External and internal topics in Yucatec Maya

EXTERNAL AND INTERNAL TOPICS IN YUCATEC MAYA* Rodrigo Gutiérrez-Bravo El Colegio de México [email protected] This paper provides a description of topicalization in Yucatec Maya (Mayan, Mexico) based on Aissen‟s (1992) classic analysis of topics and foci in Mayan languages. The relevant data for this description are taken for the most part from oral narratives. While evidence of the existence of external topics in Yucatec Maya has been recently provided by Skopeteas & Verhoeven (2009a,b), in this paper evidence is provided that Yucatec Maya also has topics that behave like internal topics. I argue that there are two different ways of interpreting this state of affairs. One is that Yucatec has both external and internal topics, a possibility that Aissen herself considers for Tzutujil. The other possibility is that the external/internal distinction is not relevant in Yucatec and that topics in this Mayan language are structurally altogether different from the topics proposed by Aissen. Some preliminary evidence is presented supporting this latter possibility. 1. Introduction Mayan languages, which are for the most part verb initial, regularly allow for argument-initial orders that result from topicalization. This is illustrated in (1) and (2) for two different VOS Mayan languages, Tzotzil (Chiapas, Mexico) and Tz‟utujil (Guatemala).1 * I would like to thank the audience of the X Encuentro de Lingüística del Noroeste (Hermosillo, Mexico, November 2008) and Line Mikkelsen for helpful feedback on numerous issues discussed in this paper. All remaining errors are my own. The research reported here was supported in part by the National Council for Science and Technology of Mexico (CONACYT), grant SEP-2004-CO1-47613. 1 The abbreviations used in this paper are the following: ABS absolutive EP epenthesis NUMC numeral classifier ASP aspect ERG ergative PART participle CAUS causative HAB habitual PASS passive CL clitic IND indicative PL plural COMP complementizer ITER iterative PPF present perfect CP completive IRR irrealis PRF perfect DET determiner LOC locative SG singular DM demonstrative NEG negation TOP topic DUR durative NEX negative existential TRNS transitive The glosses ERG and ABS correspond to what is known in traditional Mayan linguistics as the A and B pronominal series. It should be noted that ergativity in Yucatec is split on the basis of aspect, and hence not every instantiation of a pronominal element labeled as ERG or ABS necessarily bears an ergative or absolutive grammatical relation. Where relevant, the bound nature of clitics is signaled by “=”. © 2011 Rodrigo Gutiérrez-Bravo. In Representing Language: Essays in Honor of Judith Aissen, eds. Rodrigo Gutiérrez-Bravo, Line Mikkelsen, and Eric Potsdam, 105-119. Santa Cruz, Ca.: Linguistics Research Center. 106 (1) (2) Rodrigo Gutiérrez-Bravo TZOTZIL [ a ti antz-e]i jun=yo‟on TOP DET woman-CL happily „The wife stays at home happily…‟ ta=xkom ____ i ... . stays (taken from Aissen 1992:49) TZ‟UTUJIL [ja ch‟ooy]i ma x-uu-tij ta ja kéeso ____i . the rat NEG ASP-ERG.3-eat IRR the cheese „The rat didn‟t eat the cheese.‟ (taken from Aissen 1992:72) The topic position in Mayan languages is in turn different from the position occupied by foci, which corresponds to the position to the immediate left of the verb. The relative order of topic and focus is illustrated in (3) and (4) for Tzotzil and Tz‟utujil, respectively. (3) TZOTZIL [TOPIC a prove tzeb-e] [FOCUS sovra ] TOP DET poor girl-CL leftovers „It was leftovers that the poor girl was given.‟ (4) ti ch‟ak‟bat. was.given (taken from Aissen 1992:51) TZ‟UTUJIL [TOPIC ja gáarsa] [FOCUS cheqe ch‟uu‟ ] n-ee-ruu-tij . the heron only fish ASP-ABS.3PL-ERG.3-eat „It‟s only fish that the heron eats.‟ (taken from Aissen 1992:72) Although these word order facts had been regularly observed in the literature on Mayan languages, Aissen‟s (1992) seminal analysis on topics and foci in Mayan languages was the first to identify that even though all Mayan languages allow preverbal topics in the position immediately to the left of the focus position, not all these topics are equal. Aissen observed that the topics of Mayan languages like Tzotzil and Jakaltek (Guatemala) are in fact prosodically, syntactically and pragmatically different from the topics of other Mayan languages like Tz‟utujil. Aissen observes that the topics of Tzotzil (and Jakaltek) are less integrated into basic clause structure, and so she suggests that the differences between the topics of Tzotzil and those of Tz‟utujil are directly related to different structural properties. Specifically, in Aissen‟s analysis the topics of Tz‟utujil surface in [Spec, C], a clause-internal position, and consequently they are analyzed as internal topics (see also Aissen 1999). The topics of Tzotzil (and Jakaltek), on the other hand, are taken to be base-generated in a position external to the sentence as a whole (i.e. they are not derived by movement). This difference is illustrated in (5), where E stands for Expression. Aissen further takes the focus position in both languages to be [Spec, I]. External and Internal Topics in Yucatec Maya (5) a. Internal topic 107 b. External topic CP 3 topic i C‟ 3 C IP 5 …t i … E 3 topici CP # IP 5 …(proi)… In this paper, I provide a description of topics in Yucatec Maya (the Mayan language spoken in the Yucatán Peninsula, México; henceforth Yucatec). The goal of this description is to determine whether the sentence topics of Yucatec are external or internal. For this purpose, I use Aissen‟s original diagnostics for external and internal topics. My conclusion is that Yucatec has topics that show the behavior characteristic of external and topics that show the behavior characteristic of internal topics. I argue that there are two possible interpretations of this state of affairs. The first one is that Yucatec is a language that has both external and internal topics. This possibility was in fact suggested by Aissen herself (Aissen 1992:73, fn. 30), since she observed that there is some evidence that Tz‟utujil also has external topics besides the internal topics characteristic of this language, with each kind of topic serving different pragmatic functions. The second possibility is that topics in Yucatec are altogether insensitive to the external/internal distinction, and hence that they are different from both the structures in (5). I provide some preliminary evidence in favor of the latter possibility. This paper is structured as follows. In section 2, I introduce some preliminaries of the syntax of Yucatec that will later be used to identify sentence topics in the language. In section 3, I present the evidence that indicates that Yucatec has topics like the external topics found in Tzotzil. In section 4, I present the evidence that shows that Yucatec also has topics like the internal topics of Tzutujil. In section 5, I discuss the analytical implications of this state of affairs for Aissen 1992, and in section 6, I present my conclusions. 2. Topics in Yucatec Maya 2.1. Preliminary Description Most clauses in Yucatec consist minimally of the verb and a proclitic (glossed ERG) crossreferencing the subject of the verb. Most of the time, the proclitic is preceded by an auxiliary particle or verb. The main verb in turn displays a series of suffixes (glossed ABS) that agree with the object. This minimal structure is shown in (6).2 2 All Yucatec examples are presented according to the orthographic conventions of the Academia de la Lengua Maya de Yucatán and so they do not necessarily reflect their phonetic form accurately. Unless specifically indicated, the name after each example corresponds to the source in my corpus that the example is taken from; all the texts in the corpus used for this study correspond to oral narratives; they all correspond to the „eastern‟ variety of Yucatec, spoken in the east and south of the state of Yucatán and in the state of Quintana Roo. The narratives Gigante 108 (6) Rodrigo Gutiérrez-Bravo K-in w-il-ik-ech. HAB-ERG.1SG EP-see-IND-ABS.2SG „I see you.‟ For the purpose of structural description I assume that the verb heads the VP and that the aspect particle and the ergative pronoun together form the head of IP (Gutiérrez-Bravo and Monforte 2008). I further assume that an unexploded CP in turn is the maximal projection that immediately dominates IP, just as in the analysis in Aissen 1992. In Yucatec, intransitive clauses in the completive aspect and some intransitive clauses in irrealis mood are constituted by the verb alone, as in (7). In these cases I assume the same structure as the one for (6), with the exception that in these cases I0 is null. None of my assumptions regarding the nature of V0 and I0 in Yucatec are in any way crucial for the discussion that follows. (7) Sa‟at-ø. lose.PASS-ABS.3SG „He/she/it got lost.‟ Now, Yucatec has a number of syntactic properties that make it different from most other Mayan languages, word order being the one that is most immediately obvious (see Durbin and Ojeda 1978). Specifically, the precise characterization of word order in Yucatec is still an ongoing debate, with some works describing it as VOS (Bohnemeyer 2002, Skopeteas and Verhoeven 2005, Skopeteas and Verhoeven 2009a, Skopeteas and Verhoeven 2009b), and other works arguing that SVO is the unmarked word order of the language‟s transitive clauses (Briceño Chel 2002, Gutiérrez-Bravo and Monforte 2008, Gutiérrez-Bravo and Monforte 2010). Here I adopt the latter analysis and hence I assume that constructions like (8) and (9) are instances of unmarked word order and not of subject topicalization.3 (8) ko‟olel-o‟ t-u ts‟-aj-ø u ma‟alob nook‟. woman-CL CP-ERG.3 put-PRF-ABS.3SG ERG.3 good clothes „...and the woman put on her good clothes...‟ Góngora Pacheco (1990:19) Le DM (9) ch‟ak-ik-ø u che‟-il. 1SG HAB-ERG.1SG chop-IND-ABS.3SG ERG.3 tree-RL „I used to chop its trees (the cornfield‟s).‟ Teen k-in (MDG-B:136) (Roberto Salazar), Sapo (Humberto Medina) and Sonámbulo (Eduardo Pat) were collected during fieldwork in Laguna Kaná, Quintana Roo, in 1996-1997. Examples labeled as MDG-B are taken from narratives in Monforte et. al. (2011). These narratives were collected during fieldwork in Timul and Peto, southern Yucatán, in 2006-2007. The absence of a corpus reference indicates an elicited example. 