can be both bounded and unbounded. The structure in (70a) shows that it is not the verb that determines the argument structure of the sentence, but the constructional idiom itself. The form and the meaning of the VP in (70) are fixed, which could mean that it is a kind of lexical item (cf. Jackendoff 1990:222). That is, the VP consists of a verb, a weak reflexive, een weg and a PP, which roughly has the meaning of ‘travel the weg path by means of doing V’ and which has to be stored in the lexicon as such. Returning to adjectival modifiers discussed in section 3.4.3 above, it is now not so surprising that these are understood as modifying the going event. Any direct modifiers of the VP are expected to turn up as modifiers of the GO Event, as this is the superordinate Event. 83 3.9 Some of Verhagen’s weg-construction verbs This section discusses some of the verbs that are included by Verhagen in the set of verbs that can occur in the weg-construction. I will argue that these verbs do not instantiate the weg-construction, because the reflexive and the weg NP are semantic arguments of the verb. That is, these sentences in fact instantiate the ditransitive pattern. 3.9.1 Zoeken ‘search’ and vinden ‘find’ Verhagen includes the verb zoeken ‘search’ in the set of verbs that he found in the weg-construction. When combined with a reflexive, een weg and a PP, this pattern formally looks like a weg-construction and it also expresses motion, as shown in (71). (71) Bezoekers zoeken zich met computers een weg door visitors het gebouw. search REFL with computers a way through the building ‘Visitors search their way through the building with computers.’ [Verhagen 2003:30, 5b, translation mine] The semantically related verb vinden ‘find’ can also be combined with the elements of the weg-construction, which then also implies motion, as shown in (72). (72) Een traan vindt zich een weg door haar wimpers naar beneden. a tear.drop finds REFL a way through her eyelashes to down ‘A tear drop finds its way down through her eyelashes.’ www.hetschrijvertje.be/index.php?page=jijvertelt&Verhaal_Id=23931&process=story In both examples, the reflexive cannot be replaced by another NP, as shown in (73). (73) a. *Bezoekers zoeken de mensen een weg door visitors search the people het gebouw. a way through the building b. *Een traan vindt de zandkorrel een weg naar beneden. a tear.drop finds the sand.grain a way to down The fact that the reflexive cannot be replaced with another NP in both examples suggests that it is not a semantic argument of the verb and that these example 84 instantiate the weg-construction. However, there are reasons to believe that these examples do not instantiate this construction. First of all, zoeken and vinden are strictly transitive, as shown in (74). (74) a. Piet zoekt *(een leuk cadeautje voor zijn vriendin). P. searches a nice present.DIM for his girlfriend ‘Piet searches a nice present for his girlfriend.’ b. Piet vindt *(geen leuk cadeautje voor zijn vriendin). P. finds no nice present.DIM for his girlfriend ‘Piet finds no nice present for his girlfriend.’ The fact that both zoeken and vinden strictly require the presence of a thematic object suggests that the weg NP in (71) and (72) is used as a semantic argument of the verb. Consequently, we would expect it to be exchangeable for another noun denoting a path or a spot. This prediction is borne out, as shown in (75). (75) a. Bezoekers zoeken zich een plaatsje in het gebouw. visitors search REFL a spot.DIM in the building ‘Visitors search themselves a spot in the building.’ b. Een traan vindt zich een route door haar wimpers naar beneden. a tear.drop finds REFL an route through her eyelashes to down ‘A tear drop finds itself an route through her eyelashes downwards.’ Furthermore, the reflexive in this pattern can also be omitted, as shown in (76). In that case, the weg NP can also be possessive, as shown in (77) from Verhagen. (76) a. Hij vindt een weg uit de hinderlaag van oordelen en zogenaamde wetten. he finds a way out the ambush of judgements and so-called laws ‘He finds a way out of the ambush of judgements and so-called laws.’ www.katholieknederland.nl/mediapastoraat/programmas/verkondigingen/index.html 85 b. Het riviertje werd gevormd door het overtollige water dat een weg naar the river.DIM was formed by the excess water that a way to zee zocht. (INL) sea searched ‘The stream was formed by excess water that was searching a way to the sea.’ (77) a. Veel kunst vindt via vlooienmarkten zijn weg naar de kopers. (2003:48, 36) much art finds via flea.markets his way to the buyers ‘A lot of art finds its way to the buyers via flea markets.’ b. Nina and Vladimir zoeken hun weg tussen de ruïnes van hun stad. N. and V. seek their way between the ruins of their city ‘Nina and Vladimir search their way between the ruins of their city.’ [2003:48, 38, translation mine] These examples show that the reflexive and the een weg phrase are not fixed elements, which confirms that the combination of zoeken or vinden with a reflexive, een weg and a PP does not instantiate the weg-construction. A fourth reason why zoeken and vinden cannot occur in the weg-construction is that the PP can also be omitted, as shown in respectively (78a) and (78b). (78) a. Deze rock huppelt niet op de bekende maat, maar zoekt zich een expressive this rock frolics not on the known rhythm but seeks REFL an expressive weg die de tijd indringend raakt. (INL) way that the time intrusive touches ‘This rock does not frolic around on the usual rhythm, but it seeks an expressive way that really touches time.’ b. Aan de hand van de tekst en de bijbehorende afbeeldingen en on the hand of the text and the accompanying pictures and schermvoorbeelden moet de lezer zich een weg vinden. screen.examples must the reader REFL a way find ‘Based on the text and the accompanying pictures and examples on the screen, the reader must find his way.’ winkel.bruna.nl/Boeken/Computerboeken_informatica_internet/Programmeertalen.htm 86 These observations suggest that the sentences in (71) and (72) above do not instantiate the weg-construction. Instead, the weg NP is a thematic object of the verb in these examples, which can be replaced by another noun and which is not obligatorily indefinite. Since these sentences contain a reflexive in indirect object position, they must instantiate the ditransitive pattern. However, this pattern is unproductive in Dutch when the direct object is created. This either means that zoeken and vinden can exceptionally occur in the ditransitive pattern, or that the direct object is not created, but it already exists. I think the latter option is the case, because one cannot create a weg by means of zoeken or vinden. Therefore, the weg is already there, and all the subject has to do is find it. I conclude that the vinden/zoeken + reflexive + een weg + PP string does not instantiate the weg-construction, but is an instance of the ditransitive pattern. This pattern conforms to the constraint posited by Verhagen (2002:415) that the direct object in the ditransitive pattern in Dutch cannot be created. That is, for zoeken and vinden, the weg already exists. 3.9.2 Verschaffen ‘provide’ Verhagen (2004) also includes the verb verschaffen ‘provide’ amongst verbs that can be found in the weg-construction. When combined with een weg and a PP, this pattern seems to instantiate the weg-construction, as shown in (79). (79) Toen de moeder weigerde naar buiten te komen, probeerde zoonlief zich met when the mother refused to outside to come tried son.dear REFL with geweld een weg naar binnen te verschaffen. (INL) violence a way to inside to provide ‘When mother refused to come outside, her dear son violently tried to provide himself a way inside.’ This sentence looks like a weg-construction because it contains a weak reflexive, een weg and a PP, and it expresses motion. However, there are reasons to believe that this 87 pattern does not instantiate the weg-construction. First of all, both the weg noun and the reflexive can be replaced, as shown in (80).16 (80) a. Zoonlief probeerde haar toegang naar binnen te verschaffen. son.dear tried her.DAT access to inside to provide ‘Sonny tried to provide her access inside.’ b. Zoonlief probeerde zich toegang naar binnen te verschaffen. son.dear tried REFL access to inside to provide ‘Sonny tried to provide himself access inside.’ The fact that both the reflexive and the weg NP are exchangeable for other NPs indicates that they are semantic arguments of the verb.17 In fact, verschaffen appears to be strictly ditransitive, as shown in (81), where omission of one or two of the thematic objects is unacceptable. (81) Zoonlief probeerde *(zich) *(toegang) te verschaffen. son.dear tried REFL access to provide ‘Sonny tried to provide himself access.’ The fact that verschaffen is strictly ditransitive and that both the reflexive and weg can be replaced with another noun suggests that (79) does not instantiate the wegconstruction, but is an instance of the ditransitive pattern. I concluded in section 3.2.1 that strictly ditransitive verbs are incompatible with the weg-construction because they cannot be used intransitively. Another reason this sentence does not instantiate the weg-construction is because the verb does not designate the means by which the path is created (or an activity that accompanies the motion). Since one cannot move ‘by means of’ or ‘while’ verschaffen, this verb is also semantically odd in the wegconstruction. I conclude, contra Verhagen, that the verschaffen + reflexive + een weg string does not instantiate the weg-construction, because verschaffen is a strictly ditransitive verb. 16 In fact, I found no instances of zich een weg verschaffen with Google, whereas zich toegang verschaffen resulted in 217 hits. 17 Again, although the reflexives in (79) and (80b) are semantic arguments of the verb, they are weak reflexives. This will be discussed in more detail in the next chapter. 88 3.10 Summary The Dutch weg-construction consists of a verb, a weak reflexive, een weg and a PP. In the Minimalist approach adopted here, the weg-construction can be skeletally represented as follows. (82) [vP SUBJ [[VP REFL [[VP [NP een weg PP] V] V] ]v]] All the elements in (82) are obligatorily present, but none of them besides the subject are semantic arguments of the verb. Moreover, all the elements are fixed: the reflexive is obligatorily weak, the weg phrase has to be indefinite and cannot be replaced with another noun denoting a path, and the PP is necessarily a prepositional phrase. The meaning of the weg-construction cannot be predicted from its individual parts, because none of the parts entails motion, yet the construction itself denotes motion along a path. This path can be both literal and metaphorical and is created by removing obstacles or overcoming other barriers, by means of the action denoted by the verb. Consequently, the weg-construction implies motion despite external difficulty, which requires effort. In some dialects, the verb can also denote an activity that accompanies the motion. The constraints on the verb that can be found in the weg-construction are the same as those for the verb in the English way-construction. The verb has to be conceivable as a process, which means that it has to be an activity verb or else denote a repeatable bounded event. This constraint rules out verbs that do not have any internal structure, such as stative verbs, plain motion verbs, and verbs that denote an unrepeatable bounded event. When the verb denotes an event that is repeatable, the weg-construction strongly implies a repetition of that event. Another semantic constraint is that the action denoted by the verb has to be volitional, self-propelled and directed - in other words, the subject has to be an Agent. This constraint rules out unaccusative verbs, as the subject of these verbs is not an Agent but a Theme. Manner of motion verbs also occur in the weg-construction (though not for all speakers), but these verbs do not denote the means by which a path is created, but only the means by which a path is travelled. The weg-construction with a manner of motion verb implies a longer path and more external difficulty than its bare motion counterpart. Finally, the verb that enters the weg-construction must be intransitive, which rules out strictly 89 (di)transitive verbs and inherently reflexive verbs. Since the subject of the intransitive verb has to be an Agent, the verb in the weg-construction is unergative. The weg NP and the PP form a constituent that denotes both the path that is created and the path that is travelled. One constituent suffices to denote both the created path and the travelled path, because these events are the same event. I propose that the PP is an adjunct to the weg noun that specifies the trajectory of the path, so it is headed by a Path preposition. The weg noun is not referential in the wegconstruction, which is confirmed by adjectival modifiers on the weg noun, which do not modify the path but the going event. Consequently, the weg noun does not show up in the conceptual structure of the weg-construction, because it is realised as the Path function. The weg-construction is an instance of the ditransitive double object construction, but this pattern is not productive in Dutch. Therefore, the reflexive in the weg-construction is also non-referential. Verhagen explains the unusual syntax of the weg-construction by looking at old Dutch, where the ditransitive pattern was fully productive. The weg-construction may have originated from the verb banen in the ditransitive pattern. This verb is the most common verb in the weg-construction, but it has lost its meaning and only occurs idiomatically in the modern weg-construction. The weg-construction is a ‘constructional idiom’ (Jackendoff 1990), which is a kind of lexical item with a fixed form and meaning, which cannot be compositionally derived from its individual parts or from rules of syntax in Dutch. Therefore, the wegconstruction is stored in the lexicon and has to be learnt by speakers of Dutch. A Minimalist approach was adopted in this thesis to represent the structure of the weg-construction. I extended this approach to include constructions, which contain syntactic arguments that are not semantic arguments of the verb. Therefore, I assumed that the [uN] features on the verb are contributed by the construction. Furthermore, to account for the problematic fact that the weg-construction is a double object construction, I assumed the presence of an additional VP shell. This was taken to be a semi-functional head with case checking abilities, which values and checks the case features of the reflexive NP. 90 Chapter 4 The Transition to Location Construction in Dutch 4.1 Introduction This chapter introduces a second type of way-construction in Dutch, called the Transition to Location Construction (TLC). This construction consists of a verb, a weak reflexive and a PP, as illustrated in (1). (1) a. Marien zwemt zich in de finale. M. swims REFL in the final ‘Marien swims his way into the final.’ b. Paul bluft zich uit de benarde situatie. P. bluffs REFL out the awkward situation ‘Paul bluffs his way out of the awkward situation.’ c. Christine slaat zich C. door de eerste ronde. hits REFL through the first round ‘Christine hits her way through the first round.’ The example in (1a) means that Marien gets to be in the final by means of swimming, (1b) means that Paul gets to be out of the awkward situation by means of bluffing and (1c) denotes that Christine gets to be through the first round by means of hitting, say, tennis balls. The TLC is similar to the weg-construction in many ways. First of all, both constructions entail that the subject goes or gets to be somewhere by means of the action denoted by the verb, although the verb does not have to be a motion verb. Second, all elements are obligatorily present in both constructions, but none of them besides the subject is a semantic argument of the verb. Thus, like the weg-construction, the TLC violates the argument structure of the verb, as it has two syntactic complements that are not selected by the verb. Third, the reflexive in the TLC is also obligatorily weak and the PP has to be realised as a prepositional phrase as well, not as a postpositional phrase. Fourth, as will be shown below, the constraints on the verb are the same in both 91 constructions. Finally, both Dutch constructions are translated with a way-construction in English. However, there are also important differences between the two Dutch constructions. The most obvious difference is their syntax, as the TLC lacks the een weg phrase and is transitive, whereas the weg-construction is ditransitive. Moreover, the constructions have a different meaning, which will be discussed in detail in section 4.2. The weg-construction denotes motion along a path, but the TLC describes a transition to a location, which does not involve the traversal of a path - and consequently no motion either. I will argue that the TLC describes a complex event, which consists of two distinct subevents that are not co-extensive, whereas the weg-construction has a simple event structure, consisting of two co-identified subevents. The remainder of this section discusses the similarities between the two Dutch constructions. I will show that, apart from the absence of the een weg phrase in the TLC, the two constructions look the same on the surface. Section 4.2 describes the semantics of the TLC, and demonstrates the differences with the semantics of the weg-construction. Section 4.3 discusses the different event structures of the TLC and the weg-construction. Section 4.4 offers a Minimalist analysis of the structure of the TLC, and section 4.5 finishes with some potential counterexamples to the analysis presented in this paper. 4.1.1 The verb More than 120 different verbs were found in the TLC in the INL corpus and by searching with Google, which are listed in appendix 6. The majority of these verbs (86%) were also found in the weg-construction using the same corpora. As in the weg-construction, verbs of any valence are admitted into the TLC, as illustrated in (2). Springen ‘jump’ in (2a) is normally intransitive, spelen ‘play’ in (2b) can be both transitive and intransitive and slaan ‘hit’ in (2c) is normally transitive. 92 (2) a. Carl Lewis springt zich in de geschiedenis met een vierde gouden medaille … C. L. jumps REFL in the history with a fourth golden medal ‘Carl Lewis jumps his way into history with a fourth gold medal.’ www.vtm.be/spelen/index_ historisch_detail.htm?jaartal=23 b. Joyce heeft zich in de basis gespeeld van de nationale korfbalploeg. (INL) J. has REFL in the base played of the national korfball.team ‘Joyce played her way into the base of the national korfball team.’ c. Staphorster duo S. tapt zich naar landelijke finale. (INL) duo draw.beer REFL to national final ‘Duo from Staphorst makes its way to the national final by drawing beer.’ Like the weg-construction, the verb in the TLC is used intransitively. The examples in (3) show that the presence of a thematic object of the verb in direct object position is unacceptable. (3) a. *Christine slaat tennisballen zich C. hits tennis.balls REFL through the first round b. *Staphorster duo S. door de eerste ronde. (cf. 1c) tapt bier zich naar landelijke finale. (cf. 2c) duo draw.beer beer REFL to national final I propose that the presence of a semantic argument of the verb in direct object position is unacceptable because this position is occupied by the reflexive. However, the reflexive is not a semantic argument of the verb, as will be discussed in more detail in the next section. The fact that the verb is used intransitively in the TLC can account for the unacceptability of strictly (di)transitive verbs and reflexive verbs in this construction. These verbs strictly require the presence of one of two thematic objects, or, in the case of reflexive verbs, a reflexive. Consequently, these verbs cannot be used intransitively. The corpus searches did not reveal any instance of the TLC with a strictly (di)transitive or reflexive verb. The previous chapter showed that verwoesten ‘destroy’ and geven ‘give’ 93 are strictly transitive and strictly ditransitive, respectively. These verbs are infelicitous in the TLC, as shown in the following examples. (4) a. ??De olifant verwoest zich door het gewas. the elephant destroys REFL through the crops b. ??Rudy geeft zich in het Guinness Book of Records. R. gives REFL in the G. B. of R. The verbs in these examples strictly require the presence of a thematic object, so the reflexive can only be interpreted as a semantic argument of the verb. That is, the elephant is destroying himself in (4a) and Rudy is giving something to himself in (4b). This is not only semantically odd, but it is also ungrammatical because these reflexives are a semantic argument of the verb and need to be marked with zelf. Reflexive verbs are syntactically strictly transitive because they require the presence of a reflexive. These verbs are unacceptable in the TLC, as shown in (5) for REFL vergissen ‘err’ and REFL verontschuldigen ‘apologise’.1 (5) a. ??De minister vergiste zich uit het kabinet. the minister erred REFL out the cabinet ‘The minister erred his way out of the cabinet.’ b. ??De koningin verontschuldigde zich in de harten van het volk. the queen apologised REFL in the hearts of the people ‘The queen apologised her way into the hearts of the people.’ The PPs of these reflexive verbs can only be interpreted (however implausibly) as frame locatives - that is, these sentences cannot be interpreted as a TLC. The previous chapter argued that the verb in the weg-construction cannot be stative. This constraint seems to hold for the TLC as well: I found no examples of the 1 Note that the English counterparts of the Dutch verbs are acceptable in the way-construction. The acceptability of the English non-reflexive verbs in the way-construction confirms that the Dutch reflexive verbs are not infelicitous in the TLC for semantic reasons. Instead, they are unacceptable because they are strictly transitive. 