Rerum Dict: The Latin Second Supine Constuctioe 199 struction, which is also known from several other languages, has received much attention during the last wo decades, especially in transformational studies. A famous example of the English TC is (1): (1), John is easy to please. ‘This sentence is a marked expression for the more neutral ‘itis easy to please Jokn’. Example (1) may be contrasted vith example (2) which does not have an impersonal pendant: (2) Jol is eager to please (Pt is eager to please John). For the Latin counterpart of the TC compare (3): (3) virus dificlisinventu est (Seneca, NQ 3,30,8) (ct. difficile est virutem invenire). Moreorer,by answering the above question I hope to shed more light on this rather obscure phenomenon in Latin and to gain more Insight into comparable constructions in other languages. “The paper is organized along the following lines: I will suces sively del with semantic restrictions on the use of the SSC (§ 2), ‘vith restrictions oa the syntactic level (§ 3, including an excursus on the structural description of $SC-phenomena in general in § 3.2) and finally with some pragmatic and syste characteristics (§ 4). Lack of space prevents: detailed discussion of the problematic distinction between the SSC and related construcions with an adjectre plus ablative, like the ablative of respect. As regards this subject I will confine myself to tome occasional remarks throughout the text “The statements in this article on the SSC are mainly based on a sample ofall the instances of an SSC ia the works of Plautas, Ter ‘ence, Cicero, Livy, Versl, Play the Eléer, Tacitus and Stativs. The Sample consists of 311 cases 2, Semantic features In this paragraph I ill focus on the semantic features of the con- stitwent parts of the SSC, viz. the adjecve, the supine and the Sub- ject. By ‘adjective’ I mean the element in the SSC corresponding to dificil: in (3), by “supine che element corresponding to invent and invented by Postal (1974) Examples (0) and (2) were used as 1 paradigmatic stration of the necesity of transformational rules. The information in ths article on the English Tough Construction and the erm Tough Construction ave {Gken from a wecent study by van Oosten (198),200 (Caroline Kroon by “Subject che element corresponding to virts, For the sake of convenience I sIso use this terminelogy for SSC’s on the word group level. Thus, in (4) (4) ciuedem anni tem dict parvam prasteticer, ni... (Liv. 9,30,5). [call rem the Subject. First, I will consider the adjective. 2.1 Semantic features of the adjective?) “The adjectives found in the English TC largely covrespond with those which govern a second supine in Latin. Most of them could be labeled rect orientd or ealu-orinted'. Examples of success oriented adjecives are: facil, dificil, dum, asper, gravis, arduas, levi te: (5) et ine amnis Artatus nomine, hinc mons praealus et adite dif- Felis cingebat (Liv. 43,19,8) So-called value-oriented adjectives are, for instance: (in)dignus, fas et, (in)honestus, magnus, optinus, parons; acerbus,foe- dus, gratus incundus, mirabilis, minus, twpis tc (©) neque admittunt [Iibelli mei] excessus aut orationes sermonesve faut casus mirabiles vel eventus varios, ncwnda dict aut legenti- bus blanda (Plin. NH praef. 12). ‘These two groups of adjectives have in common that they are speaker-oriented. This term is used by van Oosten (1986: 111) to indicave that the adjectives involve a judgement of the speaker with regard to the success or valu of a state of affairs, or that they refer {0 the speaker's emotional reaction to it (ef inthis connection, the °) For the sake of convenience I corsider the substanives pas (meer and ‘eeu, whicn occur in predicstive SSCs, to be adjecives since in several respects {hee Sohaviwr it move ov lrs adjectival. stances Ue Cato RR 3,5 cubis sere (cf alo Dit Cas 71%, Men. 2777188; Sat. ach. 1,119) 1do ot consider SSC' but separative ablatives without peposiion. The ony exzmple ofa spine depending om a verbal predicate is Tat Agr 32: ade dice “CF. van Oosten (1986: 111), who, howerer, makes 2 threefold distinction succes-oriented, talue-orested and. pychoogicaly-orented adjectives This Tart group, which contins adjectives like acer incundie et, consider sb- category ofthe ralue-riened adjectives (dipnay, hones ete). Compare aso KuhnerStepmann’s categorisation of adjectives (1912, :724), which will prove to be inaceerate‘Raram Dist: The Lata Second Supine Construction 201 disjunct function of the adverb). The involvement of the speaker in the state of affairs of the adjectival predicate is proved by the fact, that the SSC may be extended by a dative satellite refesing to the speaker. CF. (7): ©) 2) mihi diffcde dia est... b) Sed ex cuncta Romani: ex unebris et editiorbus leis facia visa «eran (Sal. fag. 98,7). -om (7b) it appears that ‘spesker-oriented isa somewhat mislead- i term. What is actually meant by speaker is the person to whom the information conveyed by the SSC applics. Interested Party’ ‘would perhaps be a more appropriate term. ‘Another characteristic of the speaker-oriented adjectives is that they need a verbal specification. Without this specification the adjec- tive makes no sense or the meaning of the utterance changes. For ‘example, inthe sentence ‘John is easy to please’, omission of the ele- meat ‘to please’ leads to a different meaning. ‘Thirdly, speaker-oriented adjectives can he used in an impersonal construction, cf: (8) quoi verba dae dificil et (Ter. An 241) {9) thc homini dignum est divitas esse (Pl, Mil. 729. ‘This rule is not coavertble, ie. not all adjectives that occur in an impersonal construction also occur as speaker-oriented adjectives in an SSC, as appears from the ungrammaticality of (I0b): (19) a) iustum est ut tos tbi servos tuo arbitratw serviat (Pht Bacch 994) ») *princeps iustus necatu est. Speaker-oriented