3 For the specific details of how the SVO order is derived in Yucatec, I refer the reader to Gutiérrez-Bravo and Monforte 2010. External and Internal Topics in Yucatec Maya 109 A direct consequence of this assumption is that non-focal, preverbal subjects cannot be automatically taken to be sentence topics.4 With this one exception, however, identifying sentence topics in Yucatec is fairly straightforward. I now present the criteria used from here on to characterize topicalized XPs. These are the criteria with which the corpus examples were selected. 2.2. Word Order Although preverbal word order cannot be used as a diagnostic for subject topicalization in transitive clauses, it is a reliable diagnostic for topicalization in most other cases. For instance, most studies of word order in Yucatec agree that the unmarked word order of intransitive clauses is VS (see Skopeteas and Verhoeven 2005, and Gutiérrez-Bravo and Monforte 2010), as in (10). Hence in intransitive clauses with fronted definite subjects, the subject can be taken to be a sentence topic, as in (11).5 (10) (11) Lekan lúub-uk-ø le ka‟anal ja‟ … when fall-IRR-ABS.3SG DM tall water „When the rain falls...‟ (MDG-B:13) Pues si le kool-o‟ beychaj-ø-i‟. yet if DM cornfield-CL be.possible.PRF-ABS.3SG-LOC „Yet it was possible (to do) the cornfield there.‟ (Gigante-96) Similarly, given the VO order of Yucatec, in clauses with fronted definite objects, the object can be taken to be a sentence topic, as in (12). (12) [Lel-o‟]i ma‟ k il-m-aj-ø _____i to‟on-i‟. see-PPF-PRF-ABS.3SG 1PL-NEG „That, we did not get to see it.‟ DM-CL 2.3. NEG ERG.1PL (MDG-B:133) The Topic Clitic =e’ Yucatec has a phrasal enclitic =e’ that signals sentence topics, similar to the topic particles found in other Mayan languages like Tzotzil (see (1) above). Examples are presented below. 4 This is in fact entirely independent of whether or not the SVO analysis of Yucatec is assumed. Skopeteas and Verhoeven (2009b), for instance, assume that the unmarked word order of Yucatec is VOS. However, they still acknowledge that Yucatec has preverbal subjects that are neither topics nor foci. 5 It is important to control for definiteness in these cases, since, in the absence of other cues for topicality, fronted bare intransitive subjects and direct objects in Yucatec regularly function as foci and not as topics. 110 (13) (14) Rodrigo Gutiérrez-Bravo Tumen leti‟=e’ k-u beet-ik-ø jun p‟iit because 3SG=TOP HAB-ERG.3 make-IND-ABS.3SG one bit „Because he, he made a bit of it for himself.‟ u ERG.3 ti‟al-i‟. for-CL (MDG-B:132) [Tuláakal ba‟ax k‟a‟abéet-ø teen]=e’i k-in taa-s-ik-ø ____i. all what necessary-ABS.3SG 1.SG=TOP HAB-ERG.1SG come-CAUS-IND-ABS.3SG „Everything that was necessary to me, I would bring.‟ (MDG-B:30) In principle, the topic clitic should be the most reliable diagnostic for identifying a topicalized XP in Yucatec. Unfortunately, this topic clitic diagnostic has one limitation. Yucatec has a fair number of phrasal enclitics with different syntactic, semantic and intonational functions. The more prominent of these are a set of distal deictic clitics (see Hanks 1990 and Lehmann 1998) that encode the relative distance between the speaker and a nominal referent according to the following scale: =a’ D1 (i.e. “this”), =o’ D2 (i.e. “that”) and =e’ D3. Now, a number of functional elements obligatorily require the presence of one of these distal clitics. The nominal determiner le in (10) and (11) is precisely one of these elements. However, in Yucatec only one clitic is allowed to attach to the right edge of an XP. When two different clitics originating from two different sources would be expected to appear, one of them is suppressed. And, specifically, when the topic clitic and one of the deictic clitics compete for this unique position, it is the distal clitics that win. This is the reason why no topic enclitic is observed in the right edge of the topicalized XPs in the examples in (11) and (12); its presence is blocked by the distal clitic –o’. The example in (15) is particularly telling in this respect. In this example a temporal adjunct has been fronted from its unmarked postverbal position. We are further certain that it has been fronted to a topic position because it appears to the left of the focus position (see §1), which in this case is occupied by a different temporal adverbial (compare with examples (3) and (4)).6 However, the deictic clitic =o’ associated with le blocks the presence of the topic clitic =e’ and so we have a clear case of a sentence topic that does not bear the topic clitic. (15) [ Le tiempo k-in w-a‟al-ik-ø teech]=o’, [FOCUS téemprano ] DM time HAB-ERG.1SG EP-say-IND-ABS.3s 2SG=CL early k lik‟-il meyaj. ERG.1PL raise-IND work „At that time I‟m telling you (about), we would rise EARLY to work.‟ (MDG-B:26) In spite of this limitation, the topic clitic can still be used to identify topics not introduced by the determiner le and to distinguish unmarked preverbal transitive subjects (such as the subject of (9)) from true topicalized transitive subjects like the one in (13). Hence I regularly use this clitic as a diagnostic for topicality in what follows. 