94 TLC with a stative verb in the corpora, and a stative verb sounds odd in this construction, as shown in (6). (6) a. *John weet / lijkt / is / stinkt zich naar de top. J. knows seems is stinks REFL to the top b. *Marieke amuseert / verveelt / slaapt zich M. amuses bores door het toneelstuk. sleeps REFL through the play Stative verbs cannot be conceived as the means by which a location is reached, so they are semantically incompatible with the TLC. As in the weg-construction, the subject of the TLC must be an Agent, who performs a volitional and self-propelled action. The agentivity constraint can account for the unacceptability of unaccusative verbs in the TLC, as illustrated in the following examples. (7) a. *De zanger stierf zich naar de status van popidool. the singer died REFL to the status of pop.idol b. *Het raam breekt / opent zich in de kamer. the window breaks opens REFL in the room c. *De boter smelt zich van de hete kalkoen. the butter melts REFL from the hot turkey d. *De zonnebloemen groeien zich naar het licht. the sunflowers grow REFL to the light These examples are unacceptable because their subjects are Themes, not Agents. The unacceptability of unaccusative verbs in the TLC is confirmed by the results of the first questionnaire, which included the sentences in (7c) and (7d). These sentences were approved of by only two of the 22 speakers (see appendix 1).2 2 In the previous chapter I argued that the verb groeien ‘grow’ is polysemous in Dutch, i.e. it can be unaccusative as well as unergative, which may explain why about a third of the participants accepted the weg-construction version of the TLC in (7d). The reason why a TLC with groeien is unacceptable may be 95 The constraint that the subject has to be an Agent implies that it has to be animate. However, the subject of the TLC can be inanimate as well, in which case it is personified and interpreted as an Agent, as in the following example. (8) … hoe de roest zich door een chassisbalk heeft gevreten. how the rust REFL through a chassis.beam has eaten ‘How the rust has eaten its way through the chassis.’ www.lotusspecialist.nl/Werkplaats/ Restauratie/restauratie.htm The rust is portrayed as an animate entity in this example, who performs a volitional, controlled and self-propelled action, and who therefore is an Agent. The conclusion that unaccusative verbs are unacceptable in the TLC is confirmed by unaccusative verbs that have an agentive transitive alternant. The intransitive version of this agentive alternant is acceptable in the TLC, as shown in (9). (9) a. De lente smelt zich door de koude winter. the spring melts REFL through the cold winter ‘The spring melts its way through the cold winter.’ www.galerienovia.nl/galerie/ galerie.asp?Type=2&Schilderij=12 b. De Rijn breekt zich hier door een leisteenplateau ... the R. breaks REFL here through a slate.plateau ‘The Rhine here breaks its way through a slate plateau.’ www.farosreizen.nl/location.aspx?LnID=56&LTyp=21&BrCr=00P04205iP042194i&TxID The spring in (9a) is the Agent of the melting, because the spring is melting the winter snow. Likewise, the Rhine in (9b) is the Agent of the breaking, not the Theme. Although inanimate, the subject referents in both examples are interpreted as performing a volitional, controlled and self-propelled action. The verb in these examples is thus used unergatively, which is confirmed by the auxiliary that is selected. Like the wegconstruction, the TLC takes the HAVE-auxiliary, as shown in (10). because the PP naar het licht ‘to the light’ refers to a path. However, I will argue in section 4.2 that the meaning of the TLC does not entail a path. 96 (10) a. De lente heeft / *is zich the spring has door de winter gesmolten. is REFL through the winter melted ‘The spring has melted its way through the winter.’ b. De Rijn heeft / *is zich the R. has door het leisteenplateau gebroken. is REFL through the slate.plateau broken ‘The Rhine has broken its way through the slate plateau.’ The selection of HAVE instead of BE confirms that the verb in the TLC is used unergatively, and that its subject is an Agent. In sum, the constraints on the verb in the TLC are the same as for the verb in the weg-construction: the verb has to be used intransitively, which rules out strictly (di)transitive verbs and reflexive verbs. Semantic constraints include that the verb cannot be stative and its subject must be an Agent who performs a volitional and self-propelled action. The latter constraint rules out unaccusative verbs in the TLC. 4.1.2 The reflexive element The NP complement in the TLC is a reflexive, which has to agree with the subject in both person and number, as shown in (11a). The ungrammaticality of (11b) shows that the reflexive cannot be replaced with another NP. (11) a. *Marie zwemt M. mij naar de eerste plaats swims REFL.1.sg to the first place b. *Marie zwemt baantjes naar de eerste plaats M. swims laps to the first place The reflexive in the TLC is obligatorily weak and cannot be omitted, as demonstrated in (12a) and (12b), respectively. (12) a. Carl Lewis sprong zich(*zelf) in het Guinness Book of Records. C. L. jumped REFL.self in the G. B. of R. ‘Carl Lewis jumped his way into the Guinness Book of Records.’ 97 b. Carl Lewis sprong in het Guinness Book of Records. C. L. jumped in the G. B. of R. ‘Carl Lewis jumped in the Guinness Book of Records.’ The example in (12a) shows that a strong reflexive is unacceptable in the TLC. The sentence in (12b) is grammatical, but it has a different meaning: in the absence of a reflexive, the PP can only be interpreted as a frame locative that denotes the location of the jumping. Therefore, this sentence does not instantiate the TLC. There are several pieces of evidence for the claim that the reflexive in the TLC is not a thematic object of the verb (cf. Toivonen’s 2002:323-5 discussion of the Directed Motion Construction in Swedish). First of all, the reflexive cannot be replaced with another object, as was shown in (11) above. Secondly, as demonstrated in the previous section, the TLC often involves an intransitive verb, which normally does not take any semantic arguments besides a subject. Third, the reflexive object of transitive verbs is normally a strong reflexive, as shown in (13a). In the TLC, however, these verbs obligatorily take a weak reflexive, as in (13b). (13) a. De vrouw schopt zich*(zelf). the woman kicks REFL.self ‘The woman is kicking herself.’ b. De vrouw schopt zich(*zelf) naar de top. the woman kicks REFL.self to the top ‘The woman kicks her way to the top.’ The fact that the reflexive in the TLC has to be weak confirms that it is not a semantic argument of the verb. Fourth, the activity performed by the subject may be directed towards somebody or something other than the antecedent. For example, in (13b), the woman is not kicking herself but rather something else, like her competing colleagues. The ‘sloppy identity’ test provides a fifth piece of evidence that the reflexive in the TLC is not a thematic object of the verb. As discussed in the previous chapter, Sells, Zaenen & Zec (1987) argue that a sentence with a reflexive direct object that is a 98 semantic argument of the verb has three possible readings: a sloppy identity reading, a strict identity reading and an object comparison reading (see chapter 3 for a full discussion). If the reflexive in the TLC were a thematic object of the verb, we would expect it to have three readings. However, the construction only has one possible reading. Consider the following example. (14) De vrouw schopte zich sneller naar de top dan haar collega’s. the woman kicked REFL faster to the top than her colleagues ‘The woman kicked her way to the top faster than her colleagues.’ This sentence can only mean that the woman was faster in reaching the top by means of kicking than her colleagues were, which is the sloppy identity reading. Finally, the reflexive in the TLC is incompatible with the passive. A reflexive that is a thematic object of the verb can (at least marginally) be passivized by including it in a PP adjunct, as shown in (15). Compare this to (16), which gives the ungrammatical passive of the TLC in (14) (where I left out the comparative clause). (15) a. De vrouw verbaast zichzelf. the woman surprises REFL.self ‘The woman surprises herself.’ b. De vrouw werd (door zichzelf) verbaasd. the woman was by REFL.self surprised ‘The woman was surprised (by herself).’ (16) *De vrouw / zij werd naar de top geschopt (door zich). the woman she was to the top kicked by REFL These pieces of evidence indicate that the reflexive in the TLC is not a semantic argument of the verb. 99 4.1.3 The PP The same prepositions were found heading the PP in the TLC as in the weg-construction. Table 4.1 lists the prepositions that were found in the INL corpus. Table 4.1 Prepositions heading the PP of the TLC (INL corpus) Preposition naar ‘to’ door ‘through’ in ‘in’ terug ‘back’ uit ‘out’, van ‘from’ bij ‘at, with’, binnen ‘inside’, onder ‘under’, Number of hits / P 26 25 19 13 3 % of hits / P 27.7 26.6 20.2 13.8 3.2 1 1.1 over ‘across’, tussen ‘between’ Total number of hits 94 The three most common prepositions in the TLC are naar ‘to’, door ‘through’ and in ‘in’. The first two were also the two most common prepositions in the weg-construction. The same prepositions were also found in the TLC with Google in the same proportions; others included langs ‘past’, omhoog ‘up’, van ‘from’ and voorbij ‘past’. The PP can be headed by an intransitive preposition as well, such as terug ‘back’ and omhoog ‘up’. No instances of the TLC were found where the PP was realised as a postpositional phrase. 4.1.4 Summary This section showed that the individual elements of the weg-construction and the TLC are the same: both constructions accept a large variety of activity verbs with an agentive subject, and they do not allow stative verbs, unaccusative verbs and verbs that cannot be used intransitively. Both constructions select the HAVE-auxiliary, they contain superficially the same prepositions and their reflexives are both obligatorily weak. Furthermore, none of their elements besides the subject is a semantic argument of the verb. However, the two constructions differ significantly in their meaning. Consequently, the seemingly identical prepositions in fact belong to conceptually 100 different categories. Moreover, the reflexive and the PP in the TLC are in structurally different positions than in the weg-construction. The remainder of this chapter is devoted to the semantic and syntactic differences between the weg-construction and the TLC. 4.2 Semantics of the TLC The weg-construction can be paraphrased as ‘traverse the weg path by means of V-ing’, but the TLC has a different meaning. Consider again the example in (1a) above, repeated here as (17). (17) Marien zwemt zich in de finale. M. swims REFL in the final ‘Marien swims his way into the final.’ This sentence can be paraphrased as ‘Marien gets to be in the final as a result of his swimming’. This sentence does not mean Marien is moving into the final along a path by means of swimming, because a directed path is realised as a postpositional phrase in Dutch, as shown in (18a). A prepositional phrase normally denotes the location where the action described by the verb takes place, as in (18b). (18) a. ?Marien zwemt de finale in. M. swims the final in ‘Marien swims into the final.’ b. Marien zwemt (in de finale). M. swims in the final ‘Marien is swimming (in the final).’ The directed motion sentence in (18a) may be a bit odd because it can only mean that Marien literally swims into the final, where the final is interpreted as a location that is the end of a path (e.g. a separate lane in the swimming pool where the finalists are swimming). The prepositional phrase in (18b), on the other hand, is a frame locative that 101 denotes the location of the swimming, which can also be omitted. This sentence cannot mean that Marien is swimming into the final, but only that he is already in the final. The verb in the TLC does not have to be a manner of motion verb. Consider again the example in (1b) above, repeated here as (19). (19) Paul bluft zich uit de benarde situatie. P. bluffs REFL out the awkward situation ‘Paul bluffs his way out of the awkward situation.’ This sentence can be paraphrased as ‘Paul gets to be out of the awkward situation as a result of his bluffing’. It cannot mean that Paul is moving out of the awkward situation along a path, because bluffen is not a motion verb, and also because the location where Paul ends up is realised as a prepositional phrase. The sentence is unacceptable with a postpositional phrase, as shown in (20a). The sentence is also unacceptable when the reflexive is omitted, as shown in (20b). (20) a. *Paul bluft (zich) de benarde situatie uit P. bluffs REFL the awkward situation out b. *Paul bluft uit de benarde situatie. P. bluffs out the awkward situation These examples are ungrammatical because bluffen is not a motion verb. The postpositional phrase in (20a) denotes the directed path of motion and the prepositional phrase in (20b) is a frame locative headed by a Path preposition. Both phrases are incompatible with the non-motion verb bluffen. In sum, the prepositional phrase in the TLC refers to a location that is reached, but a prepositional phrase combined with a verb can normally only be interpreted as a frame locative. A frame locative does not denote a location that is reached, but only the location where the action of the verb takes place. A location that is reached, i.e. that is the end of a path, is realised as a postpositional phrase in Dutch. 102 I propose that the TLC does not denote motion along a path, but a transition to a location, which does not involve the traversal of a path. There are several pieces of evidence for this proposal, which are discussed individually below. 4.2.1 Temporal relation between the subevents 3 The weg-construction and the TLC both describe two events, namely the event represented by the verb and the one represented by the weg NP/PP. However, the temporal relation between these subevents is different in the two constructions. Recall that the weg-construction denotes motion along a path, which is created by means of the action described by the verb. For example, in Tarzan hakt zich een weg door de jungle ‘Tarzan slashes his way through the jungle’, Tarzan is moving by means of slashing, so the slashing subevent is co-extensive with his progress through the jungle. That is, the two events unfold at the same rate and they are temporally dependent: with every slash, Tarzan progresses further through the jungle. When he stops slashing, he will stop progressing. The slashing event and the progress along the path also take place at the same location, as the path is created and travelled by means of slashing. By contrast, the two subevents in the TLC are not necessarily co-extensive. For instance, in the TLC in (17) Marien zwemt zich in de finale ‘Marien swims his way into the final’, there is a swimming subevent and a getting into the final subevent, but these are not necessarily temporally overlapping or contiguous. This is because Marien may have finished swimming before he reaches the final. For example, he may have to wait for his competitors to swim their laps before he knows that he has reached the final. Furthermore, the two subevents may also take place at different locations, because Marien may swim his laps in the swimming pool and learn that he has reached the final when he is outside. Thus, in contrast to the weg-construction, the two subevents described by the TLC can be temporally disjoint. 3 The argumentation in this subsection is largely based on L&RH (1999) and Rappaport Hovav & Levin’s (2001) event structure account of the resultative construction in English. Their account will be discussed in section 4.3.1. 103 4.2.2 Gradualness The progress along the path expressed by the weg-construction evolves gradually, or incrementally. For example, in Tarzan hakt zich een weg door de jungle, Tarzan moves through the jungle in incremental steps, where with every slash he extends his path through the jungle a little bit further. Similarly, in for example John schopt zich een weg door de menigte ‘John kicks his way through the crowd’ John moves by means of kicking, so with every kick he progresses a little bit further through the crowd. The wegconstruction therefore strongly implies a repetition of the action denoted by the verb, so the above examples imply multiple slashes and multiple kicks. By contrast, the location in the TLC is not reached gradually or incrementally. Consider again the example in (12a), repeated here as (21). (21) Carl Lewis sprong zich in het Guinness Book of Records. C. L. jumped REFL in the G. B. of R. ‘Carl Lewis jumped his way into the Guinness Book of Records.’ This sentence does not mean that Carl Lewis is moving into the Guinness Book of Records along an incremental path by means of jumping, that with every jump he progresses further along the path that leads into the Guinness Book of Records. This is because he could reach the Guinness Book of Records after he finishes jumping, but also because this sentence can refer to a single jump. Carl Lewis could perform one spectacular jump and consequently end up in the Guinness Book of Records. The TLC in (21) and the weg-construction example in (22) were included in the third questionnaire, and the participants were asked whether these sentences referred to several jumps or if it could also be one jump. (22) De atlete springt zich een weg naar de finish. the athlete.FEM jumps REFL a way to the finish ‘The athlete jumps her way to the finish.’ 104 Of the 30 speakers, 87% thought the TLC in (21) could possibly refer to a single jump, and 63% of them thought the weg-construction in (22) necessarily refers to several jumps (see appendix 3). These results indicate that the TLC does not necessarily imply the repetition of a bounded event. 4.2.3 Telicity The telicity of the TLC provides another piece of evidence for the proposed difference in meaning between the two Dutch constructions. The previous chapter showed that the weg-construction can have both a telic and an atelic reading. The TLC, however, is necessarily telic. The standard telicity test of combining the sentence with a non-durative and a durative time adverbial is applied in (23). (23) a. Marien heeft zich * minutenlang / in twee minuten in de finale gezwommen. M. has REFL minutes.long in two minutes in the final swum ‘Marien swam his way into the final *for minutes/in two minutes.’ b. Christine heeft zich *urenlang / in drie C. uur door de eerste ronde geslagen. has REFL hours.long in three hours through the first round hit ‘Christine hit her way through the first round *for hours/in three hours.’ The example in (23a) was present in the second questionnaire as two separate sentences (see appendix 2). None of the 42 participants approved of the TLC with the durative time adverbial minutenlang ‘for minutes’, whereas 86% of the speakers approved of the TLC with the non-durative adverbial in twee minuten ‘in two minutes’. The TLC in (23b) is also unacceptable with the durative adverbial urenlang ‘for hours’. This TLC can only be combined with durative time adverbials like in drie uur ‘in three hours’. These data suggest that the TLC is telic, which confirms the hypothesis that the TLC does not denote motion along a path. 4.2.4 Direct predication of the PP of the subject A PP that denotes a stative location can be predicated directly of the subject in Dutch, as shown in (24). 105 (24) a. Het meisje loopt in het bos. the girl walks in the forest ‘The girl is walking in the forest.’ b. Het meisje is in het bos. the girl is in the forest ‘The girl is in the forest.’ By contrast, a path cannot be predicated directly of the subject, as shown in (25). (25) a. Het meisje loopt the door het bos / het bos door. girl walks through the forest the forest through ‘The girl is walking through the forest (atelic) / through the forest (telic).’ b. *Het meisje is door the het bos / het bos door. (with the path reading) girl is through the forest the forest through The PP of the TLC can be predicated directly of the subject, as shown in the following examples. (26) a. Een echtpaar uit Kempen heeft zich a couple from K. door de selekties gewerkt … (INL) has REFL through the selections worked ‘A couple from Kempen has worked its way through the selections.’ b. Het echtpaar is the door de selecties. couple is through the selections ‘The couple is through the selections. (27) a. Vitesse knokte zich uitstekend terug in de wedstrijd … V. fought REFL excellent back in the game ‘Vitesse excellently fought its way back in the game.’ vitesse.nl/newsitem.asp?Id=118 106 b. Vitesse was terug in de wedstrijd. V. was back in the game ‘Vitesse was back in the game.’ By contrast, the PP in the weg-construction cannot be directly predicated of the subject, as shown in (28) and (29), which contain the same verbs and prepositions as the TLC examples above. (28) a. Het echtpaar werkt zich een weg the door de menigte. couple works REFL a way through the crowd ‘The couple works its way through the crowd.’ b. *Het echtpaar is the door de menigte. (with the path reading) couple is through the crowd (29) a. De voetballer knokte zich een weg terug over het veld. the soccer.player fought REFL a way back over the field ‘The soccer player fought his way back across the field.’ b. *De voetballer was terug over het veld. the soccer.player was back over the field The fact that the PP in the TLC can be predicated directly of the subject indicates that it denotes a stative location, and not a path. By contrast, the PP in the weg-construction cannot be predicated directly of the subject, confirming that it refers to a path. The hypothesis that the same preposition can denote both a Place and a Path does not constitute a problem because, as is well known, many prepositions can alternate in their senses (cf. e.g. Talmy 1985; Jackendoff 1990). For instance, Jackendoff gives the following examples of the Path and the Place reading of the preposition through (1990:73). (30) a. Bill ran through the tunnel. b. Bill’s house is through the tunnel (from here). 107 The PP through the tunnel in (30a) denotes the Path of motion, which is dynamic, but the same PP denotes a stative Place in (30b). The CS of through as a Path preposition and as a Place preposition can be represented as in (31a) and (31b), respectively (Jackendoff 1990:74). (31) a. [Path VIA ([Place IN ([Thing TUNNEL])])] b. [Place AT-END-OF ([Path VIA ([Place IN ([Thing TUNNEL])])])] The Path preposition through in (31a) (which is represented in Jackendoff’s terms as the Place function IN embedded in the Path function VIA), functions as a Place preposition in (31b), where it is an argument of the Place function AT-END-OF. I conclude that the PP in the TLC is headed by a Place preposition; typical Path prepositions such as naar ‘to’ and door ‘through’ are used as Place prepositions in the TLC. By contrast, the preposition that heads the PP in the weg-construction is a Path preposition. This contrast is illustrated in (32), where (32a) represents the CS of the PP in the TLC in (26a) above that is headed by door. The CS in (32b) represents the PP in the weg-construction in (28a), which is headed by the same preposition. (32) a. [Place AT-END-OF ([Path VIA ([Place IN ([Thing SELEKTIES])])])] b. [Path VIA ([Place IN ([Thing MENIGTE])])] The hypothesis that the PP in the TLC is headed by a Place preposition is supported by the fact that the Place preposition in ‘in’ is one of the three most common prepositions in the TLC, whereas it is very rare in the weg-construction. 