adjectives contrast with adjectives that convey an inkerent property of the ‘Subject’ argument of the adjectival predi- 4) Ry Disunet I mean acenatituent tha! exprsies the opinion ofthe Speaker! “Witer wth respect to the content or expression of a whole prediation. For the notions Adjune: and Disiunet ef. Pinkster (1972: 96-101; 1988: 47-55), who ‘refer to Greenbaum (1963). *) Leca satelite with the function ‘Iaterestd Party’ on the level ofthe matrix predcation and not Agent ofthe embeded supine predicate This is proved for English by van Oosten (1986: 108), who poits to che ule that adjective and for. prepositions phrase can ot be separated: Fora year-old to bake tis cae would beeay (=> Fora five-year this eake would be est bake)202 Caroline Kroon cate, Examples of such property-denoting adjectives’) are ‘tall, “groen, ‘prety, ‘tof’ otc. Compare (11) (11) Lateranus ... animivalidus @ corpore ingens (Tac. Ann 15,5343). Verbal specification of these adjectives is optional or sometimes even excluded. cf (12) she is pretty (€ look 3) (13) "this Banana is green to look at Extension of property-dencting adjectives by means of a dative sat ellte is not possible, in view of the fact chat the property that is ‘expressed by the adjective does not vary with the individual experi encers. Compare (14), which isil-formed (14) this banans is green for me Finally, property-dencting adjectives as a rule do not appear in impersonal constructions. Depending on the context, however, some adjectives may be speaker-oriented as well as property-denotirg, for example pulcher in (18a) and (15b), respectively: (13) a) cui pudchowm fuit in medios dormire dies (Hor. Ep. 1,2,30) by puer pucker st The same holds «.g. for parous and magnus) (cf. (4). ‘This distinction between speaker-oriented adjectives and property- denoting adjectives is the main criterion to distinguish between an SSC and an ablative of respect. For, in general, property-denoting adjectives can be specified by an ablative of respec, speaksr-oriented adjectives by an SSC. Two apparent exceptions to the rule that the adjective in an SSC is speakeroriented are given under (16) and (17): (18) (about some sort of fish) nullo nunc in honore est, quod equi- ddem miror, cum sit rans inventu (Pn. NH 9,60) (17) (about unjustified accusations) Quam levia genere ipso! quan falsa re! quam brevia reponsu! (Cie. Cluent. 164). ‘The adjective in (16) isnot speaker-oriented, but, in any cas, it does 9 For Property-denoting adjectives van Oosten (1886: 111) uses the rather confusing term ‘Subject onentel ') Cf also Risselada (1984 211) whe observes thatthe adctive mara may be covordinated with (instead of jurtaposed to) one cart of adjeatvet when it dleaotes siz’ and wih another when t denotes Subjective evaluton™arum Diets: The Latin Second Supine Construction 203 not refer to an inherent property either. We might say that, when used in an SSC, the adjective rarus has the function Adjunct, rather than, as is usual, Disjunct. A borderine case is (17). On account of its own semantic features brevi is not a speaker-oviented adjective. However, inthis particular context brevie seems to be more oF less synonymous with facil. ‘Anyhow, the conclusion ofthis section is that the group of adjec- tives which ean occur in an SSC oF TC is limited. 22 Semantic features of the supine predicate?) Not only the adjective, but alo the supine predicate of the SSC is subjea to trong selection estictions ‘The fist selection restriction on supine verbs concerns their v2- leney, for all of them are two-place verbs that usvlly govern an Agentin the nominative and a Patient in the accusative. Inthe SSC, this Agent (which is mostly generic or provided by the context) is sever expressed, but ay be coreferential with a free dative constte ent with the adjectival predicate, ef (7): (7b) Sed ex cureta Romanis ex tenebrs et editiorbus locis facia visu. erant Galle jag. 98,7). "The constituent that is semantically the Patient of the supine predi cate (ca cuncta in (78)) for some (pragmatic) reason") syntactically functions asthe Subject ofthe adjectival predicate. As far as Tknow, ‘0 two-place verbs which govern the genitive, the dative or the abla tive rather than the accusative case form appear in the supine con- struction. For example, I did not come acrors = supine petit, nor an ‘unambiguous supine usw"). The few instances of ssw mentioned by 1) Te the Iterature itis agreed thatthe second supine is a verbal now. Some interesting remarks on verbal souns ending in -a! as contrasted to verbal nouns in stb mayrbe found in Marouzeau (1949: 47-86) and Benveniste (1948: the latter state thatthe verbal nouns ending ia “as onginaly had 2 _sbjecive those erding in ~i an objective value. This observation i in accor sce withthe face thatthe SSC (which consists of a verbal noun of the four decleion) ba a mainly evaluative, subjective character (se also 41). A more recent study on Latin verbal nouns is Rosin (191). Ch E32I A 38 9) One obvious expan Aeponentia and therefore that both sand posi are derived from 3 occur inan SSC (deponenta cannot be passive204 Caroline Kroon Sidstrand as supines are, rather, varisnts of in wsw and mean ‘in prac- tice’ oF ‘by practical experience. In English the restrictions on the use of a particular verb in the “TC seem to be less strong, as appeats, among other things, from the fact that also three-place verbs can participate in the TC and that also arguments other than Patients can function as Subject. An ‘example of a Recipient in Subject position is (18): (18) Virginia is hard to give directions to"), ‘One-place verbs are excluded from both the TC and the SSC. The instances of nat ard viet (verbsl noun derived from the verb tivere) listed by Richter and Sjostrand are, in fact, ablatives of respec, since they are never combined with a