6 In Mayan languages, the focus position to the immediate left of the verb or auxiliary is unique (see Aissen 1992 for details). Hence we can rule out the possibility that the leftmost temporal expression in this example is also a focus. External and Internal Topics in Yucatec Maya 3. 111 External Topics in Yucatec In this section I present the data that shows that Yucatec has topics like the external topics of Tzotzil and Jakaltek. Two recent studies on topicalization in Yucatec, Skopeteas and Verhoeven 2009a, and Skopeteas and Verhoeven 2009b, conclude that topics in this language are external according to Aissen‟s characterization. These works, however, do not discuss the complete number of diagnostics that Aissen gives to identify external topics. Still, I have indeed found in my corpus examples of topicalization that cover the whole range of diagnostics originally proposed by Aissen. I now present the relevant data. Aissen notes as a first defining characteristic of the external topics of Tzotzil and Jakaltek that they are separated from the sentence that follows by an intonational pause. In similar way, in an instrumental study Avelino (2009) observes that Yucatec topics that bear the topic clitic =e’ are separated from the rest of the clause by an intonational pause, represented in (17) with a comma. In contrast, in (16) no intonational pause between the preverbal subject and the remainder of the clause is observed (I refer the reader to Avelino‟s work for the complete pitch tracks and other instrumental evidence of this contrast). This is the first piece of evidence that Yucatec has external topics like those of Tzotzil and Jakaltek. (16) Le áak-o’ t-u jaan-t-aj-ø DM turtle-CL CP- ERG.3 eat-TRNS-PRF-ABS.3SG „The turtle ate grass.‟ (17) áak=e’, t-u jaan-t-aj-ø turtle=TOP CP-ERG.3 eat-TRNS-PRF-ABS.3SG „The turtle ate grass.‟ Le DM su‟uk. grass su‟uk. grass (Avelino 2009:9) This evidence, however, is probably not as strong as we would like. This is because the subject NP is introduced by the demonstrative determiner le, and so, strictly speaking, it is not possible to tell whether the =e’ clitic in (17) is truly the topic clitic or instead the homophonous distal clitic D3 discussed in §2.3.7 Fortunately, the remaining diagnostics developed by Aissen (1992) to identify external topics provide clearer results in Yucatec. Secondly, Aissen (1992) points out that external topics can be doubled by a pronominal element, very much in the way that dislocated XPs are in many European languages. This is indeed what is observed in Yucatec. Skopeteas and Verhoeven (2009a) provide the example in (18) as a case in point and further argue that it is evidence that topics in Yucatec are external. In fact these doubled topical XPs need not be lexical in Yucatec. As shown in examples (19) and (20), it is quite common for the doubled sentential topic to be a pronominal element itself. As can be seen in (17) and further examples in this section, external topics are not necessarily doubled by a pronominal element. This is presumably related to the fact that Yucatec, like Tzotzil, is a pro-drop language. One reason to think that the =e’ clitic in (17) is in fact the topic clitic is that the presence of the D2 distal clitic =o’ in (16) does not bring with it an intonational pause like the one observed in (17). It would thus be odd if some of the clitics in the distal clitic paradigm brought with them an intonational pause but others didn‟t. 7 112 (18) Rodrigo Gutiérrez-Bravo Le ah koonol-o’, leti’ DM master seller-CL 3SG „The vendor, he‟s running.‟ tun DUR.ERG.3 y-áalkab.8 EP-run (Skopeteas & Verhoeven 2009a:165) (19) To’on=e‟, jach u jaajil-e‟, mina‟an-ø to’on mix jun p‟éel quincena. 1PL=TOP very ERG.3 truth-TOP NEX-ABS.3SG 1PL not.even one NUMC salary „Because us, truly, we don‟t even have a salary‟ (Lit. „Because us, truly, not even a salary exists for us.‟) (MDG-B:13) (20) Tumen leti’=e‟ jach way Timul-e‟ leti’-e‟. because 3SG=TOP very here Timul-CL 3SG-CL „Because he, he really is from here from Timul.‟ (MDG-B:23) Thirdly, as a result of this property Aissen concludes that external topics are base-generated. She goes on to show that, consequently, it is not surprising that external topics can be extracted from island contexts in Tzotzil. Again, sentence topics can be found in Yucatec that show this behavior. In (21) the topic corresponds to the subject of the relative clause embedded in an NP, as indicated by the underscore. (21) Pues leti’ob=e‟i [NP le meyaj [RC ___i k-u beet-ik-ø-o‟ob]]=o‟ since 3PL=TOP DM work HAB-ERG.3 do-IND-ABS.3SG-PL-CL ma‟ u ti‟al-o‟ob-i‟. NEG ERG.3 for-PLUR-NEG „Because them, the work (they) did was not for their benefit.‟ (MDG-B:131) The last diagnostic proposed by Aissen is the possibility of external topics to function as hanging topics. These are understood here to be sentence topics that are not linked, via coreference or binding, to any element in the following sentence (Aissen 1992:70). Once again, there are topics in Yucatec that display this behavior. Two examples are presented below, one of which has a pronoun as a sentence topic, while the other one has a full lexical XP in this function. (22) Tuben leti=e‟ [ ma‟ p‟il-a‟an-ø u y-ich ka‟ líik‟-ij ]. because 3SG=TOP NEG open-PART-ABS.3SG ERG.3 EP-eye COMP raise-ABS.SG.PRF „Because he, his eyes were not open when he rose up.