4.2.5 Means and manner interpretation The previous chapter showed that for some speakers, the verb in the weg-construction can also denote an action that accompanies the motion of the subject referent, instead of the means by which a path is created and/or travelled. This conclusion was supported by a few instances in the corpus that had a likely manner interpretation. 108 The corpus searches did not reveal any instances of the TLC with a likely manner interpretation. I propose that this is to be expected, as the meaning of the TLC does not entail motion. Consequently, the verb cannot be interpreted as an activity accompanying the motion, but only as the means by which the location is reached. For example, in Gordon zingt zich in de top 40 ‘Gordon sings his way into the top 40’, Gordon gets to be in the top 40 as a result of his singing. It cannot mean that he gets into the top 40 by doing something else and the singing only accompanies this achievement. I conclude that the manner interpretation is not available for the TLC, which confirms that the TLC does not denote motion along a path. 4.2.6 Manner of motion verbs As concluded in the previous chapter, a manner of motion verb in the weg-construction denotes the means of motion in the traversal of the path. However, a manner of motion verb in the TLC is not interpreted as denoting the means of motion along a path. Consider the following examples from the INL corpus. (33) a. De gezusters Beltman fietsten zich dit weekeinde ook in de prijzen. the sisters B. biked REFL this weekend also in the prizes ‘The sisters Beltman also biked their way to a prize this weekend.’ b. Brandweercommandant Hendrik van 't Zand reed zich naar de vierde stek. fire.officer H. van ‘t Z. rode REFL to the fourth place ‘Fire officer Hendrik van ‘t Zand rode his way to fourth place.’ In these examples, the subject referents are not moving along a path to the location expressed by the PP. Instead, the manner of motion verbs fietsen ‘bike’ and rijden ‘ride’ denote the indirect means of reaching the stative location expressed by the PP. I propose that apparent manner of motion verbs in the TLC are interpreted as manner of action verbs. 109 4.2.7 Conclusions This section has shown that the TLC and the weg-construction differ in meaning in the following ways. First, the two subevents described by the weg-construction are necessarily co-extensive, whereas in the TLC they need not be. That is, in the TLC the action described by the verb may have finished before the location represented by the PP is reached, but this is not the case for the weg-construction. The two subevents described by the TLC may also take place at different locations, whereas they necessarily take place at the same location the weg-construction. Second, the weg-construction strongly implies an iteration of a bounded event, but this is not the case for the TLC. In the TLC, the location may also be reached by one single execution of the action denoted by the verb, that is, one jump, one kick, and so on. Third, the TLC is necessarily telic, whereas the weg-construction can also have an atelic reading. Fourth, the PP of the TLC can be predicated directly of the subject, which is only possible when the PP denotes a stative location. Consequently, the PP of the weg-construction cannot be predicated of the subject. Fifth, the weg-construction can have a manner interpretation for at least some speakers, whereas this interpretation is not available for the TLC. Finally, a manner of motion verb is interpreted as the means of motion along a path in the weg-construction, but it is interpreted as a manner of action verb in the TLC. These observations indicate that the TLC does not denote motion along a path. I propose that this construction denotes a transition to a location, which does not include the traversal of a path. The verb in the TLC denotes the indirect means by which this location is reached. The PP therefore denotes a stative location and is headed by a Place preposition. 4.3 An event structure account This section discusses the event structures of the TLC and the weg-construction. Since the weg-construction denotes motion along a path, it describes a simple event. The event structure of the TLC, however, will be argued to be complex. The reflexive in the TLC is licensed because the construction describes a complex event. My argumentation is largely based on L&RH (1999) and Rappaport Hovav & Levin’s, henceforth RH&L, (2001) 110 event structure account of the resultative construction in English, which will be summarised first. 4.3.1 The event structure of the English resultative RH&L (2001) note that some English resultatives contain a fake reflexive (e.g. Kim sang herself hoarse) and others do not (e.g. Kim danced into the room). They propose that fake reflexive resultatives consist of two distinct subevents that are not temporally dependent or co-extensive. For example, Kim sang herself hoarse entails a singing event and a becoming hoarse event, but these are not necessarily co-extensive. That is, the hoarseness may be achieved some time after the singing is over (2001:775). The reflexive in a complex event resultative is licensed by the Argument-per-Subevent Condition, which states that “there must be at least one argument XP in the syntax per subevent in the event structure” (RH&L 1998:112-13). Examples like Kim sang herself hoarse consist of two distinct subevents that are not co-extensive, which therefore each require an argument XP in the syntax. Since both subevents have the same participant, the participant of the becoming hoarse subevent expressed by the resultative AP is realised as a reflexive pronoun co-referential with the subject (2001:780). Hence, the ‘fake’ reflexive is not really fake, but a real argument of the distinct subevent described by the resultative phrase.4 By contrast, resultatives without a reflexive, which RH&L call “bare XP resultatives”, such as Kim danced into the room, have a simple event structure. Although two subevents are distinguishable (i.e. a dancing event and a getting into the room event), the temporal relation between them differs from the temporal relation in the fake reflexive resultative: in bare XP resultatives the progress of the subevent denoted by the verb and the progress towards the achievement of the result state are temporally 4 The term ‘fake’ reflexive was coined by Simpson (1983), who assumed that the reflexive is but a syntactic placeholder. Simpson observed that the result XP in the resultative construction may only be predicated of the immediate postverbal NP in direct object position, not of the subject. The ‘fake’ reflexive then serves as a syntactic device that allows verbs that do not have objects, such as unergative verbs, to occur in the resultative construction as well. The resultative XP is predicated of the fake reflexive, which is coreferential with the subject (as in Kim yelled herself hoarse). However, as discussed in this section, L&RH (1999) and RH&L (2001) argue that the reflexive is not just a syntactic placeholder, but an argument of the resultative XP. Since nothing hinges on this labelling, I will still call the reflexive in the TLC a ‘fake’ reflexive, to follow the tradition in the literature. 111 dependent, while in the reflexive pattern they need not be (2001:775). In other words, in Kim danced into the room, Kim progresses towards the room by dancing, so her dancing is co-extensive with her progress of traversing the path into the room. When she stops dancing, she will stop progressing towards the room (2001:781). L&RH propose that bare XP resultatives are represented as simple events in event structure terms because the subevents are co-identified, that is, the two subevents meet the conditions on “event identity”. Two conditions that are required for event co-identification are that the subevents must have the same location and they must be temporally dependent (cf. L&RH 1999:213 and the references therein). L&RH suggest that the reason why a bare XP resultative is preferred to a reflexive resultative when the subevents satisfy the conditions for event identity is that the preferred expression of a situation is the one that gives it the tightest event structure, a preference that follows from Grice’s (1975) Maxim of Quantity (1999:215). They propose that instances like ??Kim danced herself into the room are not ungrammatical, but disfavoured on pragmatic grounds. Since the dancing event and the event of traversing a path into the room are temporally dependent and take place in the same location, they are co-identified. Therefore, the tightest event structure is preferred (1999:216). In sum, RH&L propose that the reflexive that occurs in some English resultative constructions is a manifestation of a complex event structure that consists of two distinct, non-co-extensive subevents. The Argument-per-Subevent Condition (RH&L 1998) dictates that every distinct subevent in the event structure requires a separate argument XP in the syntax. 4.3.2 Simple and complex events Section 4.2.1 argued that the location described by the PP in the TLC might be reached some time after the event denoted by the verb is over. In addition, the two subevents described by the TLC may take place at different locations. For example, in the TLC in (21) above Carl Lewis sprong zich in het Guinness Book of Records ‘Carl Lewis jumped his way into the Guinness Book of Records’, Carl Lewis most likely finished jumping before he got to be in the Guinness Book of Records, and these two events probably also took place at different locations. The subevent described by the PP and the one 112 represented by the verb are thus not co-identified, which means that they constitute two separate events. Consequently, the TLC does not denote motion along an incremental path. The subevent described by the verb and the one represented by the PP are distinct subevents in the TLC, which each require a separate argument XP in the syntax. The argument of the verb is the subject of the sentence, and the argument of the subevent described by the PP is the reflexive. Since the reflexive is co-referential with the subject, both subevents in the TLC have the same participant. By contrast, the two subevents described by the weg-construction are temporally dependent, and they necessarily take place at the same location. This is because the subject referent moves along the path described by the weg NP, by means of the action denoted by the verb. When he stops V-ing, he will stop progressing along the path. Therefore, the subevents described by the weg-construction are co-identified and the construction describes a simple event. As stated in the Argument-per-Subevent Condition, a simple event structure only requires one argument XP in the syntax. The fact that the weg-construction also contains a reflexive, even though it describes a simple event, does not constitute a problem for this analysis. Recall that the reflexive in the weg-construction occurs in the indirect object position, which was argued to be a remnant of the (once productive) double object construction. The reflexive is thus not present to satisfy the Argument-per-Subevent Condition, but because it is a frozen indirect object. To summarise, the TLC and the weg-construction have different event structures: the TLC denotes a transition to a stative location, which is a complex event that consists of two distinct subevents that each require a separate argument XP in the syntax. The reflexive in the TLC is licensed to satisfy the need of the PP for a syntactic argument. Contrastively, the weg-construction denotes motion along a path and describes a simple event, consisting of two subevents that are co-identified. The reflexive that is present in the weg-construction therefore does not provide a syntactic argument of any distinct subevent. Because of this difference in meaning, some weg-construction examples cannot be expressed as a TLC, and some TLC examples cannot be expressed as a wegconstruction. This is illustrated in (34). 113 (34) a. Johan schopt zich ??(een weg) door J. kicks REFL de menigte. a way through the crowd ‘Johan kicks his way through the crowd.’ b. Johan schopt zich (??een weg) door J. kicks REFL de eerste ronde. a way through the first round ‘Johan kicks his way through the first round.’ The weg-construction in (34a) implies that Johan is moving through the crowd by means of kicking; he creates a path by kicking the people that are in his way and travels the path by continuously kicking. This sentence is odd without een weg, i.e. as a TLC, because this does not allow an incremental path reading. In the TLC in (34b), on the other hand, Johan is not moving through the first round along a path by means of kicking. This sentence rather entails a transition from ‘not being through the first round’ to ‘being through the first round’, which is the result of kicking, say, soccer balls. This location may be achieved some time after Johan has finished kicking and thus would be strange as a weg-construction. Sometimes the same sentence can be expressed as a weg-construction and as a TLC. These sentences then have a different meaning. Compare the weg-construction in (35a) with the TLC in (35b). (35) a. De crimineel heeft zich een weg uit de gevangenis gezwommen. the criminal has REFL a way out the prison swum ‘The criminal swam his way out of prison.’ b. De crimineel heeft zich uit de gevangenis gezwommen. the criminal has REFL out the prison swum ‘The criminal swam his way out of prison.’ In the weg-construction (35a), the criminal is literally swimming out of the prison, where he encounters external difficulty and which requires effort. For example, he may have to swim through an underground canal or through the sewage system. His progress along the path is simultaneous with his swimming: with every stroke, he moves further along 114 the path out of the prison, and he will stop swimming when he is free. Therefore, the verb denotes the means of motion along a path. By contrast, in the TLC in (35b), out of prison refers to a stative location that is reached, which is the result of swimming. For instance, the criminal could win a swimming competition, where as a prize he is released. Since he most likely will not yet be out of the prison when he stops swimming, the criminal is not literally swimming out of the prison. Consequently, the apparent manner of motion verb is interpreted as a manner of action verb. 4.3.3 Relation to ‘fake’ reflexive resultative construction The TLC is similar in both meaning and form to the fake reflexive resultative construction in Dutch. Both constructions contain a ‘fake’ reflexive and an XP denoting the result of the action described by the verb. Moreover, both the TLC and the resultative describe a complex event, consisting of two distinct subevents that are not co-identified. For example, in the fake reflexive resultative Max schreeuwde zich hees ‘Max yelled himself hoarse’, the becoming hoarse event may take place after the yelling event is finished, and these two events may also take place at a different location. That is, Max may have been yelling in the football stadium on Saturday and wake up hoarse in his bed on Sunday. The TLC can also be formally identical to the fake reflexive resultative construction, because the resultative XP can be realised as a PP in Dutch, as in the following examples. (36) a. Miranda zong zich in trance / aan flarden / te pletter. M. sang REFL in trance on rags to pieces ‘Miranda sang herself into a trance/to pieces.’ b. Miranda zoop zich in coma. M. boozed REFL in coma ‘Miranda boozed herself into a coma.’ The resultative XP describes a state which is the result of the action denoted by the verb: Miranda gets to be ‘in a trance’ or ‘to pieces’ as a result of singing in (36a), and ‘in a 115 coma’ as a result of heavy drinking in (36b). These resultatives are formally identical to the TLC, as both constructions consist of a verb, a weak reflexive and a PP. One could argue that the two constructions are in fact the same construction, as they are very similar in meaning. That is, the location that is reached in the TLC can be similar to a state, as it is often not a physical but a rather metaphorical location (such as a final, the Guinness Book of Records, and so on). However, I propose that the meaning of the two constructions differs in one respect: the location expressed by the PP in the TLC also exists independently of the subject, whereas the state described by the resultative XP does not. Compare the resultatives in (36) above with the TLC in (37). (37) Miranda zong zich in de top 40. M. sang REFL in the top 40 ‘Miranda sang her way into the top 40.’ This sentence denotes that Miranda gets to be in the top 40 as a result of her singing, which is similar in meaning to the resultatives in (36). However, the top 40 also exists without Miranda being in it, whereas a trance or a coma does not exist without her. Therefore, I propose that the PP in the TLC does not denote a property or a state, but a possibly metaphorical - location. That is, the TLC describes a transition to a location, whereas the fake reflexive resultative expresses a transition to a state. Additional evidence for the semantic difference between the TLC and fake reflexive resultative comes from English. In English, two different constructions are used for the equivalents of the fake reflexive resultative and the TLC, as shown in respectively (38a) and (38b). (38) a. The athletes ran themselves / *their way into a coma. b. The soccer player kicked ??himself / his way into the Guinness Book of Records. 116 The grammaticality judgements of these examples are based on the results of the English questionnaire (see appendix 4). Only two of the 31 participants approved of (38a) as a way-construction, whereas 77% of the speakers approved of the same sentence as a fake reflexive resultative. By contrast, only 19% of the speakers approved of (38b) as a fake reflexive resultative, as opposed to 97% who approved of the same sentence as a wayconstruction. These findings suggest that when the result described by the PP does not exist without the subject, speakers of English prefer the fake reflexive resultative, as in (38a). By contrast, when the result does exist independently of the subject, speakers prefer the way-construction, as in (38b). The result that English uses two different constructions for the equivalents of the Dutch TLC and fake reflexive resultative confirms that they are not the same construction in Dutch. I conclude that the TLC is very similar, though not identical, to the fake reflexive resultative. The resultative construction is often analysed as describing a complex causative event, consisting of a causing event expressed by the verb, and a result expressed by the XP (cf. L&RH 1999 and the references therein). This is based amongst others on the fact that resultatives can be given a causative paraphrase. For example, Sam yelled himself hoarse can be paraphrased as Sam’s yelling caused him to become hoarse (1999:201). I propose that the TLC can also be analysed as describing a complex causative event, because the PP expresses the result brought about by the action of the verb, which in the TLC is a location. The paraphrase of for example the TLC Marien zwom zich in de finale ‘Marien swam his way into the final’ is Marien caused himself to be in the final by means of swimming. 4.3.4 Conceptual structure of the TLC This section describes the conceptual structure of the TLC (cf. Jackendoff 1990). Jackendoff represents a causative event with the Event function CAUSE. I will use this function to express that the verb in the TLC describes a causative event. Jackendoff assumes the CAUSE function to have two arguments: the first argument can be a Thing or an Event, and the second argument is an Event. If the first argument of CAUSE is a Thing, it is the Agent, and if it is an Event, it is the Cause. The second argument of CAUSE is the Effect (1990:44). Jackendoff uses the Event function INCH (inchoative) to 117 describe to describe a transition to a state. Even though the TLC does not denote a transition to a state but to a location, I will still use INCH to express the meaning of the TLC, as this construction also denotes a transition. The CS of Marien zwemt zich in de finale ‘Marien swims his way into the final’ can then be represented in two ways, given in (39). In the CS in (39a), the subject is the Causer (i.e. Agent), and in (39b) the swimming event is the Cause. (39) a. [[Event CAUSE ([Thing MARIEN], [Event INCH [State BE ([Thing MARIEN], [Place IN ([Thing FINALE])])]])] [BY [Event SWIM ([Thing MARIEN])]]] b. [Event CAUSE ([Event SWIM ([Thing MARIEN])], [Event INCH [State BE ([Thing MARIEN, [Place IN ([Thing FINALE])]])]])] These CSs show that the Agent of the swimming and the entity that gets to be in the final are the same, as the latter is realised as a reflexive in the sentence. Since the reflexive represents the thing undergoing a change of location, I will consider it to be a Theme (cf. Gruber 1965; Jackendoff 1990; Dowty 1991). The CS in (39a) means that Marien causes himself to be in the final by means of swimming. In (39b), it is the swimming event that causes him to be in the final. Following Jackendoff (1990:237), who discusses two similar interpretations of a resultative construction, I propose that the former more accurately represents the meaning of the TLC. This is because only in (39a) does the surface subject have a role in the main conceptual clause - that is, only in (39a) is the subject an Agent. I argued in section 4.1.1 that the subject of the TLC is an Agent, which is only appropriately expressed in the CS in (39a). The hypothesis that the subject of the TLC is an Agent is also illustrated by the following examples. In (40a), the TLC is combined with the agentive adverb opzettelijk ‘purposefully’. The example in (40b) shows that an adverb that implies non-volitionality, such as per ongeluk ‘accidentally’, is a bit odd in the TLC. 118 (40) a. Marien heeft zich M. opzettelijk in de finale gezwommen. has REFL purposefully in the final swum ‘Marien purposefully swam his way into the final.’ b. ?Marien heeft zich per ongeluk in de finale gezwommen. M. has REFL per accident in the final swum ‘Marien accidentally swam his way into the final.’ The example in (40b) implies that Marien is swimming on purpose but gets to be in the final by accident. This is odd because the meaning of the TLC entails that the subject referent controls the result, as also argued by Hoekstra (1988:120) for the Dutch fake reflexive resultative. Since the subject referent gets to be at the location described by the PP purposefully, this suggests that the CS in (39a) most accurately represents the meaning of the TLC. I conclude that the TLC describes a causative complex event, where an agentive subject causes himself to reach a stative location, by means of the action denoted by the verb. 4.3.5 Summary The TLC describes a complex causative event and is very similar to the fake reflexive resultative, except in one respect. The resultative XP denotes a state, which cannot exist without the subject, whereas the PP of the TLC denotes a location, which does exist independently of the subject. This semantic difference is confirmed by the observation that the English equivalents of these two Dutch constructions are formally distinct. The reflexive in the TLC is licensed because the construction describes a complex event, consisting of two distinct subevents that are not co-identified, and which therefore each require a separate argument XP in the syntax. The reflexive is the argument required by the subevent described by the PP. Since the reflexive is co-referential with the subject, both subevents in the TLC have the same participant. By contrast, the weg-construction describes a simple event, as the action denoted by the verb is co-identified with the traversal of the weg path. 119 4.4 The structure of the TLC Resultative XPs are generally considered to be a sister of the verb (cf. e.g. Hoekstra 1988; Carrier & Randall 1992; L&RH 1995, Radford 2004 and others). Hoekstra argues that the postverbal NP and the resultative predicate form a constituent which is an argument of the verb. He proposes that this constituent is a Small Clause (SC), which “denotes a state of affairs which is represented as a consequence of the activity or process denoted by the verb” (1988:121). He argues that any activity verb may be combined with a complement that denotes the state resulting from that activity (1988:131). That is, each activity verb has an optional result argument. L&RH (1995), as well as L&RH (1999) and RH&L (2001), do not decide whether the postverbal NP is a syntactic argument of the verb or of the resultative XP. L&RH (1995) propose a syntactic analysis of the fake reflexive resultative construction, where the resultative XP is licensed by a lexical rule that maps an activity into an accomplishment. L&RH (1999) and RH&L (2001), on the other hand, provide a semantic account, according to which the reflexive is only licensed when the sentence describes a complex event. They argue that the reflexive provides an argument in the syntax for the distinct subevent described by the resultative phrase, but they do not discuss whether the reflexive is a sister of the verb or a subject of the resultative XP. I will assume that the reflexive NP in the TLC is a sister of the verb, not the subject of the PP, because an adverb can be inserted between the reflexive and the PP, as shown in (41). (41) a. Melanie Kuiper zwom zich M. K. vrij verrassend in de nationale ploeg. (INL) swam REFL quite surprising in the national team ‘Melanie Kuiper quite surprisingly swam her way into the national team.’ b. Hij lacht zich er manmoedig doorheen. he smiles REFL there bravely through.PRT ‘He bravely smiles his way through.’ www.blauwefeniks.nl/forum/messages/212/5323.html?1082455432 The fact that an adverb can intervene between the reflexive and the PP in the TLC indicates that they do not form a constituent. 120 Another reason to assume that the reflexive is not the subject of the PP, but a separate constituent, is the UTAH. In Adger’s (2003) Minimalist version of the UTAH, the daughter of vP is interpreted as an Agent and the daughter of VP as a Theme. I proposed in section 4.3.4 above that since the reflexive in the TLC is the entity that undergoes a change of location, it is a Theme. This suggests that the reflexive NP in the TLC is the daughter of VP, and not the subject of the PP. 4.4.1 A Minimalist approach This section offers a Minimalist analysis of the TLC, adopting Adger’s (2003) version of the Minimalist Program discussed in section 3.7.1. However, Adger, and the MP in general, do not incorporate constructions. To account for the fact that the TLC has an NP and a PP complement that are not subcategorised for by the verb, I will again assume that the [uN] and [uP] features on the verb are contributed by the construction. These features do not belong to the lexical verb itself, because the verb that enters the construction is unergative. The [uN] and [uP] features are checked by the arguments of the construction, which are the reflexive NP and the PP, respectively. The start of the derivation of Marien zwemt zich in de finale ‘Marien swims his way into the final’ then looks as follows. (42) TP qp T′ qp vP T [ucase:nom] qp NP v′ [uN] Marien qp [ucase: ] VP v qp NP zich [ucase: ] [case:acc] V′ [uN] qp PP V in de finale [uP, uN] zwemt This structure shows that the [uP] and [uN] features occur on V itself, they are not brought into the derivation by the verb. The reflexive NP checks the [uN] feature on V 121 and the PP checks the [uP] feature. The accusative case feature of the reflexive is checked and valued by v. The [uN] feature on v is checked by Merging with the external argument NP Marien. The NP Marien will move up to TP to check the strong [uN] feature on T (not shown). The nominative case feature of this NP is checked and valued by T. To check its tense features (not shown), the main verb zwemmen moves up via v to T, and ends up in C (also not shown). The structure in (42) obeys the UTAH: Marien is the daughter of vP and is interpreted as an Agent. The reflexive is the daughter of VP and is interpreted as a Theme. The reflexive is co-referential with the subject, so Marien is both the Agent of the swimming and the Theme that undergoes a change of location. Even though the Agent and the Theme are the same person, two separate NPs are needed because the swimming event is a distinct subevent from the transition to location subevent, which thus each require an argument in the syntax. RH&L’s semantic account of the licensing of the reflexive corresponds to the Unique Theta Generalization (Adger 2003 and the references therein), which states that one constituent cannot be assigned more than one theta-role. However, the reflexive seems to be licensed by the semantics rather than the syntax: there are two theta-roles because the two subevents are not co-identified. When the two subevents are co-identified, as in direct motion sentences like Marien zwemt de zee in ‘Marien swims into the sea’, a reflexive is not licensed.5 I conclude that the reflexive in the TLC is present for semantic reasons. The start of the derivation of the TLC example with the adverb in (41a) above can be represented as in (43). Following Adger (2003), I will assume that adverbs are adjuncts to vP. 5 Note that such directed motion instances may be problematic for the Unique Theta Generalization, because the subject NP appears to have both the Agent and the Theme Theta-role. In Marien zwemt de zee in ‘Marien swims into the sea’, Marien is swimming and Marien is moving into the sea. However, manner of motion verbs with a directed path phrase are assumed to be unaccusative in Dutch (e.g. Zaenen 1993), which means that Marien is in fact only the Theme. In other words, directed motion sentences focus on the change of location, rather than on the manner of motion. 122 (43) TP qp vP T qp [ucase:nom] vP Adv [uN] 3 vrij verrassend NP v′ Melanie 3 [ucase: ] VP v 3 [ucase:acc] NP V′ [uN] zich 3 [ucase: ] PP V in de nationale ploeg [uP, uN] zwemt The derivation proceeds as follows. The verb will move from V to v to T, to eventually adjoin to C. The external argument moves to TP to check the strong [uN] feature on T, and subsequently moves to CP. The end of the derivation is given in (44). Adger represents traces of moved constituents inside angled brackets, and features that are checked and consequently deleted are notated with a strikethrough. For the purpose of this analysis, I will assume that the reflexive is a clitic, that is, it is phonologically bound to the verb.6 6 The reflexive has many of the properties of a clitic: it is phonologically unstressed, it belongs to a closed class (i.e. pronouns), it has a grammatical rather than a lexical meaning, and most importantly, it always seems to occur next to the tensed verb (cf. e.g. Zwicky 1977, Anderson 1985). I will leave it to further research to determine whether the reflexive in the TLC is indeed a clitic to the verb. 123 (44) CP qp NP C′ Melanie qp [ucase:nom] C TP 3 3 T C NP T′ 3 <Melanie> 3 v T vP T 3 3 [uN, case:nom] V v Adv vP 3 vrij verrassend 3 V N NP v′ zwemt zich <Melanie> 3 [ucase:acc] VP v 3 [uN, case:acc] NP V′ <zwemt> <zich> 3 PP V in de nationale ploeg [uP, uN] <zwemt> The reflexive NP gets valued accusative case by v. The NP Melanie gets valued nominative case by T. The reflexive cliticizes to its phonological host, the lexical verb V, which yields the correct word order. In the previous chapter I argued that the meaning of the weg-construction cannot be compositionally derived from its parts, as none of these parts entails motion. The TLC, however, has a greater degree of compositionality, because the PP serves as a secondary resultative predicate on the reflexive (cf. Hoekstra 1988; Marantz 1992; L&RH 1995). Moreover, I argued that the reflexive is present because the TLC describes a complex event. The hypothesis that the reflexive is present for semantic reasons indicates some degree of compositionality. However, the TLC is an instance of the more general resultative template, which still has to be learnt by speakers of Dutch. In sum, the TLC is an instance of the more general resultative construction. The meaning of this construction is not fully compositional, because it contains two syntactic complements that are not semantic arguments of the verb. Moreover, the meaning of the main verb is demoted to a subordinate means modifier, so the syntactic head of the sentence is not the semantic head. In the Minimalist analysis suggested here, the TLC is assumed to be a structure that includes a [uN] and a [uP] feature on the verb, and two 124 internal arguments, a reflexive NP and a PP. The lexical entry of the TLC can be represented as follows. (45) a. TP qp T′ qp vP T qp [case:nom] NP v′ [uN] qp VP v qp NP REFL [case:acc] V′ [uN] qp PP V [uP, uN] b. [[Event CAUSE ([Thing ], [Event INCH [State BE ([Thing ], [Place ([Thing ])])]])] [BY [Event ([Thing ])]]] The lexical entry of the TLC has the structure in (45a), which corresponds to the meaning in (45b). Thus, the TLC is a transitive structure that includes a reflexive NP and a PP, which has the meaning ‘cause to be PP by means of V-ing’. 4.5 Discussion This section discusses some potential counterexamples to the proposed meaning of the TLC, which was argued to not include the traversal of a path. Furthermore, I will discuss some patterns that look like a TLC, but where the weak reflexive is in fact a semantic argument of the verb. 4.5.1 TLCs that imply motion The following examples seem to contradict the hypothesis that the TLC does not denote motion along a path, because they do imply motion.7 7 Since these examples imply motion, (46b) and (46c) can also be expressed as a weg-construction, which will have a slightly different meaning. The sentence in (46a) is odd as a weg-construction because the PP is headed by a Place preposition. See next chapter for further discussion. 125 (46) a. Het blaasorkest […] worstelde zich in een paar personenwagens, inclusief the brass.orchestra wrestled REFL in a couple passenger.cars including hun instrumenten. their instruments ‘The brass orchestra wrestled its way into a couple of passenger cars, including their instruments.’ jiel.nl/verhalen/2004_vietnam/ned_vietnam_2004_02.htm b. Zodra de roest zich onder je coating weet te werken …8 as.soon.as the rust REFL under your coating knows to work ‘As soon as the rust has worked its way under your coating …’ www.kaagkevers.nl/forum/ read.php?f=2&i=785&t=785 c. The kever knaagt zich door de bast. the beetle gnaws REFL through the bark ‘The beetle is gnawing its way through the bark.’ www.ako.be/koeheide/insecten.htm These examples imply motion, but the verbs are not motion verbs. When the reflexive is omitted, the sentences either become ungrammatical or they get a different meaning, as shown in (47). (47) a. ??Het blaasorkest worstelde in een paar personenwagens. the brass.orchestra wrestled in a couple passenger.cars ‘The brass orchestra wrestled in(side) a couple of passenger cars.’ b. ??De roest werkt onder je coating. the rust works under your coating ‘The rust works under your coating.’ c. The kever knaagt door de bast. the beetle gnaws through the bark ‘The beetle gnaws through the bark.’ 8 Since in this example the rust is not actually moving, it is an instance of subjective motion (see section 3.2.3). 126 In the absence of a reflexive, the PP can only be interpreted as a frame locative that describes the location where the action denoted by the verb takes place. This is why (47a) and (47b) are odd: the former means that the orchestra is wrestling inside the cars, and the PP in the latter refers to the place where the rust is ‘working’, but rust is an inanimate entity that usually does not work. The example in (47c), on the other hand, is fine but it has a different meaning than (46c), because it does not imply motion. Both examples in (46c) and in (47c) were included in the second questionnaire and the participants were asked to describe the difference in meaning (see appendix 2). The majority of the speakers (69%) said that (46c) means that the beetle completely gets through the bark with its body, whereas it is just gnawing away at the bark in (47c) without moving through it. These data suggest that the examples in (46) instantiate the TLC, because in the presence of the reflexive the sentences denote a location that is reached, but in the absence of a reflexive the PP can only be interpreted as a frame locative. I propose that the sense of motion in these examples is inferred from the fact that the location is reached. The location is a physical location that can be interpreted as the end of a path, so speakers infer that in order to reach that location, the subject referent must move. Hence, in (46c) the beetle gets to be through the bark by means of gnawing, just like for example Melanie gets to be in the national team by means of swimming in (41a) above. By contrast, when the reflexive is absent the PP does not denote an endpoint that is reached, but refers to the location where the action of the verb takes place, which can therefore also be omitted. Since the verb itself does not imply motion, the sense of motion is only due to the meaning of the TLC, which entails that the location expressed by the PP is reached. Further evidence for the hypothesis that the sense of motion in the TLC is inferred comes from the fact that the PP can be predicated of the subject. Recall that when the PP denotes a path it cannot be predicated of the subject, whereas when it denotes a stative location it can (see section 4.2.4). The PP of the examples in (46) above can be directly predicated of the subject, as shown in (48). 127 (48) a. Het blaasorkest is in een paar personenwagens. the brass.orchestra is in a couple passenger.cars ‘The brass orchestra is inside a couple of passenger cars.’ b. De roest is onder de coating. the rust is under the coating ‘The rust is under the coating.’ c. The kever is door de bast. the beetle is through the bark ‘The beetle is through the bark.’ Further evidence for the claim that the sentences in (46) instantiate the TLC comes from the fact that they are necessarily telic, as shown in (49a) for the TLC in (46c). By contrast, the sentences in (47) can also be atelic, as shown in (49b) for the sentence in (47c). (49) a. The kever heeft zich *urenlang / in een uur door de bast geknaagd. the beetle has REFL hours.long in an hour through the bark gnawed ‘The beetle gnawed its way through the bark *for hours/in an hour.’ b. The kever heeft urenlang / in een uur door de bast geknaagd. the beetle has hours.long in an hour through the bark gnawed ‘The beetle gnawed through the bark for hours/in an hour.’ I conclude that the TLC may imply motion when the PP is interpreted as a physical location that can constitute the end of a path. However, this sense of motion is derived from the meaning of the construction, which entails a location that is reached as the result of the action denoted by the verb. The sense of motion is thus inferred from the fact that the endpoint is reached. The examples in (46) and (47) therefore have a different structure, as illustrated in (50) for respectively the TLC in (46c) and the sentence in (47c). 128 (50) a. TP 3 b. T′ 3 vP T 3 [ucase:nom] NP v′ [uN] de kever 3 [ucase: ] VP v 3 NP V′ TP 3 T′ 3 vP 3 PP T [ucase:nom] vP [uN] 3 door de bast NP [case:acc] [uN] de kever [ucase: ] zich 3 [ucase: ] PP V door de bast [uN, uP] knaagt v′ 3 VP g V v [case:acc] [uN] knaagt In the TLC in (50a), the PP is a complement of the verb, whereas the PP in (50b) is an adjunct to vP. In both examples, the verb knagen is unergative; in the TLC it gets the [uN] and [uP] features from the construction, but it remains featureless in (50b). In (50a), the interpretable case feature on v is checked by the reflexive NP, whereas this feature remains unchecked in the unergative sentence in (50b). I conclude that the TLC can imply motion when the PP is interpreted as a physical location that can be the end of a path. However, this sense of motion is inferred from the meaning of the TLC, which entails that the location is reached. 4.5.2 Reflexive for pragmatic reasons The following examples also appear to contradict the proposed meaning of the TLC, because they denote motion along a path. Furthermore, the manner of motion verbs in these examples denote the actual manner of motion, instead of the means by which a location is achieved.9 (51) a. Hij slalomt zich behendig langs zijn tegenstanders, een lust voor het oog. he slaloms REFL nimbly along his opponents a lust for the eye ‘He nimbly slaloms his way past his opponents, a pleasure to watch.’ www.nrc.nl/W2/Lab/Profiel/WKvoetbal/zuidafrika.html 9 Since the verb denotes the actual way of moving along a path, these sentences can also be expressed as a weg-construction, which would imply an extended path and more external difficulty than the bare motion counterparts. See next chapter for further discussion. 129 b. Vanaf Tapah kronkelt zich een weg van 60 km lengte met meer dan from T. winds REFL a road of 60 km length with more than zeshonderd bochten naar boven. six.hundred curves to up ‘From Tapah a road of 60 km length winds its way up with more than six hundred curves.’ www.visitmalaysia.nl/cameron.htm The PP in these examples refers to a physical location, which can constitute the end of a path. However, these examples are different from those in (46) above, because they contain a manner of motion verb that denotes the actual manner of motion. Consequently, these examples denote motion along a path, where the action described by the verb and the progress along the path represented by the PP are temporally dependent and take place at the same location. In the sentence in (51a), the subject referent is moving by slaloming, and (51b) denotes subjective motion; although nothing is actually moving, since roads do not move, we perceive the sentence as denoting motion because of our incrementally built-up perception of the long and path-like object extending through space. Since the verbs are manner of motion verbs, the reflexive can also be omitted without affecting the sense of the verb, as shown in (52). (52) a. Hij slalomt behendig langs zijn tegenstanders. he slaloms nimbly past his opponents ‘He nimbly slaloms past his opponents.’ b. Vanaf Tapah kronkelt een weg naar boven. from T. winds a road to up ‘From Tapah a road winds upwards.’ For examples like those in (51) it is not immediately clear how the reflexive is licensed, because the sentences describe simple events. Therefore, they do not require a separate 130 argument XP in the syntax (which is in fact confirmed by the possibility of omitting the reflexive). L&RH (1999) propose that the bare XP pattern is favoured in the case of event co-identification because it has the tightest event structure, a preference that follows from Grice’s (1975) Maxim of Quantity. However, they also suggest that pragmatic considerations can override this Maxim of Quantity, for instance when the speaker wants to “draw attention to the two conceptually distinct events that are being composed” (1999:217). They illustrate this with the following example.10 (53) … he withdrew from the kitchen and sauntered his Bermuda-shorted self through the front door. (1999:217, 38b) L&RH state that the subject referent in this example is both an intentional Agent in subject position and a postverbal manipulatable physical object (which is also signalled by the use of the modifier Bermuda-shorted, which is used to describe the physical appearance of the object, not the subject, 1999:217). By choosing to use the reflexive variant rather than the bare XP pattern, they argue, the writer wants to draw attention to both subevents in the sentence, which otherwise would be packaged into one event in the syntax. The presence of the reflexive in the Dutch examples in (51) can be explained with the same arguments. These examples imply both an intentional Agent and an object that is manoeuvred by the Agent. That is, the subject referent in (51a) is both the Agent of the slaloming, as well as the manipulatable object that is manoeuvred past the opponents. Likewise, the road in (51b) is portrayed as an animate entity that is both actively winding, and getting itself up the mountain. Since the Agent and the object is the same entity, the latter is represented by a reflexive. By using two argument XPs for what could be packaged into one event in the syntax, the writer is drawing attention to the two conceptually distinct subevents in the sentence. 10 Strictly speaking, this example is not an instance of the reflexive pattern because it has a modifier inserted into the reflexive pronoun, but L&RH consider this example to be a variant of the reflexive resultative pattern (1999:217). 