speaker-oriented adjective”) [Not only with regard to valency but also in other respects are the selection restrictions on the Latin supine predicate more stringent than those of its English pendant. Whereas in English all kinds of verbs seem to be used, provided that they are two- or three-place predicates, in Latin there are only eight verbs of which the supine ‘occurs with some frequency (19) aire, andire, dicere, facere, intellegere, memorare, cognoscere, widere A common characteristic of these verbs is their nos-specificity or, in ‘ther words: they all are the relatively neutral representatives of ight classes of Trequently used verts, namely ‘to go"), "to hear, to say’, to do, ‘to understand’, ‘to relate’, to recognize” and 'o see’ actu, for example, is the non-specific representative ofthe semantic group of action-verbs. More specific variants like capt, necatm, sinacts, etc. are hardly ever found. ined) isto be rejected om account of ateted supine ike pees e.g Cie. Tae 220; 9 mots dies gravia, perpen aspers. ") Notice that ia the TC, ix consrast with pusive constructions the element is obligatory 9) "Tn the carious example sspene pun cord ‘their hears beat heavy’ (Sit. Theb. 4,726) pula i semantically related to puliae rather thas to two-plice Jello pulse wocld hare ben derived from pelo the meaning of aipera palin Corda would be the nonsensical hears that are hard to beat. tis probably because aie is syntaceally modifier of cords, but semanialy behaves asa snverb that Sjostand considered ita supine “) Notice chat adie may be considered the rworplace counterpart of one place ireRarum Diets: The Latin Second Supine Construcion 205 1 must be noted that these selection restrictions are statistical rather than absolute. In principle all rwo-place verbs governing an accusative Patient may occur in the SSC, as appears e.g. from (20): (20) (the plant casia) gignitur in planis quidem, sed densissimis in vepribus rubisque, dificils collectu (Plin. NE 12,89). In realy, however, Latin authors hardly use these more opecifie verbs inthe supine construction. A tentative explanation of this tend: cency is given in §4 Another semantic characteristic of she supine-predicate is the potential rather than actual/factive value of the state of affairs it refers to!) The potential value of the SSC is especially clear in ‘expressions ike facilis dtu quam re. An example is (21): (21) id dictu quam re, ut plerague, facilus erat (Liv. 31,38,4)"9, Here re seems to be preferred to e.g. the supine factu to emphasize that reference is made to an actual fact. Other examples of an explicit opposition between ‘potentiality’ and ‘actuality’ are (22) and (2b): (22) a) rarum dietu, sed aliquando viswn (Plin. NH. 14,132) 'b) Nec proinde diiudicari potest quid optimum factw fuer, ‘quam pessimum fuisse quod Jactum est (Tac. Hist, 2,39). In these examples a supine is contrasted with a perfect participle. In this potential value the SSC is comparable to the gerund and adjec tives ending in ils, like amebils and invinciil, which hare an inherent potenti value. It is, therefore, unlikely that this kind of adjectives will accurin an SSC. since this would produce a combina- tion of two verbal notions with a potential value”). Nevertheless, ‘Sjostrand registers some examples of gerunds and adjectives of the amabili-type in an SSC, e.g. miserbilis and miserandus. The poten- tial value ofthese adjectives, however, jst ike that of homrbil tr- ribils ete, seems to be weakened. Other instances, for example verdibiles in (23), I do nct consider SSCs" %) Cf Benveniste (1944: 102-1) and Blue! (1979: 1). 4 CE also Liv 10,38,13 and Cie. Paradox. 8,98. Cf Seanty (965: 382-3) forthe opposte view Other istanes of adjectives ending in Jism Sjostrand’ material resut fom a nisinterpreation ofthe senence, eg. relene itr amas murda dxi- mua it pores anmerbile deta mela et sera? "or would He Be MOFE Conect speak of heavens as any or infinite in momber? (Cie. Tim. 12); CL206 Caroline Kroon (23) (about some sort of grapes) et quae forenses vocantur, vendibiles aspect, ... (Pl. NH 14,42). ‘Tomy mind, aspectu isa causal abla (aspectu) they are salable (vendibiles: Finally, from the English data") it appesrs that the state of affairs to which the infinitive fers is permanent. This can be proved by the fact that not all types of causal satelites can be added to the ini tive, but only satellites that have a permarent validity, cf: (24) a) he is impossible to talk to, because he is always out of town / as stubborn as a mule by +he is impossible to talk to, hecaute he is out of town Gf. i is impessible to talk to him, because he is out of town) 1: because of their appearance Te is to be expected that the state of affairs to which the Latin SSC refers is also permanent. In any case, I did not find a counterexan- ple 23 Semantic features ofthe Subject In considering the semantic features of the Subject argument of the SSC we have to realize that there are two types of SSC, The first type of SSC functions as an attribute on the word group level or as predicate to an NP-Subject. This type is usually called ‘personal supine construction’ (examples (3) and (4)). The ather type is the so-called impersonal construction of adjective plus copula. About 30% of the sample of 311 cases consist of such impersonal construe- tions. An example of s so-called impersonal SSC is (25): (25) difficile dicts est quanto in odio simas (Cic. de Lege Men. 63) (TC: how much we are hated is not easy to sy’) In this example, the indirect question could be considered the Sub- ject of the SSC, as is clear from the personal construction in (26): (26) id, quod tuepssimnum dicta es... (Cie. de Leg. 1,52) I do net, therefore, differentiate between personal and impersonal ‘SSC's, but consider them to be essentially the same: the former cate- {gory has a nominal, the latter a sentential Subject. Plato, Tin 31A; in this sentence dct depends on melas and aot as Sostened suggests on immwmeabils *) Cr. van Oosten (1986: 119,208 Caroline Kroon From this examination of the semantic features of the Subject constituent we may conclude that the Latin SSC mainly predicates something of a sate of affairs and not of concrete enies. In tht respect itis more limited than the English Tough Construction, or stated more precisely: the Latin SSC is less fully developed than its English pendant since it observes the selection restrictions on the Subject For in principle, the selection restrictions put on the Subject by adjectives like diffs, facilis ee. are such as to exclude concrete entities. In the English TC, on the other hand, these selection restic- ‘tions seem to be less stringent. 