‟ (Sonámbulo-16) (23) Le Nenela’-o‟ [ t-u máan u tóok-la‟an-t-ø-o‟ob u Nenela-CL DUR-ERG.3 pass ERG.3 burn-ITER-TRNS-ABS.3-PL ERG.3 y-otoch y-éet wíinik-il-o‟ob]. EP-house EP-fellow man-RL-PLUR „As to (the town of) Nenelá, they went burning down the houses of their fellow men.‟ DM (MDG-B:133) 8 The gloss and free translation of this example are my own. External and Internal Topics in Yucatec Maya 113 It might be argued that (22) is not the best example of a hanging topic. This is because the topicalized third person pronoun leti’ can be claimed to be the true lexical possessor of the nominal expression u yich, „his eyes‟, as in u yich leti’, „his eyes of him‟. In this kind of construction, the possessor is a full XP that is cross referenced by the third person ergative proclitic u, which functions as an instance of possessor agreement in nominal expresions in Yucatec (as in many other Mayan languages). Consequently, it could be claimed that leti’ in (22) is not a hanging topic, but rather the possessor NP extracted via movement. Even if this were the case, this claim cannot be extended to (23), though. In this case, the nominal expression Nenela’ is not linked to any element of the clause that follows it. Observe that Yucatec does actually have the grammatical means of cross-referencing displaced locative XPs via a locative pronominal enclitic =i’, which regularly attaches to the main verb of a clause, as in example (11). This clitic, however, is not observed in (23), which points to the conclusion that Nenelá in this case is indeed a hanging topic. Further evidence for the existence of hanging topics in Yucatec can also be found in Skopeteas and Verhoeven 2009a. As such, Yucatec displays topics that show all the diagnostics for external topics. This corroborates the observation by Skopeteas and Verhoeven (2009a,b) that there are topics in Yucatec that are external in nature. However, in the following section I provide evidence that shows that Yucatec also has topics like the internal topics of Tz‟utujil. 4. Evidence for Internal Topics in Yucatec In this section I present the data that shows that Yucatec also has internal topics and, as such, it cannot be grouped with languages that only have external topics. Aissen observes four properties of the internal topics of Tz‟utujil that distinguish them from the external topics of Tzotzil and Jakaltek. As I discuss in what follows, there are topics in Yucatec that show three of these four properties, specifically, topicalization of 3rd person pronouns, the possibility of embedded topicalization, and the possibility to have topicalized XPs functioning as continuing topics.9 Regarding the first property of internal topics, Yucatec readily allows the topicalization of 3rd person pronouns as illustrated in the examples below. (24) (25) 9 Pero leti’=e‟ k-u p‟áat-al te‟ jool-o‟. but 3SG=TOP HAB-ERG.3 stay-IND LOC door-CL „But she would stay at the door (of the corral).‟ (MDG-B:27) Entonces leti’=e‟ ka t-u k‟eb-aj-ø u jool le caja-o‟. then 3SG=TOP COMP CP-ERG.3 open-PRF-ABS.3SG ERG.3 lid DM box-CL „Then he opened the box‟s lid.‟ (Sapo-13) The fourth property that distinguishes internal from external topics is the absence of an intonational pause between the topic and the clause that follows. However, a full instrumental analysis, which cannot be undertaken here, is required to accurately determine whether or not this intonational pause is absent in the data presented in this section. See also §5 below. 114 Rodrigo Gutiérrez-Bravo It could be argued that the possibility of having topicalized 3rd person pronouns cannot be taken as conclusive evidence that Yucatec has internal topics, since examples (20), (21) and (22) also show topicalized 3rd person pronouns, but they are instead clear cases of external topics. Granting that this argument might be correct (see §5), Yucatec still shows the other two syntactic diagnostics that characterize the internal topics of Tz‟utujil. First, Yucatec shows embedded topics. This is illustrated in examples (26) and (27), where fronted topics are included in regular brackets and the embedded CP that functions as the complement of the matrix verb is signaled in boldface. This property is expected if topics are allowed to have [Spec, C] as their landing site,10 but it is unexpected if topics are only allowed to surface in a base-generated position external to the utterance as a whole. (26) ts‟o‟ok ka in w-a‟al-ø teech-e‟11 [CP [TOP le tiempo táan ERG.3 end COMP ERG.1 EP-say-ABS.3SG 2SG-CL DM time DUR u meen-t-ik-ø]=o‟, [TOP kan taa-k-ø]=e‟, táan u ERG.3 do-TRNS-IND-ABS.3SG=CL when come-IRR-ABS.3SG=TOP DUR ERG.3 jats‟-ik-en]. beat-IND-ABS.1SG „Lastly, I‟ll tell you that at that time when he used to do it (get intoxicated), when he came (back home), he would beat me.‟ (MDG-B:29) (27) J-taal12 in w-a‟al-ik-ø teech-e‟ de [CP que [TOP le kool]i =o‟, CP-come ERG.1SG EP-say-IND-ABS.3SG 2SG-CL of that DM cornfield=CL ts‟o‟ok-a‟an-ø ___i ]. finish-PART-ABS.3SG „I‟ve come to tell you that the cornfield has (now) been done.