131 Since the reflexives in (51) are present to focus on the two conceptually distinct subevents in the sentence, I consider them to instantiate the TLC, even though they imply motion along a path and their reflexives can also be omitted. I take the reflexive in the TLC to be a manifestation of a complex event structure, which also seems to be the intention of the writer: the reflexives in (51) are present to focus on the two different subevents in the sentence. Thus, what could otherwise be represented as a bare motion sentence with a simple event structure is now represented as a complex event - in other words, as a TLC. 4.5.3 Verbs that can take a weak reflexive as a thematic object The following sentences look like a TLC, as they contain a weak reflexive and a PP that denotes the result of the action described by the verb. (54) a. Piet gooit zich voor de trein. P. throws REFL in.front.of the train ‘Piet throws himself in front of the train.’ b. De jongens slepen zich the boys door de dode uren. (Verhagen 2004: 341, 34) drag REFL through the dead hours ‘The boys are dragging themselves through the dead hours.’ c. Oma hijst zich uit haar stoel. grandma heaves REFL out.of her chair ‘Grandma heaves herself out of her chair.’ d. De boeg van het vrachtschip boorde zich in de zijkant van de veerboot. (INL) the bow of the cargo.ship bored REFL in the side of the ferry ‘The bow of the cargo ship bored its way into the side of the ferry.’ e. Wij wringen / wurmen / persen ons we squeeze door de nauwe opening. REFL through the narrow opening ‘We squeeze ourselves through the narrow opening.’ The verbs in (54) very frequently pattern with a weak reflexive and a PP in Dutch. Verhagen (2004:341) includes the verbs slepen ‘drag’, boren ‘bore, drill’, wurmen 132 ‘squeeze’, persen ‘squeeze’ and proppen ‘stuff’ in the set of verbs that he found in the ‘zich-verplaatsings-constructie’, which he exemplifies with the sentence in (54b). However, there are reasons to believe that these sentences do not instantiate the TLC, because the weak reflexive is in fact a semantic argument of the verb. As discussed above, one of the pieces of evidence for the hypothesis that the reflexive in the TLC is not a semantic argument of the verb is that it is obligatorily weak; reflexives that are a semantic argument of the verb are normally strong. However, the weak reflexives in (54) can be replaced by a strong reflexive, as well as by another full NP, as shown in (55). (55) a. Piet gooit zichzelf / zijn tas voor de trein. P. throws REFL.self his bag in.front.of the train ‘Piet throws himself/his bag in front of the train.’ b. De jongens slepen zichzelf / hun verveelde makkers door de dode uren. the boys drag REFL.self their bored mates through the dead hours ‘The boys are dragging themselves/their bored mates through the dead hours.’ c. Oma hijst zichzelf / opa uit haar stoel. grandma heaves REFL.self grandpa out her chair ‘Grandma heaves herself/grandpa out of her chair.’ d. Het vrachtschip boorde zichzelf / zijn boeg in de zijkant van de veerboot. the cargo.ship bored REFL.self his bow in the side of the ferry ‘The cargo ship bored itself/its bow into the side of the ferry.’ e. Wij wringen / wurmen / persen onszelf / onze tassen door we squeeze de nauwe opening. REFL.self our bags through the narrow opening ‘We squeeze ourselves/our bags through the narrow opening.’ The fact that the weak reflexives in (54) can be replaced with a strong reflexive or a full NP, suggests that they are in fact semantic arguments of the verb. However, the sentences in (54) with a weak reflexive and those in (54) with a strong reflexive have a slightly different meaning. Veraart (1996) discusses the difference in meaning between similar examples with a weak and a strong reflexive, which she refers to as ‘presupposed’ and ‘asserted’ reflexivity, respectively. She assumes that 133 presupposed reflexivity occurs when the reflexive cannot be replaced by a non-reflexive element. Asserted reflexivity occurs when the reflexive can be replaced with a nonreflexive element (1996:19). Presupposed reflexivity prefers a weak reflexive and asserted reflexivity prefers a strong reflexive. The sentences in (54) with the weak reflexives then instantiate presupposed reflexivity: their readings do not allow a nonreflexive NP, so the subject referents are not literally throwing, dragging, etc. themselves. The example in (54a) means that Piet is jumping in front of the train, (54b) implies that the boys are moving through the dead hours, and so on. Therefore, these sentences denote a change in the body position of the subject referent. The sentences with a strong reflexive in (55), on the other hand, involve asserted reflexivity: these readings also allow a non-reflexive element, so the subject referent is physically moving his own body, or alternatively, something else. Veraart points out that “it takes some imagination to create a context in which zichzelf is possible when the predicate has this meaning” (1996:19). That is, it takes some imagination to interpret how Piet in (55a) is literally throwing himself, in the same way that he is throwing for example his bag. In sum, the sentences with a weak reflexive denote a change in the body position of the subject referent, whereas the same sentence with a strong reflexive implies a physical displacement of one’s own body. The ambiguity of verbs that can denote both a change in one’s body position and physically moving one’s own body has been a longstanding issue in the philosophy of language (Kate Kearns p.c., cf. Chisholm 1964; Davidson 1967). Verbs like throw, heave, raise, squeeze and so on systematically display this ambiguity, so they are generally considered to be polysemous (Kate Kearns, p.c.). I propose that these verbs are polysemous in Dutch as well: gooien ‘throw’ slepen ‘drag’, hijsen ‘heave’, and so on can both refer to a change in the body position of the subject referent (in which case the reflexivity is presupposed), and to physically moving one’s own body (in which case reflexivity is asserted). What is important to note here is that in both (54) and (55), the reflexives are semantic arguments of the verb, even though they are weak in (54). This is evidenced by the fact that a strong reflexive can be used as well, with only a slight difference in meaning. Moreover, the weak reflexives in (54) transfer the action of the verb back to their antecedent: Piet is throwing himself, grandma is heaving herself, and so on. I argued in section 4.1.2 above that the reflexive in the TLC 134 cannot transfer the action of the verb back to its antecedent, because it is not a semantic argument of the verb. The proposed difference in meaning between the fake reflexive in the TLC and a weak reflexive that is a semantic argument of the verb is confirmed by the results of the second questionnaire, which contained the following two sentences. The participants were asked for the difference in meaning. (56) a. Piet gooit zich voor de trein. (= 54a) P. throws REFL in.front.of the train ‘Piet throws himself in front of the train.’ b. Piet gooit zich naar goud. P. throws REFL to gold ‘Piet throws his way to gold.’ These sentences look formally identical, as they consist of a verb, a weak reflexive and a PP. The majority of the 42 speakers (63%) said that (56a) means that Piet is throwing himself (which they expressed with a strong reflexive), but that he is throwing something else in (56b), like a discus or darts. Hence, the reflexive in (56a) transfers the action of the verb back to the antecedent, whereas the reflexive in the TLC does not.11 Furthermore, the weak reflexive in (56a) can be replaced with a strong reflexive, whereas this is unacceptable for the TLC in (56b), as shown in (57). (57) a. Piet gooit zichzelf voor de trein. (= 55a) P. throws REFL.self in.front.of the train ‘Piet throws himself in front of the train.’ b. ??Piet gooit zichzelf naar goud. P. throws REFL.self to 11 gold Although it was not mentioned by any of the participants, (56b) could mean that Piet is throwing himself, because the thematic object of the verb is implied in TLC. For example, Piet could win a gold medal by throwing himself into a sandbox the fastest or the furthest of all competitors. However, this thematic object is not represented by the reflexive. 135 The fact that the weak reflexives in (54) can be replaced with a strong reflexive and that they transfer the action of the verb back to their antecedent suggest that they are semantic arguments of the verb. Although these sentences superficially look like a TLC, they do not instantiate this construction. The difference between the sentences in (56) is further supported by English, which uses the formally different constructions given in respectively (58a) and (58b). (58) a. Piet throws himself / *his way in front of the train. b. Piet throws his way / ??himself to gold. The grammaticality judgements in these sentences are based on the results of the English questionnaire (see appendix 4). None of the 31 participants approved of (58a) as a wayconstruction, whereas everybody accepted the same sentence with a reflexive. The questionnaire included a sentence very similar to (58b), with the verb kick instead of throw (see 37b above). All but one of the participants approved of this sentence as a wayconstruction, but only 19% of the speakers accepted the same sentence with a reflexive. The result that English uses two different constructions for the two Dutch sentences in (56) above confirms the difference between the two Dutch sentences. I conclude that some patterns may look like a TLC in Dutch, as they consist of a verb, a weak reflexive and a PP. However, the weak reflexive may be a semantic argument of the verb, in which case they do not instantiate the TLC. 4.5.4 Verhagen’s reflexive verbs In the analysis proposed here, reflexive verbs are incompatible with the TLC because they cannot be used intransitively. However, Verhagen (2004) includes the following reflexive motion verbs among the verbs that he found in his zich-verplaatsingsconstructie: bewegen ‘move’, verspreiden ‘spread’, begeven ‘go’, haasten ‘hurry’, spoeden ‘speed’ and voortplanten ‘propagate’. When combined with a PP, they may look like a TLC, as shown in (59), where (59a) is provided by Verhagen (2004) as an instance of the zich-verplaatsings-constructie. 136 (59) a. Hij bewoog zich door de geluidssluis naar ... (2004:341, 35) he moved REFL through the sound.gate to ‘He moved through the sound gate to …’ [translation mine] b. Dan verspreidt zich een behaaglijke gloed door de kleine, gezellige ruimte. then spreads REFL a comfortable glow through the small cosy space ‘Then a comfortable glow spreads through the small cosy space.’ (INL) c. Moeder haast zich in de kamer. mother hurries REFL in the room ‘Mother is hurrying inside the room.’ These sentences look like a TLC because they consist of a verb, a weak reflexive and a PP. However, there are reasons to believe that these examples do not instantiate the TLC. First of all, the PP can also be omitted, as shown in (60). (60) a. Hij beweegt zich. he moves REFL ‘He is moving.’ b. De gloed verspreidt zich. the glow spreads REFL ‘The glow is spreading.’ c. Moeder haast zich. mother hurries REFL ‘Mother is hurrying.’ The fact that the PPs in (59) can be omitted indicates that they are frame locatives, which contrasts with the obligatoriness of the PP in the TLC. Secondly, the sentences describe a simple event, which means that the reflexive cannot be present to provide an argument for a distinct subevent in the sentence. Instead, the reflexive is present because the verb is inherently reflexive. Consequently, as can be seen in (59), these sentences cannot be translated with a way-construction in English, which is another reason why they do not instantiate the TLC. 137 I conclude that the reflexive verbs that Verhagen includes amongst the verbs that he found in the zich-verplaatsings-constructie cannot occur in the TLC. Reflexive verbs are incompatible with the TLC because they cannot be used intransitively. 4.6 Summary The TLC in Dutch consists of a verb, a weak reflexive and a PP. In the Minimalist approach adopted here, the TLC can be skeletally represented as follows. (61) [vP SUBJ [[VP REFL [PP V]] v]] The TLC is similar to the weg-construction discussed in the previous chapter in many ways. Both constructions have two syntactic complements that are not semantic arguments of the verb, one of which is a weak reflexive. Both constructions furthermore have a PP that is headed by the same range of spatial prepositions. The constraints on the verb that can occur in the constructions are also the same: the weg-construction and the TLC only allow agentive intransitive activity verbs. They do not allow unaccusative verbs, stative verbs, strictly (di)transitive verbs and verbs that are inherently reflexive. Finally, both the TLC and the weg-construction are translated with a way-construction in English. However, the two Dutch constructions have very different meanings: the wegconstruction denotes motion along a path, whereas the TLC denotes a transition to a location, which does not involve the traversal of a path. That is, the location expressed by the PP may be achieved some time after the event denoted by the verb is finished, and these two events do not necessarily take place at the same location. The PP of the TLC refers to a stative location that is reached, which is the result of the action denoted by the verb. A manner of motion verb is therefore not interpreted as denoting the means of motion (as it is in the weg-construction), but as a manner of action verb that describes the indirect means by which the location is attained. The PP is headed by a Place preposition. The TLC describes a complex causative event, consisting of two distinct, non-coidentified subevents, which each require an argument XP in the syntax (cf. Argumentper-Subevent condition, L&RH 1998). The reflexive in the TLC fulfils the requirement 138 for an argument XP of the subevent described by the PP. Since a reflexive is coreferential with the subject, both subevents have the same participant. By contrast, the weg-construction denotes motion along a path, so the two subevents are necessarily coextensive and they also take place at the same location. The weg-construction thus describes a simple event, which does not require a separate argument XP in the syntax. The reflexive in the weg-construction is not a syntactic argument of a distinct subevent, but an indirect object contributed by the construction. The TLC has a greater degree of compositionality than the weg-construction, because the PP serves as a resultative predicate on the reflexive NP. The reflexive (which is the subject) ends up at the location described by the PP, where there is no sense of motion that needs to be accounted for. The TLC is an instance of the more general template of the fake object resultative construction, and it can even be formally identical to the fake reflexive resultative. The semantic difference between the TLC and the fake reflexive resultative is that the resultative XP expresses a state, which cannot exist without the subject, whereas the PP in the TLC refers to a location, which does exist independently of the subject. Moreover, the TLC is translated with a way-construction in English but the fake reflexive resultative is not. In the Minimalist analysis proposed in this thesis, the [uN] and [uP] features on the verb are assumed to be contributed by the construction. These features are not part of the verb itself, because the verb that enters the construction is unergative. The [uN] and [uP] features are checked by the internal arguments of the construction. 139 Chapter 5 Discussion and conclusions The previous two chapters described two different way-constructions that exist in Dutch, called the weg-construction and the Transition to Location Construction (TLC). This chapter gives an overview of the conclusions reached so far and subsequently discusses some potential counterexamples to the proposed difference in meaning between the two Dutch constructions. Furthermore, the meaning of the English way-construction will be discussed. I will argue that the English construction is in fact ambiguous between a motion along a path reading and a transition to a location reading. This ambiguity has not been recognised in the literature as far as I am aware, and it may reconcile some of the existing controversial analyses of this construction. Section 5.1 summarises the similarities and differences between the wegconstruction and the TLC as determined so far. Section 5.2 discusses some apparent overlaps in meaning between the weg-construction, the TLC and simple manner of motion sentences. In section 5.3 I will show that the way-construction in English has an additional meaning, equivalent to the meaning of the TLC. Section 5.4 briefly discusses motion constructions in other Germanic languages that have been described in the literature and suggests that these may in fact be ambiguous as well. Section 5.5 summarises the conclusions drawn in this thesis and section 5.6 finishes with some interesting issues for future research. 5.1 Overview of the weg-construction and the TLC As has been demonstrated in the previous chapters, the weg-construction and the TLC are similar in many ways. Both constructions take a wide range of activity verbs, which have to be unergative. That is, only verbs that can be used intransitively and that have an agentive subject are allowed into the constructions. This constraint rules out unaccusative verbs, as the subject of these verbs is a Theme, as well as reflexive verbs and strictly (di)transitive verbs, as these verbs cannot be used intransitively. The verb in both 140 constructions also cannot be stative, as stative verbs cannot be conceived as the means by which a path is travelled or a location is reached. Furthermore, both constructions contain a reflexive that is obligatorily weak and a PP that can only be realised as a prepositional phrase. The PP is headed by the same range of spatial prepositions in both constructions, and can refer to both literal and metaphorical space. Both constructions violate the argument structure of the verb, as they contain two syntactic complements that are not semantic arguments of the verb. Consequently, the constructions represent a mismatch in the syntax-semantics mapping, first of all because the thematic arguments of the verb cannot be mapped onto the syntactic complements of the verb in the construction. Secondly, the meaning of the main verb is demoted to a subordinate means or manner modifier, so the syntactic head of the sentence is not the semantic head. The weg-construction and the TLC are both translated with a way-construction in English. However, the two constructions were argued to be syntactically as well as semantically distinct. Table 5.1 lists the semantic differences between the two Dutch constructions. Table 5.1 Semantic differences between the weg-construction and the TLC weg-construction Basic interpretation TLC Basic interpretation Incremental traversal of a path by means Transition to a stative location by means of (or while) V-ing Temporal relation between subevents Action described by the verb is coextensive with traversal of the path Repeatable bounded event The verb denotes the means of traversing an incremental path, so iteration of bounded event of V-ing Temporal relation between subevents Action described by the verb is typically temporally disjoint from the transition to the location Repeatable bounded event No path traversal, so bounded event may be interpreted as a single action 141 Table 5.1 - continued weg-construction Telicity Traversal of path event, so may be telic or atelic Typical verb Impact verb, which denotes means of creating a path / removing obstacles Compositionality Not compositional: meaning cannot be predicted from individual parts or syntax Manner of motion verb TLC Telicity Transition event, so must be telic Typical verb Any agentive intransitive action verb, which denotes indirect means of reaching location Compositionality Higher degree of compositionality, because PP is resultative predicate on the reflexive NP Apparent manner of motion verb Interpreted as means of motion (reflexive Interpreted as manner of action (reflexive and weg NP can be omitted without cannot be omitted without affecting the affecting the meaning of the verb) meaning of the verb) The differences in meaning between the two constructions are reflected in their structures. I have proposed the following Minimalist analyses, where (1) gives the lexical entry of the weg-construction and (2) gives the lexical entry of the TLC. The conceptual structures in (1b) and (2b) represent the meanings of the structures in (1a) and (2a) (the dashed line indicates optional arguments). 142 (1) a. TP qp T′ qp T vP qp NP [case:nom] [uN] v′ qp VP v qp [case:acc] [uN] NP REFL V′ qp VP V qp NP [case:acc] [uN] V een weg PP [ucase: ] [uN] b. [Event GO ([Thing ], [Path ([Place AT-END-OF ([Path ([Thing -----------------------------[BY/WITH [Event ([Thing ])]]] (2) a. ])])])]) ---- TP qp T′ qp vP T qp [ucase:nom] NP v′ [uN] qp VP v qp NP REFL [case:acc] V′ [uN] qp PP V [uP, uN] b. [[Event CAUSE ([Thing ], [Event INCH [State BE ([Thing ], [Place ([Thing ])])]])] [BY [Event ([Thing ])]]] The verb that enters the constructions has to be unergative - that is, it should not have any [uN] or [uP] features of its own, as these are assumed to be contributed by the construction, and its subject should be an Agent. The weg-construction has two [uN] features, one on V and one on the additional VP, which is assumed to have case checking 143 abilities. These [uN] features are checked by the weg NP and the reflexive NP. The TLC has a [uN] and a [uP] feature on V, which are checked by the reflexive NP and the PP, respectively. As can be seen in (1b) and (2b), the reflexive elements of both constructions do not appear in the CS of the constructions. This is for different reasons. The reflexive in the weg-construction used to be a beneficiary indirect object, but it has lost its meaning in the modern weg-construction. The reflexive of the TLC, on the other hand, is a syntactic argument of the subevent described by the PP. The reflexive ensures that the participant of the subevents represented by the verb and the one described by the PP is the same person. Both structures obey the UTAH, which states that identical thematic relationships are represented by identical structural relationships. That is, an Agent is always the daughter of vP and a Theme is always the daughter of VP. The subject of both constructions is the daughter of vP, so it is interpreted as an Agent. The reflexive in the TLC was argued to be a Theme, and consequently is the daughter of VP. However, the weg NP in the weg-construction is also the daughter of VP, but it is not a Theme. This is not necessarily a contradiction, because the UTAH only works in one direction: a Theme is the daughter of VP. However, the daughter of VP does not have to be a Theme. 