2.4. Summary of he semantic characteristics of the SSC In summary, there are a number of selection restrictions on the constiwent parts of the SSC (adjective, supine and Subjec). The dis- tinctive semantic feature of the adjective ists speaker-ovientednes. Speaker-oriented adjectives have the following characteristics: they may be extended by a dative satellite referring to the speaker: they need a verbal specification; they can be used in impersonal construc- tions. Property-denoting adjectives (the type of adjective that is con stitutive for the ablative of respect construction) lack these charae- teristics. Secondly, supine predicates are restricted to 1 group of non-specific bivalent verbs that govern a Patient in the accusative. ‘The state of affairs they refer to may rather be described as potential than a3 actual. Lastly, the Subject of the SSC tends to be noa-cot- crete and inanimate: embedded sentences and neuter pronouns pre- val in Subject postion. In this respect too the SSC differs from the ablative of respect construction, With regard to the supine and the Subject the Latin SSC is far more restricted than the English Tough Construction. 2: Syntactic characteristics 3.1 Syntactic restrictions An investigation of most SSC’ in Latin makes clear that the semantic restrictions recounted in chapter 2 coincide with restric tions on the syntictic evel. The SSC is syntactically restrcted in two respects. First, the internal structuee of the SSC is nen-complex. Sec- ‘ondly ithas a highly independent status with regard to surrounding predicationsaww Dict: The Latin Second Supine Construaion 209 “The first restriction, the non-complexity of the syntacti structure, is mainly a consequence of the inability of the supine to govern arguments). Moreover, extension of the supine by a satellite is rare. In ay sample I have found only one instance: (28) ex his {urbibus] digns memoratu aut Latino sermone dicts (Pin, NH 3,7) Extension of the adjectival predicate (i.e. of the SSC as a whole) by means of satellites is alto very rare). The only eegulary used type of extension of the (constituent parts of) the SSC is modification of the adjectival predicate by means of particles like etiam, quoque, vix, etc. “The other ype of syntactic resuictedness of the SSC concerns the fact that as a whole it is syntactically highly independent. This appears from the folloving observation (i) Ithardly ever occurs in embedded clauses, and when it dots 50, only in a first-degree embedding"). i) The attibutive SSC’s do not combine with nouns in all case forms: the adjective and the Subject argument only appear in the nominative and, athough less frequently, in the accusative. (29) is an example of an SSC in the accusative: (29) eivsdem anni rem dicta parvam practerirem, ni... (Liv 9,39,5). In the sample there are only three exceptions to the rule that the SSC ‘occurs in the nominative or accusative”). These exceptions have in ‘common thatthe SSC’sall function as a restrictive attribute, which is i Gide amie. 12 quo de genere nots difficile dieu est Ge Fam 17,2 diffi dct ex de snguis = Gell 9,7, Sed de ius rarivs dict et mrabilos est [Addition of dhe PP's inthe Cicero examples nap probably posible because diff ‘ile dieu was by that time fl tobe an idiomatic expression. The SSC in Glas ‘coma to Lean litical contamination of two constrctons (i, od de fidivs dtu, arias et mizabils est i), de fdibusraro dict et qaod diturmirablis et "These are, for instance, + few example of extension bya casa slice sei diffi ada protr harevous slime (Phin. NH 6,46) Insramen: tal satelites ("homo felis hac hata ednerats) are nok aested. 4) These are manly the evakative quod: dauses. See below. | Dhin NH 29,61 (genive); Pin. NAT 10,81 (lave absolute); Liv. 21,13 (date)210 Caroline Kroon 4 deviant and relatively marginal category of SSC’s, as Iwill demon- strate below ( 4). (ii) A considerable part of the material consists of supine con- structions with a parenthetical status i.e a sentence within another sentence but without any syntactic relation with the matrix sentence. ‘The sample of 311 cates contains 48 real parentheses, inchiding 4 exclamations; compare (30) miramgre dic do hostiom agmina ...in sva ruebant (Tac Ann 2,17) Besides parentheses in a strict sense there area lot of constructions in the material that are related to the parenthesis in that they also are to some extent syatactcally and prapmatically independent. These ‘parenthetical clauses’ have the function Disjunet with regard to the surrounding prediction. Tey se listed in table 2 Table 2: parenthetical clauses parenthesis (incl. exclamations) 48 evaluative quad-clause B pseudo-concessive clause 2 pseudo-conditional clause 3 apposition 4 80 [An example of an evaluative quod-clase is (31: (1) is... auctoritatem Paulini...criminando, quod failimum factu st, pravus et calidus bonos et modest anteibat (Vac. Hist. 18. ‘The sample contains 13 of such evaluative clauses. An example of a pseudo-concessive™) clause is (22): (2) quaseo oroque vas, patres conscript