‟ (Gigante-89) U The examples above are from oral narratives, but it is also significant that I have found robust and productive use of internal topics in elicitations carried out for the purpose of determining how pronominal reference works in Yucatec. Specifically, when presented with the Spanish equivalent of clauses like (28), where the subject of the embedded complement clause must be different from the subject of the matrix clause, speakers systematically produce constructions like (29), where the embedded subject is realized as a topicalized third person pronoun. Observe that examples like (29) provide particularly strong evidence for the existence of internal topics in Yucatec, since they simultaneously show two of the defining properties of internal topics, namely, topicalization of third person pronouns and the possibility of the sentence topic to appear in an embedded context. 10 Or to adjoin regularly to CP when there is more than one topic, as in (26), where two temporal expressions are topicalized. 11 The clitic =e’ in this example and in (27) and (29) is homophonous but different from the topic clitic (see §2.3). Specifically, it seems that this clitic optionally signals the right edge of an intonational phrase, just like the clitic =un in Tzotzil described in Aissen 1992. 12 In this context, the verb taal „to come‟ should be inflected with the 1 st person singular absolutive suffix /-en/. I take the absence of this suffix to simply be a speech error. External and Internal Topics in Yucatec Maya 115 (28) The man said that she got lost in the jungle. (29) Le nojoch máak=o t-u y-a‟al-aj-ø=e‟ [leti=e’ sa‟at-ø DM big person-CL CP-ERG.3 EP-say-PRF-ABS.3SG=CL 3SG=TOP lose.PASS-ABS.3SG ich k‟áax]. in jungle „The man said that she got lost in the jungle.‟ Secondly, in the examples below I present the evidence that these topics can be continuing topics, which is again a property that corresponds to internal and not to external topics. Aissen notes that external topics correspond to new or shifted topics. Consequently, once that a referent has been established as a topic it is not referred to again using an external topic until the topic shifts. This is not what is observed in Yucatec. Instead topics in Yucatec are like the topics of Tz‟utujil in that they do not need to shift or refer to a new topic, but instead they can (and in fact regularly do) refer to a continuing topic over a span of discourse. This can be observed in the two adjacent clauses in the text fragment presented in (30)-(31). The preceding part of the narrative that this text fragment comes from mentions how the narrator and her family were about to go out to the town square for a festival when they suddenly noticed two strangers lurking in the vicinity. The strangers later turned out to be thieves.13 (30) (31) Pero to’on=e‟ ma‟ k ojel máax-i‟… but 1PL=TOP NEG ERG.1PL know who-NEG „But we did not know who they were….‟ jóok‟-o‟on, to’on=e‟ bin-o‟on. 1PL=TOP COMP exit-ABS.1PL 1PL=TOP go-ABS.1PL „We, when we walked out (of the house), we just left.‟ (MDG-B:26) to’on=e‟, ka (MDG-B:26) Summing up, the data presented in this section show that topics in Yucatec display most of the properties that are characteristic of the internal topics of Tz‟utujil. As such, it is not accurate to classify Yucatec with Mayan languages that have external topics only, such as Tzotzil and Jakaltek. In the following section, I briefly discuss the implications that this has for the standard distinction made between Mayan languages that have external topics only and those that only have internal topics. 5. Discussion Having presented evidence that Yucatec has both external and internal topics, it is natural to ask if there is some morphosyntactic cue that systematically distinguishes the two kinds of topics in this language. However, so far I have found no evidence to this effect. For instance, the topic clitic =e’ might seem to be a natural candidate for such a morphosyntactic cue. But this clitic In (30), ojel „know‟ is a verboid that does not appear with the aspect particles characteristic of Yucatec clauses (§2.1); see Verhoeven 2007. 13 116 Rodrigo Gutiérrez-Bravo does not appear exclusively with either kind of topic. It is found both with external topics (as in examples (17) and (19-22)), and with internal topics (as in examples (24), (25), and (29-31)). Another altogether different cue that might distinguish internal and external topics in Yucatec might be the intonational break that is observed after external topics in general (as in (17)), but not after the internal topics of Tz‟utujil (Aissen 1992:73). However, to corroborate this possibility, a complete instrumental analysis is required of the data where internal topics are observed, and such an analysis cannot be undertaken here. Because of this absence of a systematic cue to distinguish internal and external topics in Yucatec, one might be inclined to analyze every topic in this language as being external, or alternatively, to analyze every topic as internal. A possible analysis of the data presented so far could then be that all topics in Yucatec are external; it would just happen