5.2 Apparent overlap in the meaning of the weg-construction and the TLC So far I argued that the weg-construction and the TLC have a different meaning: the former denotes motion along a path and the latter does not. However, as discussed in section 4.5, some instances of the TLC can imply motion along a path. This may be the case when the PP is a location that can be interpreted as the end of a path, or when the reflexive in the TLC is present for pragmatic reasons. Since these instances imply motion, they can also be expressed as a weg-construction. This section compares such TLC examples to the semantically similar weg-construction examples. 144 5.2.1 TLCs with a PP that can be interpreted as the end of a path Section 4.5.1 discussed some instances of the TLC that imply motion along a path, such as the example in (3a) from the previous chapter, as well as (3b) that implies metaphorical motion. (3) a. The kever knaagt zich door de bast. (= 45c) the beetle gnaws REFL through the bark ‘The beetle is gnawing its way through the bark.’ b. Haar favoriet was Maar vanavond heb ik hoofdpijn, waar ze her favourite was but een kraai doorheen a zich zo vals als tonight have I headache where she REFL as false as krijste. crow through.PRT shrieked ‘Her favourite was But tonight I have a headache, that she shrieked her way through terribly out of tune.’ (Panorama magazine, volume 39, 2005) Since these examples imply motion along a path, they can also be expressed as a wegconstruction, as shown in (4). (4) a. The kever knaagt zich een weg door de bast. the beetle gnaws REFL through the bark ‘The beetle is gnawing its way through the bark.’ b. … waar ze zich zo vals als een kraai een weg doorheen where she REFL as false as a crow krijste. a way through.PRT shrieked ‘… that she shrieked her way through terribly out of tune.’ The TLC examples in (3) and the weg-construction examples in (4) imply (metaphorical) motion. Since knagen ‘gnaw’ and krijsen ‘shriek’ are not motion verbs, the reflexive and een weg cannot be omitted without changing the meaning of the sentence: without these elements the PP can only be interpreted as a frame locative, which thus does not denote 145 motion. These examples seem to contradict the hypothesis that the TLC and the wegconstruction have a different meaning, as they both seem to denote motion. However, I argued in the previous chapter that the sense of motion in the TLC is inferred from the fact that the endpoint is reached. When the PP in the TLC expresses a location that is interpreted as a Goal, speakers infer that in order to reach that Goal, the subject must move. This hypothesis is confirmed by the fact that the TLC is telic, whereas the weg-construction can have both a telic and an atelic reading, as shown in (5) for the examples in (3a) and (4a). (5) a. The kever knaagde zich *urenlang / in twee uur door de bast. the beetle gnawed REFL hours.long in two hours through the bark ‘The beetle gnawed its way through the bark *for hours/in two hours.’ b. The kever knaagde zich urenlang / in twee uur een weg door de bast. the beetle gnawed REFL hours.long in two hours a way through the bark ‘The beetle gnawed its way through the bark for hours/in two hours.’ The TLC in (5a) can only be combined with the non-durative phrase in twee uur ‘in two hours’, indicating that it is telic. By contrast, the weg-construction in (5b) can be combined with both a durative phrase and a non-durative phrase, suggesting that it has both a telic and an atelic reading. These results support the proposed meanings of the Dutch constructions: the TLC entails a transition to a location and is telic, whereas the weg-construction describes motion along a path, which can be both telic and atelic. In sum, a TLC may imply motion when the PP describes a location is interpreted as a Goal. However, this sense of motion is inferred from the fact that the Goal is reached. 5.2.2 Instances of the TLC with manner of motion verbs Section 4.5.2 discussed some instances of the TLC with a manner of motion verb that denote motion along a path, as in (6a). Consequently, the sentence can also be expressed as a weg-construction, as shown in (6b). Since the motion verb denotes the means of 146 motion, the reflexive and een weg can also be omitted without affecting the meaning of the verb, as shown in (6c). (6) a. Camara dribbelt zich C. door de defensie … dribbles REFL through the defence ‘Camara dribbles his way through the defence.’ www.sporting.be/content/verslag/0405/bdeinze.asp b. Camara dribbelt zich een weg door C. de defensie … dribbles REFL a way through the defence ‘Camara dribbles his way through the defence.’ c. Camara dribbelt C. door de defensie … dribbles through the defence ‘Camara dribbles through the defence.’ Following L&RH (1999), I proposed that the reflexive in (6a) is present for pragmatic reasons, to focus on the two conceptually distinct subevents of the sentence, which could otherwise be packaged into one simple event (as in 6c). By giving each subevent a separate argument XP in the syntax, the writer is drawing attention to both the dribbling event and the traversal of the path through the defence. Thus, the reflexive is a manifestation of a complex event structure in (6a), which is the reason why such examples were taken to instantiate the TLC. The TLC in (6a) has a slightly different meaning than the weg-construction in (6b) and the simple motion sentence in (6c): the latter two describe a simple event, where the trotting and the progress along the path are co-identified. These sentences focus on the traversal of the path. By contrast, the TLC describes a complex event, which focuses on the endpoint that is reached. The TLC therefore is telic, where the other two sentences can be either telic or atelic, as shown in (7). (7) a. Camara dribbelde zich *mintenlang / in twee minuten C. door de defensie. dribbled REFL minutes.long in two minutes through the defence ‘Camara dribbled his way through the defence *for minutes/in two minutes.’ 147 b. Camara dribbelde zich mintenlang / in twee minuten een weg door de defensie. C. dribbled REFL minutes.long in two minutes a way through the defence ‘Camara dribbled his way through the defence for minutes/in two minutes.’ c. Camara dribbelt mintenlang / in twee minuten C. door de defensie. dribbled minutes.long in two minutes through the defence ‘Camara dribbled through the defence for minutes/in two minutes.’ These examples show that the weg-construction in (7b) and the simple motion sentence in (7c) have a telic and an atelic reading. The difference in meaning between these two sentences is that the weg-construction implies more effort and a longer path than the simple motion sentence. The difference in meaning between these three sentences is furthermore confirmed by the fact that the PP can be predicated of the subject for the TLC in (6a), whereas this is not possible for the weg-construction and the simple motion sentence, a shown in respectively (8a) and (8b). (8) a. Camara is C. door de defensie. is through the defence ‘Camara is through the defence.’ b. *Camara is C. door de defensie. (with the path reading) is through the defence These observations support the proposed difference in meaning between the wegconstruction and the TLC: the PP of the TLC denotes a location, whereas the PP in the weg-construction describes a path. To summarise, a reflexive may be present for pragmatic reasons in a sentence with a manner of motion verb, to focus on the conceptually distinct subevents of the sentence. Even though such sentences denote motion along a path and the verb denotes the actual means of motion, they are considered to instantiate the TLC. This is because the reflexive is a manifestation of a complex event structure. The following difference in 148 the interpretation of manner of motion verbs in the weg-construction and the TLC can now be added to table 5.1. Table 5.1 - continued weg-construction TLC Apparent manner of motion verb Manner of motion verb Interpreted as means of motion (reflexive and weg NP can be omitted without affecting the meaning of the verb) Does not denote manner of motion, but is interpreted as manner of action (reflexive cannot be omitted without affecting the meaning of the verb). However, can be pragmatically interpreted as means of motion. 5.3 The English way-construction is ambiguous In the literature, the way-construction in English is considered to denote motion along a path, which can be paraphrased as ‘traverse the PP path by or while doing V’ (Jackendoff 1990; Marantz 1992; Goldberg 1995; L&RH 1995). This section proposes an additional meaning of the way-construction, which is the equivalent of the meaning of the Dutch TLC. 5.3.1 Temporal relation between subevents The two subevents described by the way-construction can be temporally dependent, as in the following examples. (9) a. Willy jumped his way into Harriet’s arms. (Jackendoff 1990:223) b. Kelly laughed her way out of the room. (L&RH 1995:198) c. For hours, troops have been shooting their way through angry, unarmed mobs. (Goldberg 1995:204) 149 In these examples, the subevent described by the verb and the one represented by the PP are co-extensive: in (9a), Willy is traversing the path that leads into Harriet’s arms by means of jumping, so the jumping and the progress along the path unfold at the same rate. With every jump, Willy progresses further on the path that leads into Harriet’s arms; when he stops jumping, he will stop progressing. These subevents also take place at the same location. Likewise, Kelly is traversing the path that leads out of the room by means of (or ‘while’) laughing in (9b), and the troops are progressing through the mobs by means of shooting in (9c). With every shot, the troops progress further through the crowd. Since the two subevents in these examples are necessarily co-extensive and they necessarily take place at the same location, they are co-identified. These wayconstruction examples therefore have a simple event structure and describe motion along a path. By contrast, the subevents in the following examples are not necessarily coextensive, nor do they necessarily take place at the same location. (10) a. Babe Ruth homered his way into the hearts of America. (Jackendoff 1990:212) b. Corporate executives wined, dined and golfed their way to a record 4.98 trillion yen … (L&RH 1995:137) c. Joe bought his way into the exclusive country club. (Goldberg 1995:205) The location expressed by the PPs in these examples may be reached some time after the action denoted by the verb is over. That is, in (10a) Babe Ruth may be hitting homeruns for years without entering the hearts of the American people. He could even end up in their hearts after his death. Moreover, these two events may take place at different locations: the homering presumably takes place on the baseball field, whereas the entering the hearts of Americans can take place in their living rooms, behind the television screen. Similarly, the corporate executives in (10b) can obtain their trillion yen contract after they have finished wining, dining and golfing, and the signing of the contract most likely also takes place at a different location. Joe in (10c) may also have finished bribing before he gets to be a member of the exclusive country club, and the bribing of people can take place at a different location than in the country club. 150 In sum, the two subevents described by the way-constructions in (10) are not coidentified. Therefore, these examples have a complex event structure, and they do not denote motion along a path. This hypothesis is in fact confirmed by Jackendoff (1990:219), who notes that “in certain cases of ‘metaphorical’ movement such as [10a], neither go or get is appropriate; instead, the motion verb enter turns up as fairly acceptable”. He illustrates this observation with the following paraphrases of the wayconstruction in (10a) above (1990:219, 24). (11) a. ?*Babe Ruth went into the hearts of America homering. b. ?*Babe Ruth got into the hearts of America (by) homering. c. Babe Ruth entered (into) the hearts of America (by) homering. Jackendoff’s observation that some way-construction instances cannot be paraphrased with go or get confirms that these instances do not denote motion. 5.3.2 Repeated action In the way-construction examples in (9) above, the subject referents are moving by means of the action denoted by the verb. In the example in (9a), Willy is getting closer to Harriet with every jump he makes, so he is gradually moving towards her in incremental steps. Jackendoff observes that this sentence strongly implies several jumps (1990:224). In fact, Goldberg argues that the repeated action constraint accounts for the unacceptability of the following examples (1995:212). (12) a. *With a single bullet, Jones shot his way through the crowd. b. *She jumped her way over the ditch. Goldberg suggests that these examples are unacceptable because they do not denote a repeated action. However, the following attested way-construction examples do not necessarily imply a repetition of the action denoted by the verb. 151 (13) a. Junior Andra Manson leapt his way to first place in the high jump. www.cstv.com/sports/c-xc/uwire/022706aab.html b. German schoolgirl Annika Irmler has licked her way into the Guinness Book of Records with her whopping seven centimetre Tongue. http://news.sydney.gothic.org.au/viewthread.php?tid=1777. c. Bill lied his way into the army. Speakers interpret the way-construction in (13a) to possibly refer to a single jump, because Andra could perform one fantastic jump and consequently become first. Likewise, the example in (13b) could possibly refer to one lick, as the German schoolgirl gets to be in the Guinness Book of Records because of the length of her tongue, not because of repeated licking. Jackendoff mentions the example in (13c) in a footnote and points out that some people have suggested to him that this way-construction could refer to one lie, provided this lie is Bill’s means for entry (1990:298, fn. 2). In sum, in contrast to what has been claimed in the literature, some instances of the way-construction do not necessarily a repetition of a bounded event. 5.3.3 Telicity None of the authors in the literature discusses the telicity of the way-construction. Marantz (1992) and L&RH (1995) propose that the construction is parallel to the fake object resultative, thereby implying that the way-construction only has a telic reading. However, Goldberg gives the example in (9c) above of a way-construction combined with the durative adverbial for hours, which indicates that this example is atelic. To investigate the telicity of the way-construction, the following two wayconstructions denoting motion along a path were included in the English questionnaire. One was combined with the durative time adverbial for days and the other with the nondurative time adverbial in two days. (14) a. Tarzan slashed his way through the jungle for days. b. Tarzan slashed his way through the jungle in two days. 152 Of the 31 participants, 84% approved of the atelic sentence in (14a) and 87% approved of the telic sentence in (14b) (see appendix 4). The majority of the speakers (71%) approved of both sentences. These results suggest that the way-construction can have both a telic and an atelic reading. However, speakers do not approve of the durative time adverbial for two hours in the following way-construction with the same preposition; they only accept this example with the non-durative adverbial in two hours. (15) a. *Venus Williams hit her way through the first round for hours. b. Venus Williams hit her way through the first round in two hours. The two subevents in (15b) are not co-identified, because Venus Williams may have finished hitting before she gets to be through the first round, for instance because she has to wait for her competitors to play their games. The unacceptability of the durative time adverbial in (15a) suggests that this way-construction only has a telic reading. 5.3.4 Direct predication of the PP over the subject A PP that refers to a stative location can be predicated directly of the subject, as illustrated in (16). By contrast, a PP referring to a path cannot be predicated directly of the subject, as shown in (17). (16) a. The girl is running in the forest. b. The girl is in the forest. (17) a. The girl ran into the forest. b. *The girl is into the forest. (with the path reading) As to be expected, the PP of a way-construction that denotes motion along a path cannot be predicated directly of the subject, as shown in (18a) for the example in (14) above. By 153 contrast, the PP in the way-construction in (15) can be predicated of the subject, as shown in (18b).1 (18) a. *Tarzan is through the jungle. (with the path reading) b. Venus Williams is through the first round. The observation that the PP in the way-construction in (15) can be predicated of the subject suggests that it does not denote a path, but rather a stative location. 5.3.5 Means and manner interpretation For some speakers, the verb in the way-construction can also denote a manner accompanying the motion of the subject referent. For instance, I knitted my way across the Atlantic can mean that the subject referent moved across the Atlantic while knitting, rather than by means of knitting (Goldberg 1995:213). However, a manner interpretation is not available for the following wayconstruction examples. (19) a. The final heat […] is the last chance for bar tenders to shake their way into the final … brightonfoodfestival.co.uk/index.php?id=32 b. … a few hundred notable women broke barriers and wrote their way onto the front pages of metropolitan newspapers. unp.unl.edu/bookinfo/4375.html . The example in (19a) cannot mean that the bar tenders are moving into the final by doing something else, and that the shaking only accompanies this movement. Likewise, (19b) cannot mean that the women are moving onto the front page by doing something else and 1 Note that some Path prepositions such as into and onto cannot be predicated directly of the subject, suggesting that they cannot be used as Place prepositions (cf. e.g. *Babe Ruth is into the hearts of America, *The German schoolgirl is into the Guinness Book of Records). On the other hand, typical Place prepositions are not allowed in the way-construction (cf. *Babe Ruth homered his way in the hearts of America, *The schoolgirl licked her way in the Guinness Book of Records). I do not have an explanation for this, so I will leave it to further research. 154 that the writing is only an accompanying action. The lack of a possible manner interpretation in these examples confirms that these examples do not denote motion along a path. 5.3.6 The two meanings of the English way-construction The above sections showed that some English way-constructions describe two subevents that are not necessarily co-identified. These way-constructions do not imply a repetition of the action denoted by the verb, they only have a telic reading, their PP can be directly predicated of the subject and they do not have a manner interpretation. These observations suggest that some way-construction instances do not denote motion along a path, but rather a stative location that is reached. I propose that the English way-construction in fact has an additional meaning: besides motion along a path, it can also describe a transition to a location. The former is the meaning that is described in the literature, which is the equivalent of the Dutch wegconstruction. The second meaning of a transition to a location reading has not been recognised in the literature as far as I am aware, apart from some footnotes mentioning problematic cases. This meaning is the equivalent of the Dutch TLC. For ease of exposition throughout the remainder of this chapter, let us call the English equivalent of the weg-construction the way-path-construction, and the equivalent of the TLC the waytransition-construction. Since the way-path-construction and the way-transition-construction are homomorphic, some way-construction instances can be ambiguous. Consider the following example. (20) The criminal swam his way out of the prison. This sentence can mean that the criminal is literally swimming out of the prison, where the swimming subevent and the progress along the path are co-extensive: with every stroke, he progresses further along the path that leads out of the prison. The motion takes place with effort, and the manner of motion verb is interpreted as the means of motion along a path. Alternatively, the PP out of the prison can refer to a stative location, which 155 is reached as a result of swimming. That is, the criminal could win a swimming competition, where as a prize he is released. These events are not co-extensive, because he will probably be released after he has finished swimming, and these events can also take place at different locations. The verb is therefore interpreted as a manner of action verb, which denotes the indirect means by which the location is reached. This reading thus describes a complex causative event. The conceptual structures of the way-path-construction and the way-transitionconstruction are given in (21a) and (21b), respectively. (21) a. [Event GO ([Thing CRIMINAL], [Path FROM ([Place IN ([Thing PRISON])])]) [BY [Event SWIM ([Thing CRIMINAL])]]] b. [[Event CAUSE ([Thing CRIMINAL], [Event INCH [State BE ([Thing CRIMINAL], [Place AT-END-OF ([Path FROM ([Place IN ([Thing PRISON])])])])]])] [BY [Event SWIM ([Thing CRIMINAL])]]] The superordinate event of the CS of the way-path-construction in (21a) is GO, whereas the superordinate event of the way-transition-construction in (21b) is CAUSE. This difference reflects that the former describes a simple event of motion along a path, whereas the latter describes a complex causative event of a transition to a location. The PP expresses the result brought about by the action of the verb, which is similar to the meaning of the fake object resultative. 5.3.7 Similarities between the way-construction and the ‘fake’ object resultative Chapter 1 summarised the debate in the literature about whether or not the wayconstruction is parallel to the fake object resultative construction. Marantz (1992) and L&RH (1995) argue that the two constructions are exactly parallel: the way NP can be regarded as a reflexive path NP and the PP serves as a resultative predicate on the way NP. Jackendoff (2002) and Goldberg (1996), on the other hand, assume that although the two constructions are related, they should be regarded as distinct constructions. The differences between the two constructions pointed out by Goldberg (1996) were 156 summarised in chapter 1. This section reconsiders these points and shows that not all of them hold for the way-transition-construction. 