ut primo, i oit wel acer ‘on auitu vel incredible a M. Cicerone esse dicts, sccipiatis sine offensione quod dixero (Cie. Phil. 7,8)"). ) Psewdo-concesve cause is not a very accurte term. What i actually meant by his term is that, due to i spec postion ia the sentence, the conces sive clause moze or ls behaves as a parenthesis: ther is seeming lack of inte- ‘ration of the eoncemive dause witha the mattis asse 19) "This sentence is par ofa very extensive ‘real’ parenthesisRarun Dict: The Lata Second Supine Constuction 211 Examples of a pseudo-coné jonal cause and an apposition are (33) and (34), respectively 83) Humanus autem animus ... cam sli nullo xi hoe far et dicts, comparan potest (Cie. Tusc. 5,38) (34) Practerea dloacas, opus onium dics maximum, ... (Plin. NEF 36,108). (On account ofits independent status in relation to the ‘antecedent’, the non restrictive apposition may be regarded as a parenthesis on word group level: there is no syntactic Head-Attribate relation between ‘antecedent’ and apposition, as is clear from the fact thatthe ‘antecedent’ may be left oat and replaced by the apposition without the sentence becoming ungranmatical. In this respect appostions differ e.g. from relative causes ‘As appears from table 2, about a quarter of the sample of 311 cases has the syntactical status of 2 parenthesis or a parenthetical clause, When we leave out of consideration the 137 attibutive con- structions, almost 50% of the SSC’s turn out to occur in parentheti- cal clauses. On the other hand, most of these attrbutive SSC's abo are relatively independent as regards their syntactic and pragmatic characteristics. This is especially clear from the fact that they prefer- ably occur in a series of non restrictive attributes without relative proncuns. An example is (35) (95) Lathyris ... habet geemira molta, quac cum inarucre,eximun- tur grana piperis magnitudine, candida, dulia, fail purgatu (Plin, NE 27,95). To sum up the syntactic features of the construction we observed thus far: the SSC has a nen-complex internal structure. Furthermore, it as a high degree of syntacic independence with regard to sur” rounding predicstions. This independence appears from the con- straints on embedding and case-marking and from the fact that the S5C often occurs in a parenthetical clause. The internal syntax of the English Tough Construction caa be more complex. it also can be indefinitely far embedded in s matrix clause. Even within a Tough Construction embedding is possible, cf. (36): (36) the question is unreasonable for you to expect John to be able In this example the Subject has its co-referent in a third level of embedding.an Caroline Kroon 32 Description of the syntactic structure of the SSC In 3.1 we examined the restricted syntactic behaviour of the SSC, with special atention for its relatedness to surrounding predications, ‘This section contains some remarks on how the phenomenon should be accounted for in a formal description of its interaal steucture™), in view ofthe fact that it isnot absolutely clear how the constcuent parts of the SSC are tobe defined in syntactic terms such as Subject, Object and Complement. These problems inthe description are con: nected with the fact that the SSC shows a discrepancy between the semantic and the syntactic structure: the element that is semantically the Goal) of the supine predicate syntactically behaves as a Subject with respect to the governing predicate (the adjective). In the sections 3.2.1-3.23 below three types of detcrition that are proposed for constructions like the Latin SSC will be discussed: raising (3.2), the supine a3 a Purpose expression (3.2.2) and the SSC as complex adjectival predicate (3.2). In these sections [will raise arguments for and against each type of description, without hhaving the intention to be complete or to say the final words sbout it. The cemarks will toa cercain extent also apply to comparable phe- rnomens in other languages than Lat 3.24 Raising In a ‘ruising’-analysi™) (ardently adhered to by transformatios lists) the SSC is considered to be a complex predication, ie. a pre cation which contains ancther predication embedded within this matrix predication. Or, in the terminology of Functional Grammar: fone or more term positions (= argument or satellite positions) of the predicate (h.L the adjective) ace specified by a predication (the supine) In the SSC the ‘Object’ constituent (Geal) of the embedded predicate has been promoted to Subject of the main predicate, « phe nomenon which could be called Object-to-Subject raising”). Sche- matically the raising operation may be shown as follows: ") Inthe descripon I will use the rerminology of Funcioral Grammar (Dik 1978) ") In Functional Grammar the tern Goa is used intead of Paint %) This anilysis for SSC-phenomena i, ameng others, proposed by Dik (85:85), ) Probably one could also speak of Subject-te-Subject raking: itis ne clear ‘whether in the embedded sentece the syntactic function Subject oc Object hat ‘een antigned tothe displiced constiventRarum Diet: The Latin Second Supise Construction 213 Figure 1: raising-analysis applied to the sentence Diffcie est hoc genus exornationis inventu (Rhet. Her. 4,39). Predication Predicate “Argument sema difficile invenire genus exornationis syntactic difficile genvs'exorn, inventu a (Source: Pinkster, 1988: 204), Within the theory of Functional Grammar the raising analysis) implies that the SSC has the same underlying predication ss the ‘impersonal construction, viz. (37) 97) difficile, (invenre(0)gg (genus exornaionit)gay Subscript A means: adjectival precicate. This predicate has one argu tment all information between the outer brackets serves as Subject of the adjectival predicate difficile and has the semantic function Zero (in G7) marked by subseript 0). Fr some (perhaps pragmatic) ceason, which still