that external topics in Yucatec (in contrast with those of Tzotzil and Jakaltek) show some of the properties that are characteristic of the internal topics of Tz‟utujil, such as topicalization of third person pronouns and the possibility to function as continuing topics. However, this analysis is hard to reconcile with the data from embedded topicalization in (26), (27), and (29), where it is clear that the topicalized NPs in these examples are sentence-internal and not sentence external. The other alternative (i.e. an analysis where all the sentence topics of Yucatec are clause internal and hence derived by movement) seems in turn to be readily falsified by the Yucatec data showing hanging topics and extraction of topics from island contexts. As such, two different analyses of the Yucatec data come to mind. Following Aissen‟s (1992:73, fn. 30) observation that Tz‟utujil appears to have both external and internal topics (see §1), it can be claimed that Yucatec has the two kinds of topics illustrated in (5). If this is the correct analysis, then the relevant crosslinguistic distinction would be between Mayan languages that have both external and internal topics, like Yucatec and Tz‟utujil, and Mayan languages that have external topics only, like Tzotzil and Jakaltek. It is worth mentioning that the possibility of a language having both internal and external sentence topics is actually not as odd as it would appear to be from a purely Mayan perspective. Albeit with a number of important differences, evidence can be found in Spanish that some topics are very closely integrated with the clause, whereas others instead seem to be prefixed or adjoined to the clause (see Zubizarreta 1998 and Gutiérrez-Bravo 2005). There is, however, a different possibility altogether, namely, that topics in Yucatec are insensitive to the external/internal distinction, and so that they are altogether different from both the topics of Tzotzil (and Jakaltek) and those of Tz‟utujil. In order to situate this possibility in the bigger picture provided by Aissen‟s (1992) analysis, it is worth considering its central contributions again. At the heart of Aissen‟s analysis is the proposal that in Mayan languages topics generated by movement (internal topics) systematically have a landing site, [Spec, C], which is different from the structural position occupied by base-generated topics (external topics), which is external to the clause. Now, as previously observed, topics in Yucatec cannot be analyzed as being exclusively internal or external. Rather, if anything, they would appear to show the properties of both external and internal topics. An analysis can thus be developed that derives a third kind of topic by combining two of the structural properties that Aissen proposes to establish the external/internal distinction. Specifically, it can be proposed that the topics of Yucatec are base-generated in the left edge of the clause and not derived via movement (like the External and Internal Topics in Yucatec Maya 117 topics of Tzotzil and Jakaltek), but that they are base generated in [Spec, C] (thus occupying the same structural position as the internal topics of Tz‟utujil), as in (32). (32) CP 3 topici C‟ 3 C IP 5 …(proi)… This analysis has a number of advantages. It accounts for the fact that topics in Yucatec can function as hanging topics and that they are insensitive to island contexts, since the topic does not undergo movement at any point. Crucially, this analysis also accounts for the fact that topics in Yucatec are observed in embedded contexts like (26), (27), and (29), where, as just mentioned, it is clear that the topicalized NPs in these examples are sentence-internal and not sentence external. This analysis further makes an interesting prediction. Since this analysis does not make a distinction between external and internal topics, it is expected that we should find topicalized XPs that simultaneously exhibit properties characteristic of external and internal topics, for instance, a hanging topic in an embedded context or a topic extracted from an island that simultaneously functions as a continuing topic.14 Now recall from §4 that the use of topicalized third person pronouns cannot be used as conclusive evidence for the internal nature of a topic since topicalized third person pronouns are found in Yucatec in contexts that clearly correspond to external topics such as (20), (21) and (22). These are precisely cases where we see a topic simultaneously displaying a property characteristic of external topics and a property characteristic of internal topics, and hence provide evidence in favor of the analysis in (32). This evidence should probably only be considered preliminary at this point, since (in principle, at least) a more thorough corroboration of this analysis requires a larger number of cases of topics simultaneously showing the complete range of properties of external and internal topics. A corpus larger than the one used for this paper is required for this purpose, and hence I leave this issue open for future investigation. 