1) The way-construction implies the creation of a path by removing obstacles or involving other external difficulty, but the resultative does not. The way-transition-construction does not imply the creation of a path either, or any external difficulty, as it does not denote motion along a path. 2) The way-construction at least marginally allows a manner interpretation, whereas the resultative does not. For the way-transition-construction, the action denoted by the verb causes the result expressed by the PP. Consequently, this meaning does not have a manner interpretation either. 3) Dutch is a language that has fake object resultatives but does not have a wayconstruction (Annie Zaenen, p.c.). This thesis has shown that Dutch has two way-constructions: Verhagen (2002) described the weg-construction in a response to Goldberg’s remark, and this thesis has introduced a second type of way-construction. 4) The way-construction can be used with a wide variety of verbs, whereas the verb in the fake object resultative is highly restricted. I leave it to further research to investigate whether this point is also valid for the waytransition-construction. These observations suggest that, apart from the last point, Goldberg’s points are not valid for the way-transition-construction. Therefore, this construction is parallel to the fake object resultative construction. Both constructions denote a transition, which is a complex causative event and which is telic. The proposed meaning of the way-transitionconstruction is not ‘traverse the PP path by or while V-ing’, but rather ‘cause NP to become PP by V-ing’. This meaning is parallel to the meaning of the resultative construction as proposed by Jackendoff (2002:176). Marantz (1992) and L&RH (1995) also argue that the way-construction is parallel to the resultative construction, but they assume the way-construction to denote motion along a path. Their analysis therefore implies that the way-path-construction is 157 necessarily telic, but we have seen in section 5.3.3 above that this is not the case. I propose that only the way-transition-construction is parallel to the fake object resultative. That is, only the English equivalent of the Dutch TLC is parallel to the resultative construction. 5.3.8 Differences between the way-construction and the ‘fake’ object resultative The fake object resultative describes change of state, whereas the way-transitionconstruction denotes a change of location. Recall that the location in the TLC can be very similar to a state (cf. e.g. Hij rende zich in de prijzen ‘He ran his way to a prize’). I argued that the difference between the TLC and the fake reflexive resultative is that a location also exists independently of the subject, whereas a state does not. To investigate the difference in meaning between the way-transition-construction and the fake reflexive resultative with a PP resultative phrase, the following pairs of sentences were included in the English questionnaire. The grammaticality judgements are based on the results (see appendix 4). (22) a. The soccer player kicked his way/*himself into the Guinness Book of Records. b. The athletes ran *their way/themselves into a coma. c. The patient coughed *?his way/himself into a haemorrhage. These results suggest that speakers of English prefer a way-construction when the PP describes a location that exists independently of the subject, such as the Guinness Book of Records in (22a). By contrast, a fake reflexive resultative is preferred when the result described by the PP does not exist without the subject, such as a coma in (22b) or a haemorrhage in (22c). 5.3.9 A Minimalist approach In the literature, the English way-construction is assumed to have two syntactic complements that are not semantic arguments of the verb. To represent the structure of the way-construction in a Minimalist framework, several additional assumptions have to be made to account for the fact that the construction contains arguments that are not 158 subcategorised for by the verb (see discussion in section 3.7.1). As for the Dutch constructions, I will adopt and extend Adger’s (2003) version of the Minimalist Program. Again, I will assume that the [uN] and [uP] features on the verb in the way-construction are contributed by the construction. That is, they do not belong to the verb itself, because the verb that enters the construction is unergative: it has no features of its own. The wayconstruction in (22a) can be represented as follows. (23) TP 3 T′ 3 T vP [case:nom] 3 [uN] NP v′ the soccer player 3 [ucase: ] v VP [case:acc] 3 [uN] NP V′ his way 3 [ucase: ] V PP [uN, uP] kicked into the Guinness Book of Records The [uN] feature on the verb is checked by the way NP and the [uP] feature is checked by the PP. The accusative case feature on the way NP is checked and valued by v, and the nominative case feature on the subject NP is checked and valued by T. The lexical verb moves to v and the external argument moves to T (not shown). The lexical entry of the way-construction can be represented as in (24). (24) a. TP 3 T′ 3 T vP [case:nom] 3 [uN] NPi v′ 3 v VP [case:acc] 3 [uN] NP V′ POSSi way 3 [ucase: ] V PP [uN, uP] 159 b. [Event GO ([Thing ], [Path ([Place AT-END-OF ([Path ([Thing -----------------------------[BY/WITH [Event ([Thing ])]]] ])])])]) ---- c. [[Event CAUSE ([Thing ], [Event INCH [State BE ([Thing ], [Place ([Thing ])])]])] [BY [Event ([Thing ])]]] The structure in (24a) corresponds to the two meanings in (24b) and (24c). This structure has two fixed complements that are not semantic arguments of the verb: a possessive way NP and a PP. These complements are checked by the [uN] and [uP] features that are present on V. The conceptual structures in (24b) and (24c) represent the two meanings of the structure in (24a): the CS in (24b) represents the motion along a path reading (which has an optional telic reading), and the CS in (24c) represents the transition to a location reading. The way NP does not appear in either one of the CSs, which is in line with Jackendoff’s, Marantz’s and L&RH’s analyses, who all assume that the way NP is meaningless. However, according to Goldberg the way NP is a meaningful element, which therefore should appear in the CS of the way- construction. In chapter 3 it was suggested that external modification of the way NP is only possible for way-constructions with verbs that occur idiomatically in it, such as make or wend. Therefore, I will assume that in way-constructions with other verbs, the way NP is non-referential. This lexical entry suggests that, although the way-path-construction and the waytransition-construction have a different meaning, they both have the same structure. The proposed structure in (24a) is the same as the structure of the Dutch TLC (see 2a above), except that the daughter of VP is a reflexive NP in Dutch and a way NP in English.2 I leave it to further research to determine whether the way-path-construction requires a different structure than the way-transition-construction, or that the structure of the wayconstruction can be ambiguous as well. 5.3.10 Constructional idiom or decompositional analysis? The different analyses in the literature vary as to whether the meaning of the wayconstruction can be compositionally derived from its individual elements or not. Jackendoff (1990) and Goldberg (1995) propose a constructional analysis, because they 2 The structures also differ in word order, because English has SVO word order and Dutch has SOV. 160 claim that the sense of motion along a path cannot be predicted from its individual parts. Marantz (1992) and L&RH (1995) suggest that the meaning of the way-construction can be compositionally derived from its individual parts, because the PP functions as a resultative secondary predication on the way NP. Like the fake object resultative construction, the meaning of the way-construction can be compositionally derived from the interpretation of the eventuality as an accomplishment (1995:49-50). I propose that the constructional idiom analysis proposed by Jackendoff is most appropriate for the way-path-construction: this meaning entails motion along a path, which cannot be derived from its individual elements. However, the way-transitionconstruction has a higher degree of compositionality. We have seen that this wayconstruction entails a causative complex event, which is necessarily telic and thus similar to the fake object resultative construction. Therefore, the PP serves as a resultative predicate and the way NP and the construction belongs to the more general resultative template. As suggested by Marantz and L&RH, the fact that the way-construction contains a way NP instead of a fake reflexive can account for the change of location meaning, rather than a change of state. 5.3.11 Summary This section proposed an additional meaning of the English way-construction, which does not denote motion along a path, but a transition to a location. The PP expresses a stative location that is reached, which is the result of the action denoted by the verb. The waytransition-construction does not have the characteristics described in the literature: it does not imply a repetition of a bounded event, it cannot be paraphrased with go or get, and its PP does not denote a path. The way-transition-construction is similar to the fake object resultative construction in both meaning and form: both constructions contain an NP and an XP complement that are not semantic arguments of the verb, and both constructions describe a complex causative event. The only formal difference between the way-transitionconstruction and the fake object resultative with a resultative PP is that the former contains a POSS way phrase, whereas the latter contains a fake reflexive or an inalienably possessed body part. The only semantic difference is that the way-construction denotes a 161 transition to a location, which also exists independently of the subject, whereas the fake object resultative describes a transition to state, which does not exist independently of the subject. 5.4 Way-constructions in other Germanic languages Toivonen (2002) describes a construction in Swedish that is very similar to the wayconstruction, called the Directed Motion Construction (DMC). This construction consists of a verb, a weak reflexive and a PP, as illustrated in (25).3 (25) …han till sist kravlade sig uppför ravinens motsatta sida. (2002:315, 3b) he to last crawled REFL upwards ravine.the’s opposite side ‘… he finally crawled his way up to the opposite side of the ravine.’ Toivonen proposes that the DMC conveys the sense of directed motion: the subject moves by the means specified by the verb in the direction specified by the PP (2002:314). Seland (2001) describes a similar construction in Norwegian, called the Reflexive Caused Motion construction (RCM). This construction also consists of a verb, a weak reflexive and a PP, as illustrated in (26). (26) Eventyrerne kuttet seg gjennom og ut av den gjengrodde regnskogen. adventurers.the cut REFL through and out of the overgrown.the rain.forest.the ‘The adventurers cut their way through and out of the overgrown rain forest.’ (2001:41, 3c) Seland argues that the RCM denotes motion to a goal, by means of the action denoted by the verb. However, besides the motion along a path reading, these constructions seem to have an additional meaning. Consider the following Swedish DMC from Toivonen (2002:318) and the Norwegian RCM from Seland (2001:74). 3 Toivonen glosses the reflexive with the Swedish form in English, because “it is not clear that it is equivalent to English reflexives” (2002:314 fn. 3). However, for the sake of consistency I will gloss the reflexives in Toivonen’s examples with REFL, as this is how I glossed the reflexives throughout this thesis. 162 (27) a. Karin försöker springa sig in i Guinness Rekordbok. K. tries to.run REFL in in Guinness Record.book ‘Karin is trying to run her way into the Guinness Book of Records.’ b. Han kastet seg til he favorittstatus før konkurransen. threw REFL to status.as.a.favourite before competition.the ‘He threw his way to a status as a favourite before the competition.’ Toivonen assumes that the DMC in (27a) denotes figurative motion, and Seland takes the motion in the RCM in (27b) to be metaphorical. However, she notes that the repeated action constraint does not hold for this example, because it can also refer to one throw. Furthermore, preliminary research suggests that two different way-constructions exist in German as well, one with a reflexive NP and a Weg NP and one with only a reflexive (Ludwig 2005). Consider the following examples. (28) a. Mann pinkelte sich den Weg man aus Lawine. (Ludwig 2005:11, 33) peed REFL the way out.of avalanche ‘Man peed his way out of the avalanche.’ b. Er hat sich aus dem Gefängnis geschwommen. (Seibert 1993:62, 6b) he has REFL out of the prison swum ‘He swam his way out of the prison.’ (translation mine) The example in (28a) denotes that the man gets out of the avalanche by means of peeing. Seibert (1993) provides the example in (28b) outside the context of the way-construction. She notes that it can refer to an event where, say, the subject wins a swimming competition, where as a prize he is released. In other words, this example denotes a transition to a location, which does not involve the traversal of a path. These observations suggest that all Germanic languages may have two different way-constructions, one which denotes motion along a path and one which denotes a transition to a location. In Dutch and perhaps German, the path/transition distinction is visible in the syntax, as these languages use two formally different constructions. In English, Swedish and Norwegian, this distinction is not visible in the syntax, because the 163 way-path-construction is formally identical to the way-transition-construction. I think this is an interesting topic for further research. 5.5 Conclusions This thesis has given a detailed account of two way-constructions in Dutch, based on corpus data and questionnaires. These constructions are the weg-construction and what I have called the Transition to Location Construction (TLC). The constructions show that it is not always the verb that determines the complement configuration of the sentence, because both constructions have two syntactic complements that are not semantic arguments of the verb. Furthermore, the syntactic head of a sentence is not always the semantic head, because the main verb in the weg-construction and TLC is subordinate to respectively a GO and a CAUSE function. The weg-construction consists of a verb, a weak reflexive, een weg ‘a way’ and a PP. The reflexive NP and the weg NP, which includes the PP, are syntactic complements of the verb: the reflexive occurs in indirect object position and the weg NP in direct object position. The weg-construction denotes motion along a path that is created by the subject referent, by means of the action denoted by the verb. Since obstacles have to be removed or other barriers overcome in order to create and travel the path, the motion takes place despite external difficulty. The subevent described by the verb is co-identified with the subevent of traversing the path, so the construction expresses a simple event. The weg-construction was argued to be a constructional idiom, for the several reasons: first, the meaning of motion along a path cannot be derived from the individual elements of the construction, as none of them entails motion. Second, the constructions contain elements that are not selected by the verb, and the main verb is demoted to a subordinate means or manner modifier. Third, the weg-construction is an instance of the double object construction, but this pattern cannot normally be used in Dutch when the direct object is created. The verb in the TLC has a syntactic NP and PP complement, which are not semantic arguments of the verb. The NP is a weak reflexive. This construction denotes the achievement of a location, which is the result of the action denoted by the verb. The two subevents described by the TLC are typically temporally disjoint, and they may also 164 take place at different locations. Therefore, this construction describes a complex event. The PP is interpreted as a resultative predicate on the reflexive NP, so the TLC has a higher degree of compositionality than the weg-construction. The TLC is an instance of the more general resultative construction, and can be formally identical to the fake reflexive resultative. The two constructions are semantically very similar, though not identical: the fake reflexive resultative denotes a transition to state, whereas the TLC describes a transition to a location. The weg-construction and the TLC are very productive: as long as certain constraints are obeyed, any kind of verb is allowed into the construction. The constraints on the verb are the same for both constructions: the verb has to be able to be used unergatively. The verb cannot be stative, and the unergative constraint rules out unaccusative verbs. The difference in meaning between the two constructions has implications for the interpretation of the verb that occurs in them. A manner of motion verb in the weg-construction denotes the means of motion along a path, whereas a manner of motion verb in the TLC is interpreted as a manner of action verb denoting the indirect means of reaching a location. The weg-construction strongly implies an iteration of a bounded event, but the TLC does not have such implication. Furthermore, for some speakers, the verb in the weg-construction can be interpreted as an activity that accompanies the motion, but this interpretation is not available for the TLC. I presented the syntactic structures of the Dutch constructions within a Minimalist framework. Several additional assumptions had to be made to account for the fact that both constructions have two syntactic complements that are not subcategorised for by the verb. I assumed that the verb that enters the construction does not have any features of its own, and that the features on the verb are contributed by the constructions. The wegconstruction has two [uN] features, which are checked by the weg NP and the reflexive NP. The TLC has a [uN] and a [uP] feature, which are checked by the reflexive NP and the PP, respectively. The weg-construction and the TLC are both translated with a way-construction in English. Based on the difference in meaning between the two Dutch constructions, I have shown that the English way-construction is in fact ambiguous between a motion along a path reading and a transition to a location reading. Therefore, the controversial analyses 165 of the way-construction may be reconciled to some degree if we acknowledge that this construction has two meanings. In particular, the way-transition-construction, but not the way-path-construction, is comparable to the fake object resultative. Finally, evidence was presented that the path-type and the transition-type constructions identified in this thesis are likely to be found in Germanic languages generally. These two types are not formally realised in the same way across these languages: they may be two distinct constructions, as in Dutch and perhaps German, or they may be ambiguous, as in English, Swedish and Norwegian. 5.6 Issues for future research One major issue to be investigated is how (or if at all) constructions can be incorporated into the Minimalist Program. In the Minimalist approach presented in this paper, I assumed that the features on the verb are contributed by the construction, but this goes against the lexicalist principles of Minimalism. Moreover, what problems do constructions pose for the foundational assumption that the syntax reflects the semantics? The existence in other Germanic languages of the two types of constructions identified in this thesis should be further investigated as well. Are the path-type and the transition-type equally productive? Germanic languages differ from for example Romance languages in their expression of motion, as observed by Talmy (1985). Hence, do both types of constructions identified here exist in all Germanic languages, and are they absent in Romance languages? These questions provide exciting topics for further research. 166 References Adger, David. 2003. Core Syntax: A Minimalist Approach. New York: Oxford University Press. Anderson, Stephen R. 1985. Inflectional morphology. In Language typology and syntactic description, ed. Timothy Shopen, 150-201. Cambridge; New York: Cambridge University Press. Baker, Mark. 1988. Incorporation: a Theory of Grammatical Function Changing. Chicago: University of Chicago Press. Burzio, L. 1986. Italian Syntax. Dordrecht: Reidel. Carrier, Jill and Janet Randall. 1992. The argument structure and syntactic structure of resultatives. Linguistic Inquiry 23:173-234. Chisholm, Roderick. 1964. The Descriptive Element in the Concept of Action. The Journal of Philosophy 61:613-625. Chomsky, Noam. 1993. A Minimalist Program for Linguistic Theory. In The View from Building 20. Essays in Linguistics in Honor of Sylvain Bromberger, ed. Kenneth Hale and Samuel Jay Keyser, 1-52. Cambridge, MA: MIT Press. Chomsky, Noam. 1995. The Minimalist Program. Cambridge, MA: MIT Press. Chomsky, Noam. 2000. Minimalist Inquiries: The Framework. In Step by Step: Essays on Minimalism in Honor of Howard Lasnik, ed. David Michaels and Juan Uriagereka Roger Martin, 89-155. Cambridge, MA: MIT Press. Chomsky, Noam. 2001. Beyond Explanatory Adequacy. Ms. Cambridge, MA. Davidson, Donald. 1967. The Logical Form of Action Sentences. In The Logic of Decision and Action, ed. Nicholas Reschler, 81-120: University of Pittsburgh Press. Dowty, David. 1991. Thematic proto-roles and argument selection. Language 67:547619. Goldberg, Adele E. 1995. Constructions: A Construction Grammar Approach to Argument Structure. Chicago & London: The University of Chicago Press. Goldberg, Adele E. 1996. Making One’s Way Through the Data. In Grammatical Constructions - Their Form and their Meaning, ed. Masayoshi Shibatani & Sandra A. Thompson, 29-53. New York: Oxford University Press. 167 Goldberg, Adele and Ray Jackendoff. 2004. The English resultative as a family of constructions. Language 80.3:532-568. Grice, H.P. 1975. Logic and Conversation. In Syntax and Semantics 3: Speech Acts, ed. P. Cole and J. Morgan, 41-58. New York: Academic Press. Gruber, J. S. 1965. Studies in Lexical relations, PhD dissertation. MIT. Hoekstra, Teun. 1988. Small Clause Results. Lingua 74:101-139. Jackendoff, Ray. 1983. Semantics and Cognition. Cambridge, MA: MIT Press. Jackendoff, Ray. 1990. Semantic Structures. Cambridge, MA: MIT Press. Jackendoff, Ray. 2002. Foundations of Language. New York: Oxford University Press. Kayne, R. 1984. Connectedness and Binary Branching. Dordrecht: Foris. Langacker, Ronald W. 1990. Concept, image and symbol. The cognitive base of grammar. vol. 1: Cognitive linguistics research. Berlin: Mouton de Gruyter. Larson, Richard K. 1988. On the Double Object Construction. Linguistic Inquiry 19:335-391. Levin, Beth and Malka Rappaport Hovav. 