requires a thor: ‘ough investigation, in the SSC the syntactic function Subject is assigned to a lower-level Gosl and not the entire enbedded predi- cation”). This differenes in Subject assignment accounts forthe dif ferences in capresiion between the SSC end the impertonal con. struction, ‘Objections agaisst the raising-analysis are raised by van Oosten (196: 113) and Vee (1981: 156-7). The later demonstrates thatthe precication in (38a) involves the reference to a state of affairs, ‘whereas in (986) it concerns the entity in Subject position (~ le Concierge). This is why (388) entals (38a), but not the other way around. Orin the terminology of FG: Pseudo argument formation (= displace iment + moephelogiladjestmest. "Probably the sane pragmatic esto as entail the comparable operaion of passivation; in FG pasiviation is alo devribed as Sebjectassigement to the Gos! la this ease, hovever, the Goal is part ofthe matrix predication and not of tan embedded preicaon. In view of oor dita it eems to be the case that the Subitct ofthe SSC ha the status of Tepe (oles oft, Focss) in the sentence a5 a hole wheres nthe impersonal eostracton the whole embeded predicae tion, and sot one clement efi, bas Tepe or Focus stanae Caroline Kroon (38) 4) illest difficile de corrompre la concierge ) Ia concierge est cifficle & corrompre This observation slso applies tw SSC-phenomena which are governed by speaker-oriented adjectives other than diffcilis (Engl. ‘Affcalt, Fe, dificile resp). Therefore, as the two constructions are semantically not equivalent), they can not be derived from the same underlying predication. ‘A further problem of the raising-analysis and the underlying pred. ication (37) is, according to Vet, that in the SSC-phenomena the Agent must always remain unspecified, in contrast with other imper- sonal constructions which have ‘raised’ equivalents"). Compare examples such as: (29) il semble que Jeanne embrasse son mar 3.2.2 Purpose expressions Another possibility forthe descrition of SSC-phenomena may be derived from a proposal by Dik (1985) for constructions ofthe fol- lowing type, which show some similarities with the SSC): (40) 2) you are too young to participate b) you are old enough tc participate In these examples the to-nfinitive resembles the expression of Pur- pose satellites. They also seem to be semantically related to Purpose satellite as appears from the following paraphrase of (40a) (41) ‘you are younger than needed for a particular purpose, viz. to participate. Dik therefore assumes the following underlying predication for (40a: (42) too young, (you), (0 participate) rey This seman diflerenceis proved, among ther things, by the causal stl- lie test. Ch exarple (24). >) FG avoids doping elements (hl. an Agent argument) inthe urdertyng prediction when they do not occur in the acta expression Dis remarks coacern Dutch sentences wth the paticle om The sume Ios, however, for English and other languagesaru Dict: The Latin Second Supine Constrection 2158 In this analysis 1 participate is the second argument of = two-place (adjectival) predicate (t00 young],” Application of this description to SSC-phenomena would lead to the following analysis of the sentence ‘John is easy to please’ (43) easyy oh, (6 plea) ngewe “This analysis, however, yields difficulties for the SSC-phenomena in general and for the Latin SSC in particular) First, om the semantic level the supine (or infinitive) can hardly be considered Purpose expression asappearsfromthe paraphrasein(44): (44) The is easy for 4 particular purpose, viz. to please’ Purpose satellites require 2 controllable state of affairs. Isis obvious that the adjectives that occur inthe SSC-phenomens do not refer to such controllable states of affsirs. According to this condition of controllability, examples like (40) cannot have the semantic function Purpose either. These constructions, however, at least have an ob- ligatory constituent with the semantic function Result, which is related to the function Purpose. Such a Condition-Result relation ‘an not be discerned in the SSC-phenomena.Itis also very question- able whether the supine (or infinitive resp.) should be considered an argument at all®), Figure 2 (p.216) shows the difference between 85C’s and ‘Purpose’ expressions. For the sake of completeness the ablative of respect is added. In figure 2 nr.1, a sate of affairs referred to by predicate Y is con- ditioned by a predicate X In the ease of an SSC there is no such relation between two independent predicates (ie. two separate events) X and Y: id dificile div est can not be paraphrased if difi- tile et et (consequently) dicitw. Figure 2 also shows the similarity between the SSC and the ablative of respect: the property o is irrele- vant or not known. Therefore, it cannot be expressed. AS a conse- quence the SSC and the ablative of respect have the same syntactic representation, 2”) x, (4) i in fact, a derived predicate-frme which resuited from the absorption ofa Parpose satelite ino an exiginallyone-placepredcate-frame.T wal not go ino thie mater of derived predates, 8) CE Dik (1985: 41-8), who dicuses the difficuies for Dutch 2») The fact that te supine ean not be omited without causing an ill formed sentence (eee § 21) need not be an indication for argument sats: this seems ‘thera result ofthe fai that Speaker-onented sdjectres have thefuncton DS FncU/Adjunc and therelore preferably oceor together mith some verbal xeme16 Caroline Kroon Figure 2: Schematical representation of the semantic situato 1, Purpose expr. (e.g. ad + ger): Xis the case > Y 2. 