6. Conclusions In this paper I have provided a description of topicalization in Yucatec Maya following Aissen‟s (1992) classic analysis of topics and foci in Mayan languages. The data for this description is taken mostly from oral narratives. An inspection of the data corroborates the observation made in Skopeteas and Verhoeven (2009a,b) that topics in Yucatec show properties similar to those of 14 Moreover, such behavior can be used to empirically distinguish between this proposal and the previous alternative where it is instead claimed that Yucatec has both external and internal topics, each kind with the formal properties originally proposed in Aissen‟s analysis, since in the previous alternative we do not expect to find a topic simultaneously displaying the properties of external and internal topics. I am thankful to Line Mikkelsen for detailed discussion of these latter issues. 118 Rodrigo Gutiérrez-Bravo the external topics of Tzotzil and Jakaltek. However, the data also clearly indicates that topics in Yucatec can exhibit the properties of the internal topics of Tz‟utujil. I have concluded from these data that topics in Yucatec cannot be characterized as being exclusively external or exclusively internal. I have suggested that there are two ways of understanding this state of affairs. The first one is that Yucatec is a language that has both external and internal topics, a possibility originally suggested by Aissen for Tz‟utujil. If this is the correct analysis, the relevant crosslinguistic distinction should be understood as a distinction between Mayan languages that have both internal and external topics (Yucatec and Tz‟utujil) and Mayan languages that have only external topics (Tzotzil and Jakaltek). Alternatively, topics in Yucatec can be analyzed as being altogether different from both the topics of Tzotzil/Jakaltek and those of Tz‟utujil. Specifically, I have proposed an alternative where topics in Yucatec are base-generated in [Spec, C]. Hence they occupy the same structural position as the internal topics of Tz‟utujil, but they are not derived via movement, just like the external topics of Tzotzil and Jakaltek. Further research should clarify which of these two alternatives is correct. References Aissen, Judith. 1992. Topic and focus in Mayan. Language 68:43-80. Aissen, Judith. 1999. External possessor and logical subject in Tz'utujil. External possession, eds. Doris L. Payne and Immanuel Barshi, 167-193. Amsterdam/Philadelphia: John Benjamins. Avelino, Heriberto. 2009. Intonational patterns of topic and focus in Yucatec Maya. New perspectives in Mayan linguistics, eds. Heriberto Avelino, Jessica Coon and Elisabeth Norcliffe, 1-22. Cambridge, Mass: MIT Working Papers in Linguistics. Bohnemeyer, Jürgen. 2002. The grammar of time reference in Yukatek Maya. Munich: Lincom Europa. Briceño Chel, Fidencio. 2002. Topicalización, enfoque, énfasis y adelantamiento en el maya yukateco. La organización social entre los mayas prehispánicos, coloniales y modernos, eds. V. Tiesler Blos, R. Cobos and M. Greene Robertson, 374-387. Mexico City/Mérida: INAH/UADY. Durbin, Marshall, and Fernando Ojeda. 1978. Basic word order in Yucatec Maya. Papers in Mayan linguistics, ed. Nora England, 69-77. Columbia: University of Missouri. Góngora Pacheco, María Luisa. 1990. U tzikbalilo'ob Oxkutzcaab yéetel Mani. Mérida, Yucatán: Maldonado Editores. Gutiérrez-Bravo, Rodrigo. 2005. Structural markedness and syntactic structure: a study of word order and the left-periphery in Mexican Spanish. New York: Routledge/Taylor and Francis. Gutiérrez-Bravo, Rodrigo, and Jorge Monforte. 2008. La alternancia sujeto-inicial/verbo-inicial y la Teoría de Optimidad. Teoría de Optimidad: Estudios de sintaxis y fonología, eds. Rodrigo Gutiérrez-Bravo and Esther Herrera Zendejas, 61-99. Mexico City: El Colegio de México. External and Internal Topics in Yucatec Maya 119 Gutiérrez-Bravo, Rodrigo, and Jorge Monforte. 2010. On the nature of word order in Yucatec Maya. Information structure in indigenous languages of the Americas, eds. J. Camacho, R. Gutiérrez-Bravo and L. Sánchez, 139-170. Berlin: Mouton de Gruyter. Hanks, William. 1990. Referential practice: Language and lived space among the Maya. Chicago: The University of Chicago Press. Lehmann, Christian. 1998. Possession in Yucatec Maya. Munich: Lincom Europa. Monforte, Jorge, Lázaro Dzul, and Rodrigo Gutiérrez-Bravo. 2011. Narraciones mayas. Mexico City: Instituto Nacional de Lenguas Indígenas. Skopeteas, Stavros, and Elisabeth Verhoeven. 2005. Postverbal argument order in Yucatec Maya. Sprachtypologie und Universalienforschung (STUF) 58:347-373. Skopeteas, Stavros, and Elisabeth Verhoeven. 2009a. Distinctness effects on VOS order: Evidence from Yucatec Maya. In New perspectives on Mayan linguistics, eds. Heriberto Avelino, Jessica Coon and Elisabeth Norcliffe, 157-174. Cambridge, Mass.: MIT Working Papers in Linguistics. Skopeteas, Stavros, and Elisabeth Verhoeven. 2009b. The interaction between topicalization and structural constraints: Evidence from Yucatec Maya. The Linguistic Review 26:239-259. Verhoeven, Elisabeth. 2007. Experiential constructions in Yucatec Maya. Amsterdam/ Philadelphia: John Benjamins. Zubizarreta, María Luisa. 1998. Prosody, focus and word order. Cambridge, Mass.: MIT Press.