1995. Unaccusativity: at the syntax-lexical semantics interface. vol. 26: Linguistic Inquiry Monographs. Cambridge, Massachusetts: MIT Press. Levin, Beth and Malka Rappaport Hovav. 1999. Two Structures for Compositionally Derived Events. In SALT IX,. eds. Tanya Matthews and Devon Strolovitch, 199-223. Ithaca, NY: Cornell University. Lieber, Rochelle and Harald Baayen. 1997. A semantic principle of auxiliary selection in Dutch. Natural Language and Linguistic Theory 15:789-845. Ludwig, Ilka. 2005. The way-construction in German. Honours paper, Department of Linguistics, University of Canterbury. Marantz, Alec. 1992. The way-construction and the Semantics of Direct Arguments in English: a Reply to Jackendoff. In Syntax and the Lexicon, ed. Tim Stowell and Eric Wehrli, 179 - 188. San Diego: Academic Press. Oehrle, R. 1976. The Grammatical Status of the English Dative Alternation, PhD dissertation, MIT. Radford, Andrew. 2004. Minimalist Syntax: Exploring the structure of English. Cambridge: Cambridge University Press. 168 Rappaport Hovav, Malka and Beth Levin. 1998. Building verb meanings. In The projection of arguments: Lexical and compositional factors., eds. Miriam Butt and Wilhelm Geuder, 97-134. Stanford, CA: CSLI Publications. Rappaport Hovav, Malka and Beth Levin. 2001. An event structure account of English resultatives. Language 77:766-797. Reinhart, Tanya and Eric Reuland. 1993. Reflexivity. Linguistic Enquiry 24:657-720. Seibert, Anja J. 1993. Intransitive Constructions in German and the Ergative Hypothesis. vol. 14: Working Papers in Linguistics. Dragvoll, Norway: University of Trondheim. Seland, Grete. 2001. The Norwegian Reflexive Caused Motion Construction. A Construction Grammar Approach, Department of Linguistics, University of Oslo. Sells, Peter, Annie Zaenen and Draga Zec. 1987. Reflexivization variation: relations between syntax, semantics, and lexical structure. In Working papers in grammatical theory and discourse structure: interactions of morphology, syntax, and discourse., ed. Masayo Iida, Wechsler, Stephen & Zec, Draga, 169-238. Stanford, CA: CSLI Publications. Simpson, Jane. 1983. Resultatives. In Papers in lexical-functional grammar, ed. M. Rappaport & A. Zaenen L. Levin, 143-157. Bloomington: Indiana University Linguistics Club. Talmy, Leonard. 1985. Lexicalization Patterns: Semantic Structure in Lexical Forms. In Language typology and syntactic description, ed. Timothy Shopen. Cambridge; New York: Cambridge University Press. Toivonen, Ida. 2002. The directed motion construction in Swedish. Journal of Linguistics 38:313-345. van Valin, Robert. 1990. Semantic parameters of split intransitivity. Language 66:221-260. Veraart, Fleur. 1996. On the Distribution of Dutch Reflexives. MIT Working papers in Linguistics 10. Verhagen, Arie. 2002. From parts to wholes and back again. Cognitive Linguistics 13:403-439. Verhagen, Arie. 2003. The Dutch way. In Usage based approaches to Dutch, ed. Arie Verhagen & Jeroen van de Weijer (red.). Utrecht: Utrecht: LOT. 169 Verhagen, Arie. 2004. Hoe het Nederlands zich een eigen weg baant - Vergelijkende en historische observaties vanuit een constructie-perspectief. Nederlandse Taalkunde 8:327-346. Zaenen, Annie. 1993. Unaccusativity in Dutch: integrating syntax and lexical semantics. In Semantics and the Lexicon, ed. J. Pustejovsky, 129-161: Kluwer Academic Publishers, Dordrecht. Zwicky, Arnold M. 1977. On clitics. Bloomington, IN: Indiana University Linguistics Club. 170 Appendix 1: First Dutch questionnaire (22 speakers) % of ‘+’ Tarzan hakt zich een weg door T. de jungle. slash REFL a way through the jungle 95.5 ‘Tarzan slashes his way through the jungle.’ Tarzan hakt zich T. door de jungle. slash REFL through the jungle 36.4 ‘Tarzan slashes his way through the jungle.’ Marianne fluit M. zich een weg door de tunnel. whistles REFL a way through the tunnel 36.4 ‘Marianne whistles her way through the tunnel.’ Marianne fluit M. zich door de tunnel. whistles REFL through the tunnel 27.3 ‘Marianne whistles her way through the tunnel.’ Pieter van den Hoogenband zwemt zich in de finale. P. van den H. swims REFL in the final 86.4 ‘Pieter van den Hoogenband swims his way into the final.’ Pieter van den Hoogenband zwemt zich een weg in de finale. P. van den H. swims REFL a way in the final 0 ‘Pieter van den Hoogenband swims his way into the final.’ Elien zwemt zich letterlijk het nieuwe jaar in. E. swims REFL literally the new year in 77.3 ‘Elien literally swims her way into the new year.’ Elien zwemt zich letterlijk in het nieuwe jaar. E. swims REFL literally in the new year 4.5 ‘Elien literally swims her way into the new year.’ Elien zwemt letterlijk het nieuwe jaar in. E. swims literally the new year in 100 ‘Elien literally swims into the new year.’ De dikke man perst zich door de nauwe opening. the fat man squeezes REFL through the narrow opening ‘The fat man squeezes his way through the narrow opening.’ 100 171 % of ‘+’ De dikke man perst zich een weg door de nauwe opening. the fat man squeezes REFL a way through the narrow opening 40.9 ‘The fat man squeezes his way through the narrow opening.’ De klont boter smelt zich een weg van de hete kalkoen. the lump butter melts REFL a way off the hot turkey 4.5 ‘The lump of butter melts its way off the hot turkey.’ De klont boter smelt zich van de hete kalkoen. the lump butter melts REFL off the hot turkey 9.1 ‘The lump of butter melts its way off the hot turkey.’ Frank heeft zich een weg uit de gevangenis gegraven. F. has REFL a way out the prison dug 81.8 ‘Frank dug his way out of prison.’ Frank heeft zich uit de gevangenis gegraven. F. has REFL out the prison dug 77.3 ‘Frank dug his way out of prison.’ Het riviertje kronkelt zich door het dal. the river.DIM winds REFL through the valley 40.9 ‘The river winds its way through the valley.’ Het riviertje kronkelt zich een weg door het dal. the river.DIM winds REFL a way through the valley 54.5 ‘The river winds its way through the valley.’ De zonnebloemen groeien zich naar het licht. the sunflowers grow REFL to the light 9.1 ‘The sunflowers are growing their way to the light.’ De zonnebloemen groeien zich een weg naar het licht. the sunflowers grow REFL a way to the light 31.8 ‘The sunflowers are growing their way to the light.’ De schichtige kinderen wagen zich de straat op. the timid kids dare REFL the street on ‘The timid kids venture onto the street.’ 81.8 172 % of ‘+’ De schichtige kinderen wagen zich op straat. the timid kids dare REFL on street 95.5 ‘The timid kids venture onto the street.’ De studenten roken en drinken zich door de nacht. the students smoke and drink REFL through the night 77.3 ‘The students are smoking and drinking their way through the night.’ De studenten roken en drinken zich een weg door de nacht. the students smoke and drink REFL a way through the night 13.6 ‘The students are smoking and drinking their way through the night.’ De studenten roken en drinken zich de nacht door. the students smoke and drink REFL the night through 59.1 ‘The students are smoking and drinking their way through the night.’ De studenten roken en drinken zich een weg de nacht door. the students smoke and drink REFL a way the night through 13.6 ‘The students are smoking and drinking their way through the night.’ De vluchtelingen zoeken zich the refugees tussen de puinhopen. search REFL between the rubble.heaps 4.5 ‘The refugees are searching their way in between the rubble.’ De vluchtelingen zoeken zich een weg tussen de puinhopen. the refugees search REFL a way between the rubble.heaps ‘The refugees are searching their way in between the rubble.’ 68.2 173 Appendix 2: Second Dutch questionnaire (42 speakers) PART I % of ‘+’ Het voetbalteam wringt zich een weg in het busje. the soccer.team squeezes REFL a way in the bus.DIM 19.0 ‘The soccer team squeezes its way inside the van.’ Het voetbalteam wringt zich in het busje. the soccer.team squeezes REFL in the bus.DIM 97.6 ‘The soccer team squeezes itself inside the van.’ Johan blufte J. zich uit de benarde situatie. bluffed REFL out the awkward situation 81.0 ‘Johan bluffed his way out of the awkward situation.’ Johan blufte J. zich een weg uit de benarde situatie. bluffed REFL a way out the awkward situation 26.2 ‘Johan bluffed his way out of the awkward situation.’ Johan blufte uit de benarde situatie. J. bluffed out the awkward situation 0 ‘Johan bluffed out of the awkward situation.’ Tarzan heeft zich dagenlang een weg T. door de jungle gehakt. has REFL days.long a way through the jungle slashed 61.9 ‘Tarzan slashed his way through the jungle for days.’ Tarzan heeft zich in twee dagen een weg door T. has REFL in two days de jungle gehakt. a way through the jungle slashed 78.6 ‘Tarzan slashed his way through the jungle in two days.’ Opa hijst zich een weg uit zijn stoel. grandpa heaves REFL a way out his chair 0 ‘Grandpa heaves his way out of his chair.’ Opa hijst zich uit zijn stoel. grandpa heaves REFL out his chair ‘Grandpa heaves himself out of his chair.’ 100 174 % of ‘+’ Gordon zong de hitparade in. G. sang de hit.parade in 2.4 ‘Gordon sang into the hit parade.’ Gordon zong zich de hitparade in. G. sang REFL de hit.parade in 81.0 ‘Gordon sang his way into the hit parade.’ Gordon zong zich G. in de hitparade. sang REFL in de hit.parade 42.9 ‘Gordon sang his way into the hit parade.’ Urenlang ploegden de vermoeide wielrenners zich een weg door het zand. hours.long ploughed the tired cyclists REFL a way through the sand 69.0 ‘The tired cyclists ploughed their way through the sand for hours.’ In een uur ploegden de vermoeide wielrenners zich een weg door het zand. in an hour ploughed the tired cyclists REFL a way through the sand 76.2 ‘The tired cyclists ploughed their way through the sand in an hour.’ De mol graaft zich een gang onder de grond. the mole digs REFL a tunnel under the ground 31.0 ‘The mole digs himself a tunnel under the ground.’ Met één trap trapte Jan zich een weg door de deur. with one kick kicked J. REFL a way through the door 33.3 ‘With one kick, Jan kicked his way through the door.’ Met één trap trapte Jan zich door de deur. with one kick kicked J. REFL through the door 28.6 ‘With one kick, Jan kicked his way through the door.’ Het meisje duwt zich een weg door de menigte. the girl pushes REFL a way through the crowd 50.0 ‘The girl pushes her way through the crowd.’ Het meisje duwt zich door de menigte. the girl pushes REFL through the crowd ‘The girl pushes her way through the crowd.’ 69.0 175 % of ‘+’ Het meisje duwt door de menigte. the girl pushes through the crowd 21.4 ‘The girl pushes through the crowd.’ Moeder heeft zich de kamer in gehaast. mother has REFL the room in hurried 54.8 ‘Mother has hurried into the room.’ Moeder is zich de kamer in gehaast. mother is REFL the room in hurried 31.0 ‘Mother is hurried into the room.’ Pieter heeft zich minutenlang in de finale gezwommen. P. has REFL minutes.long in the final swum 0 ‘Pieter swam his way into the final for minutes.’ Pieter heeft zich in twee minuten in de finale gezwommen. P. has REFL in two minutes in the final swum 85.7 ‘Pieter swam his way into the final in two minutes.’ De alcoholisten dronken zich in een staat van bewusteloosheid. the alcoholics drank REFL in a state of unconsciousness 35.7 ‘The alcoholics drank themselves into a state of unconsciousness.’ De alcoholisten dronken zich bewusteloos. the alcoholics drank REFL unconscious ‘The alcoholics drank themselves unconscious.’ 97.6 176 PART II (32 speakers) Is there a difference in meaning between the following pairs of sentences? If so, what is that difference? 1) De tor knaagt zich door de bast heen. the beetle gnaws REFL through the bark PRT ‘The beetle gnaws its way through the bark.’ 2) De tor knaagt door de bast heen. the beetle gnaws through the bark PRT 68.8%: the beetle goes through the bark with its body in (1) but not in (2) ‘The beetle gnaws through the bark.’ 1) Piet gooit zich voor de trein. P. throws REFL in.front.of the train ‘Piet throws himself in front of the train.’ 2) Piet gooit zich naar goud. 62.5%: Piet throws himself in (1) but he throws something else in (2) P. throws REFL to gold ‘Piet throws his way to gold.’ 1) De kat wurmde zich door het veel te kleine the cat squeezed REFL through the much too small kattenluikje. cat.flap.DIM ‘The cat squeezed its way through the way too small cat flap. 2) De kat wurmde zich een weg door het veel te the cat squeezed REFL a way through the much too kleine kattenluikje. small cat.flap.DIM ‘The cat squeezed its way through the way too small cat flap. 15.6%: (2) implies mote effort and more time 37.%: (2) is strange because it implies that the cat flap is a tunnel 177 Appendix 3: Third Dutch questionnaire (30 speakers) % of ‘+’ Het depressieve meisje worstelt zich een moeizame weg the depressed girl wrestles REFL a door het laborious way through the 63.3 bestaan. existence ‘The depressed girl wrestles her laborious way through her existence.’ De geur verspreidt zich een weg door de kamer. the odour spreads REFL a way through the room 16.7 ‘The odour spreads its way through the room.’ De mensen begeven zich een weg naar buiten. the people go REFL a way to outside 10.0 ‘People are going their way outside.’ Het jongetje slaapt zich een weg door de saaie les. the boy.DIM sleeps REFL a way through the boring lecture 20.0 ‘The little boy sleeps his way through the boring lecture.’ De bejaarden wandelen zich een weg door the elderly stroll het park. REFL a way through the park 3.3 ‘The elderly are strolling their way through the park.’ Tarzan heeft zich dagenlang een weg T. has REFL days.long door de jungle gehakt. a way through the jungle slashed 90.0 ‘Tarzan slashed his way through the jungle for days.’ Kraijceck heeft zich K. urenlang door de eerste ronde geslagen. has REFL hours.long through the first round ‘Kraijceck hit his way through the first round for hours.’ hit 13.3 178 What is the difference in meaning between the following two sentences? a) De voetballer slalomt zich een weg langs zijn 53.3%: (a) is more difficult tegenstanders. (e.g. obstacles, effort) the soccer.player slaloms REFL a way past his 20.0%: (a) is a longer path opponents 6.7%: (a) is more directed ‘The soccer player slaloms his way past his opponents.’ b) De voetballer slalomt langs zijn tegenstanders. the soccer.player slaloms past his 3.3%: (a) is repetition 3.3%: no difference opponents ‘The soccer player slaloms his way past his opponents.’ Assume that the following sentence is correct. What does it mean for you? Jan boert zich een weg uit het restaurant. J. belches REFL a way out the restaurant ‘Jan belches his way out of the restaurant.’ a) Jan walks out of the restaurant belching. a) manner: 36.7% b) Jan uses the belching as a means to get out of the b) means: 63.3% restaurant, for example to frighten the guests in such a way that they are going out of his way. Do the next sentences refer to several jumps or can it be also one jump? a) De atlete springt zich een weg naar de finish. the athlete.FEM jumps REFL a way to the finish a) more jumps: 63.3% ‘The athlete jumps her way to the finish.’ b) Carl Lewis springt zich in het Guinness Book of Records. C. L. jumps REFL in the G. B. of R. ‘Carl Lewis jumps his way in to the Guinness Book of Records.’ b) one jump: 86.7% 179 Appendix 4: English questionnaire (31 speakers) % of ‘+’ The soccer player kicked his way into the Guinness Book of Records. 96.8 The soccer player kicked himself into the Guinness Book of Records. 19.4 The alcoholics drank themselves into oblivion. 90.3 The alcoholics drank their way into oblivion. 32.3 The cat squeezed its way through the narrow opening. 77.4 The cat squeezed itself through the narrow opening. 80.6 Mary had to wrestle her way into her tight jeans. 77.4 Mary had to wrestle herself into her tight jeans. 41.9 Grandma heaves herself out of her chair. 96.8 Grandma heaves her way out of her chair. 22.6 The athletes ran their way into a coma. 6.5 The athletes ran themselves into a coma. 77.4 The depressed girl drags her way through life. 38.7 The depressed girl drags herself through life. 83.9 Tarzan slashed his way through the jungle for days. 83.3 Tarzan slashed his way through the jungle in two days. 90.0 He threw himself in front of the train. 100 He threw his way in front of the train. 0 Susie flirted her way into a string of unhappy relationships. 74.2 Susie flirted herself into a string of unhappy relationships. 32.3 180 % of ‘+’ She ate herself to death. 90.3 She ate her way to death. 9.7 John kicked his way through the door with a single kick. 87.1 John kicked himself through the door with a single kick. 6.5 The ivy is winding itself around the tree. 83.9 The ivy is winding its way around the tree. 90.3 The patient coughed his way into a haemorrhage. 22.6 The patient coughed himself into a haemorrhage. 58.1 She had to push her way through the crowd. 100 She had to push herself through the crowd. 45.2 181 Appendix 5: Verbs found in the weg-construction with a Google search for “zich een weg” (the verbs that are underlined were also found in the TLC) 1. Unaccusative verbs Groeien ‘grow’ 2. Unergative verbs Baggeren ‘wade’ Banen ‘?’ Banjeren ‘pace’ B-boyen ‘b-boy’ Benen ‘leg it, hare’ Bladeren ‘leaf’ Blazen ‘blow’ Bluffen ‘bluff’ Bodyslammen ‘body slam’ Boksen ‘box’ Borrelen ‘bubble’ Branden ‘burn’ Brullen ‘roar’ Buffelen ‘wolf down, gobble’ Bulldozeren ‘bulldoze’ Carven ‘carve’ Copuleren ‘copulate’ Counteren ‘counterattack’ Dansen ‘dance’ Dribbelen ‘dribble’ Dromen ‘dream’ Drummen ‘drum’ Ellebogen ‘elbow’ Flippen ‘flip’ Fluisteren ‘whisper’ Forceren ‘force’ Frauderen ‘commit fraud’ Gijpen ‘gybe’ Glibberen ‘slither’ Glijden ‘slide’ Glimlachen ‘smile’ Golven ‘golf’ Grappen ‘joke’ Grommen ‘growl, grumble’ Hamerslingeren ‘hammer-throwing’ Kabbelen ‘ripple’ Kirren ‘coo’ Klauwen ‘claw’ Klikken ‘click’ Klimmen ‘climb’ Klippen ‘clip’ Kluiven ‘gnaw’ Knabbelen ‘nibble’ Knallen ‘bang’ Knokken ‘fight’ Krabbelen ‘scratch’ Krakatauën ‘krakatau’ (nonce word) Kreunen ‘groan’ Krijsen ‘shriek’ Kronkelen ‘wind’ Kruipen ‘crawl’ Lachen ‘laugh’ Liegen ‘lie’ Moorden ‘murder’ Moshen ‘mosh’ (nonce word) Netwerken ‘network’ Neuken ‘fuck’ Niesen ‘sneeze’ Peddelen ‘peddle’ Piepen ‘squeak’ Ploeteren ‘plod’ Praten ‘talk’ Puffen ‘pant, puff’ Puzzelen ‘solve puzzles’ Rausen ‘?’ Redeneren ‘reason’ Rennen ‘run’ Roeien ‘row’ Rommelen ‘mess around’ Ruziën ‘argue, fight’ Schaken ‘play chess’ Scharrelen ‘rummage’ Scheuren ‘drive very fast’ Schreeuwen ‘yell’ Schuifelen ‘shuffle’ Schuimen ‘foam’ 182 Sjezen ‘race’ Slalommen ‘slalom’ Slapen ‘sleep’ Slenteren ‘saunter’ Slijmen ‘brown-nose, sweet-talk’ Slingeren ‘wind’ Sloffen ‘shuffle’ Sluipen ‘steal, sneak’ Snellen ‘rush’ (zijn and hebben aux) Snuisteren ‘nose about, pry into’ Soleren ‘give a solo performance’ Spartelen ‘thrash about’ Spoelen ‘wash, sluice’ Sprankelen ‘sparkle’ Springen ‘jump’ Steunen ‘groan’ Stralen ‘beam, radiate’ Struikelen ‘trip, fall’ Surfen ‘surf’ Tasten ‘grope’ Toeteren ‘honk’ Trommelen ‘drum’ Vechten ‘fight’ Vliegen ‘fly’ Vloeken ‘swear, curse’ Voetballen ‘play soccer’ Werken ‘work’ Worstelen ‘wrestle’ Wroeten ‘root’ Zappen ‘zap’ Zigzaggen ‘zigzag’ Zuchten ‘sigh’ Zuigen ‘suck’ Zwemmen ‘swim’ Zwoegen ‘drudge’ 3. Verbs that can be both transitive and intransitive Beitelen ‘chisel’ Beuken ‘batter’ Bijten ‘bite’ Boren ‘drill’ Breken ‘break’ Dringen ‘push, squeeze’ Drinken ‘drink’ Dwingen ‘force’ Eten ‘eat’ Graven ‘dig’ Hakken ‘chop, slash’ Happen ‘bite’ Houwen ‘chop’ Kloppen ‘beat’ Knagen ‘gnaw’ Knippen ‘cut’ Kopen ‘buy’ Likken ‘lick’ Maaien ‘mow’ Malen ‘grind’ Manipuleren ‘manipulate’ Manoevreren ‘manoeuvre’ Navigeren ‘navigate’ Persen ‘press’ Pikken ‘peck’ Plagiëren ‘plagiarize’ Ploegen ‘plough’ Plukken ‘pick, pluck’ Roken ‘smoke’ Schieten ‘shoot’ Schoffelen ‘hoe’ Schoppen ‘kick’ Schrijven ‘write’ Schroeien ‘singe’ Schroeven ‘screw’ Schudden ‘shake’ Schuren ‘rub’, sand’ Slaan ‘hit’ Slepen ‘drag’ Smelten ‘melt’ Snijden ‘cut’ Spinnen ‘spin, weave’ Spuwen ‘spit’ Stomen ‘steam’ Traden ‘trade’ Trekken ‘pull’ Voelen ‘feel, grope’ Vreten ‘eat’ Wringen ‘wrench’ Zagen ‘saw’ Zingen ‘sing’ 183 4. Transitive verbs Bezemen ‘broom’ Bombarderen ‘bomb’ Duwen ‘push’ Gooien ‘throw’ Hacken ‘hack’ Innoveren ‘innovate’ Kappen ‘chop’ Klieven ‘cleave’ Klutsen ‘whisk’ Kussen ‘kiss’ Lezen ‘read’ Maken ‘make’ Slijpen ‘grind, polish’ Slopen ‘demolish’ Spanken ‘spank’ Stoten ‘thrust’ 184 Appendix 6: Verbs found in TLC with Google (the verbs that are underlined also were found in the weg-construction) 1. Unergative verbs Banen ‘?’ Benen ‘leg it, hare’ Blazen ‘blow’ Bluffen ‘bluff’ Branden ‘burn’ Brullen ‘roar’ Counteren ‘counterattack’ Crossen ‘race’ Dansen ‘dance’ Dribbelen ‘dribble’ Dromen ‘dream’ Drummen ‘drum’ Duiken ‘dive’ Ellebogen ‘elbow’ Fietsen ‘bike’ Fladderen ‘flutter’ Fluisteren ‘whisper’ Forceren ‘force’ Glibberen ‘slither’ Glijden ‘slide’ Golven ‘golf’ Grappen ‘joke’ Hinkstappen ‘triple jump’ Kabbelen ‘ripple’ Klauwen ‘claw’ Klikken ‘click’ Knallen ‘bang’ Knokken ‘fight’ Krabbelen ‘scratch’ Kreunen ‘groan’ Krijsen ‘scream’ Kronkelen ‘wind’ Kruipen ‘crawl’ Lachen ‘laugh’ Liegen ‘lie’ Lopen ‘walk, run’ Moorden ‘murder’ Neuken ‘fuck’ Ploeteren ‘plod’ Praten ‘talk’ Puffen ‘pant, puff’ Puzzelen ‘solve puzzles’ Redeneren ‘reason’ Rennen ‘run’ Rocken ‘rock’ Roeien ‘row’ Rommelen ‘mess around’ Ruziën ‘argue, fight’ Schaatsen ‘skate’ Schaken ‘play chess’ Scharrelen ‘rummage’ Scheuren ‘drive very fast’ Schreeuwen ‘yell’ Schrijden ‘stride’ Schuifelen ‘shuffle’ Schuimen ‘foam’ Slalommen ‘slalom’ Slapen ‘sleep’ Slijmen ‘brown-nose, sweet-talk’ Slingeren ‘wind’ Sluipen ‘steal, sneak’ Solliciteren ‘apply for’ Spartelen ‘thrash about’ Springen ‘jump’ Stralen ‘beam, radiate’ Sukkelen ‘trudge’ Surfen ‘surf’ Vechten ‘fight’ Vloeken ‘swear, curse’ Voetballen ‘play soccer’ Waggelen ‘totter’ Werken ‘work’ Worstelen ‘wrestle’ Wroeten ‘root’ Zappen ‘zap’ Zigzaggen ‘zigzag’ Zuchten ‘sigh’ Zuigen ‘suck’ Zwemmen ‘swim’ Zwoegen ‘drudge’ 185 2. Verbs that can be both transitive and intransitive Beuken ‘batter’ Breken ‘break’ Dringen ‘push, squeeze’ Drinken ‘drink’ Eten ‘eat’ Graven ‘dig’ Hakken ‘chop, slash’ Happen ‘bite’ Knagen ‘gnaw’ Knippen ‘cut’ Lezen ‘read’ Likken ‘lick’ Manipuleren ‘manipulate’ Manoevreren ‘manoeuvre’ Navigeren ‘navigate’ Rijden ‘ride’ Roken ‘smoke’ Schieten ‘shoot’ Schoppen ‘kick’ Schroeien ‘singe’ Schroeven ‘screw’ Schuren ‘rub’, sand’ Slaan ‘hit’ Slikken ‘swallow’ Smelten ‘melt’ Snijden ‘cut’ Spelen ‘play’ Spinnen ‘spin, weave’ Spuwen ‘spit’ Trouwen ‘marry, get married’ Vreten ‘eat’ Wringen ‘wrench’ Zagen ‘saw’ Zingen ‘sing’ 3. Transitive verbs Drukken ‘press’ Duwen ‘push’ Hacken ‘hack’ Innoveren ‘innovate’ Kappen ‘chop’ Klieven ‘cleave’ Ploegen ‘plough’ Stoten ‘thrust’ Tappen ‘draw beer’