88C + [wis the case] ~ XY 3. Ablativ of respect XY is the eae Explanation of symbols: X= some adjectival predicate Y = some verbal (or substantval) predicate 2 = some non-expressed property-denoting adjective X + Yee. = X conditions ¥ ‘The second objection that cn be raised against « description of the SSC as a Purpose expression is based on evidence from the Latin data, In Latin, in contrast, for example, English, the semantic dif ference between the SSC and ‘Purpose’ expressions corresponds to a difference in expression: Purpose expressions and Condition-Result relations appear to be almost exclusively expressed by ad + ger- und). To conclude: the description of the SSC in terms of a two-place predicate requiring a second argument with the semantic function Purpose mects with a number of dificules. It is time to consider « third type of description. 3.2.3 Complex adiectival predicates ‘As appears from figure 2 it might probably be mors fruitful not to. describe the SSC as consisting of a main predication and an ‘embedded predication, but as one complex adjectival predicate requiring only one ergument*) (45) diffcilis, nventuy(sirtus) ‘The following observations may be adduced in favour of this description: +) Although itis generally aged that ad ++ gerund isan alternative con- struction for the SSC teompare Kuner~Stegmann I: 726.0, there are several indications against the supposed exchangesbility e.g. with respect othe type of Adjective on which both constructions depend. ‘8 Sucha description is also suggested by Vee (1981) and Dik (1985) in the articles on French and Dutch tes.218 Caroline Kroon ition of compound adjectives in these languages is very productive). ‘Compare the SSC in (47), which can also be described as a substitute for some non-existing ‘composite’ adjectival predicate: (47) se difficile fate est a deis ons fidem aon habere (Cic. Tim. 11,38) = (Plato Tim. 49D: déivarov obv Geddy nasoiv én. aveiv), Classical Latin has 10 words corresponding to the English possible/ impossible, In (47) the SSC is an excellent means to cope vith this problem, considering that the SSC has an iherent potential value (see §2.2) ‘There is also some evidence that the SSC can function as a com- pound or compound-like adjective, namely in cases where in Latin both an SSC and a compound adjective with roughly the same mean- ing ate avsilable. For example, Livy uses dignus memoratn, hand par. ‘ous dictu et. in exactly the same way as he uses the compound-like Adjective memorabils. Examples (48) through (51) show some of ‘many remarkable instances: (48) Consules L. Sergius Fidenas iterum Hosts Lucretius Tricipiti- nus. sihil dignem dctw actum his consulibus (Liv. 4,36,4). (49) Proximo anno Num, Fabio Vilano T. Quinctio Capitolini Capitolino consulibus ductu Fabii... nihil digmum memoratu sctum (Lir. 4,431). (60) In Gallia nihil sane memorable ab Sex. Actio consule gestum (Liv. 32,25 1). (51)... tribunos miitum consulari potestate creavere,....Q. Servi= lium Fidenater iterum Q. Sulpicium Camecinum iterum. His, tribunie ad Veioe nihil edmodum memorable actum est (Liv 5,146). These four examples are identical in several respects: context; sub- ject-matter, introduction to a new paragraph. It is not quite dear ‘what factor determines the choice between an SSC or a ‘compound’ ©) CE Zebian (1968: 257,12), Leumann (1986: 150-2) and Seanty (1965: 384-5 Gegentatr um Griechichen st im Latein die Fihigeit nur Komp: ‘ntvickel’. The Romane themselves seemed to be aware of ‘hie sation, at appear from Quint 15,7: (ahowt compote ce) "Sed res ‘ota magis Graecos deer, nobis miaussucedit nee i fien natura puto, sed ale tis favems idecque exm atoeaewe miati sims. incrvicercicm vx 2 sa detendimus.Rarum Dictu The Latin Second Supine Construction 219 adjective) In other examples the choice seems to be determined by ragmatic/stylistc factors, see for example (52) and (53) (52)... gnarus facilen’ macaow gentem, wt segnem ad pericula, its infidam ad occasiones (Tac. Ana. 14,23). (53) Sabinus suopte ingenio mitis, ubi formido incesisst, facilis mu tain etn alieno diserimine sibi pavens, ne ... (Tac. Hist. 2,63) In these examples, ficils matatw has more or less the same meaning ‘as mutablis®). Both SSC’s, however, appear in a context which is composed of a series of adiectives plus extension. The SSC is apt to ‘maintain this parallelism. Word length seems to be another stylistic factor that may determine the choice for an SSC instead of a simple lexeme. As, among others, Szantyr') observes, Latin words seldom surpass the limit of three of four syllables. To avoid longer words authors may use descriptive phrases, like for example: (54) et si sequemur ea quie de brevitate praccepta sunt, nam quo brevior, dilusidior et cognityfucior narratio (Rhet. Her. 1,15). In this example some properties of the sentence topic nerratio are ‘enumerated by means of adjectives. Because of its length the alterna- tive conprehensibilior seems to be a non-preferred form. Instead of this form a descriptive phrase has been chosen that does not disturb the parallelism in the expression of the several predicates, viz. the ssc. (ix) The SSC and the Indo-European double dative ‘The description of the Latin SSC as a complex adjectival predicate seems to be in accordance with the hypothesis shat the Latin second construction is related to the Indo-European double “This hypothesis requires some explanatory remarks. In the Indo-European double dative construction two types are represented: “) Particles like se and admodn can alo cecur with SSC's end therefore sre not deci o) matail is abo arsed in Tacs. 1) 1965: 756 4) Alin Carstol (pc) see also Blume! (1978: 108). Fora eral survey of the lteture an the diashroni desslopment and interpretation of the Indo Etropean dovble dive constriction see Heth (1984), Hetrch himself expel os tothe aisumad Indo-European bass ofthe double dative comet. tions in sever LE languages. He raher ties te provide evidence for parallel je prcester from syrtatclly independent dative constitucts it alized systagins of interdependent connttwests20 (Carolive Kroon (i) the substantive inthe dative is Agent of the verbal noun. Com- pare Vedic) Indra pltave “for Indra te drink’, for Indra to be drunk, (ii) the substantive in the dative is Patient of the verbal noun. Compare Vedic Vrniya hénueve ‘Yor Vetea to kill, for Vitra to be iilled: ‘According to the hypothesis, the Latin SSC represents the Fest ‘ype of double dative construction: in (7b), ¢.g. Romanis corre sponds to the ‘Agent’ dative, facilia vi w the predicative dative. (7b) ea cuncta Romanis... facili visa erant (Sal. fug. 98,7) ‘The relation between the SSC and the Indo-European double dative ‘ean be mide more clear by comparing the SSC to the ‘regu’ double dative construction in Latin in which both types of the Indo- European construction are represented®"): (65) hace mibi curse est (66) mihi amictui ese Scythicum tegimen (Cie. Tise. 5,99). First, one of the constituents in the dative is a verbal noun’), which can never occur in the plural. Secondly, the verbal noun can be modi- fied by an adjective (but, in concrast to the adjectives of the SSC, ‘only those that indicate gradation). Adverbs are much less com- mon?!). Examples (37)-(58) illustrate the relative similarity between the SSC and the double dative) 9%) The examples from Vedic ae fom J, Hovey (197: 111-2) 5) Another texm for this consvucton is prediative dative, There ae some remarkable instances this prediative dative ending in in Plautus, eg, Rad. 294 hse hami aque hace harundines sant nobis quae et calla 10) There area few exceptions, eg. aigul miki ext cord. These, however, all are variants of fequertly wed verbal nouns (condi = care) 53) some cates the vert noun is mode by qualfirs such a ius ma. Team aocbeaccemparied by any other substantival modifier or slit than the dative of peson (Rosin, 1°81: 148-50). In this tespea, tow the double dave conseactioneesenbles the SSC. One hypothesis might be that in (5) dificil dic is + Iter devdlopment orginal dif dit (dative, As esl of the fact thatthe miki costa ft uially remained snenrenved when i fenesoned ay Agent of the veal oun, "df dw kad gradualy come tbe fel as predicate ofa copula con- ‘Arsction and wor thus expressed inthe nominative care form. A that age of ‘he development the sepine was no Tonger fl 4 dative but rather asanabla- thn, tinee the corsrution had some more orl identical withthe ablative of respect From this moment an, abo the adjective digna (osualyconstaed withRarum Diets: The Latin Second Supine Constuction 221 (52) i mihi maximo honori est (58) id mii difficile dictu est. In (57) maximo honor functions as a syntactic unity (viz. predicate). Adopting tke hypothesis of the double dative-origin of the SSC, it might - at least diachronically- be justified to consider also difficile dictu in (58) 4 syntactic wai (bul. complex adjectival predicate) For the moment it s not posible to provide a more detailed dis- cussion about the double dative-hypothesis and its temability. Un- doubtedly there are sill some considerable problems attached to it T should like, however, © touch on ore more interesting point, viz. that the hypothesis would shed aew light on the controversy in the literature with regard tothe case form of the second supine. On accourt of the occurrence of instances like PI. Bacch 62. istace lepida sunt memoratui several sttempts have been made to explain these ew supines in vs, e.g -a in these cases is nota dative, but an ahlative ending (-e and -ki are interchangeable) oF inthe second supine several cases have merged (dative, ablative and locatve)), the instances of spines ending in 1d are archaisms. However, from Priscianus) onwards, most grimmarians”) view the second Supine 4 originally an ablative in form and function. Besides all the arguments adduced in this section in favour of describing the SSC as a complex adjectival predicate, a few objec- tions ean, on the other hand, be essed: (i) the fact chat the complex of adjective and supine may be prag- matically split up: seen. 42, where itis stated that the supine can bear (contrstive) focus. Compare also examples lke Liv, 34,58,4 where a particle has only one element of the complex as its scope: enimoero 1d anditu clam dicere indignum ene Hepesianax (Hegesianse ssid that ths [suggestion] was unworthy even to listen to [let alone co execute}. However, t appears from example (59) that in ther lan- uages t00, compourd adjectives need not be a pragmatical unity: (69) deze pen is niet blauw.-, maar awact-schtijvend (Dutch) this pen is not blue, but Dlack-writing, ‘a8 ablative) could acer inthe SSC, and the supine could be co-ordinated with a ‘hetntive ie the slate (compare example (21 in $2.2) %) Choe-g WH Linésny (1695). 9) TK, 9 $48: equens wero form quae in a termina, ablaivs miki vider isis womins gun ine pat, tres ser pacestione. Quid *} Eg J. Golig (186).22 Caroline Koon i) the occurrence of examples like Liv, 21,32,7: visu quam dicts feediora;, compare also Plin. NH 28,1 (Gi) the fact that adjective and supine may be at a great distance of each other in the senterce. Within this context it may be noted that at least the ablative of respect can be descrbed as a complex adjectival predicate: first, because mort of the relatively few compounds in Latin are the equi alent of an ablative of respect (magnaninus <--~ magnus animo); second, because the ablative cannot be considered an independent argument or satellite since it does not have its own complete term structure: the substantive in the ablative cannot be modified by another adjective. Consider the ill-formed sentence: (60) *Britanni cani capells longis sunt. 4. Pragmatic and style characteristics In considering the syntactic characterises I have aeady briefly mentioned tome pragmatic factors that play a pat in the wre of the $8C. In § 41 I will discuss these pragmatic factors more systemati cally. In tis section Iflly confine myself to the Latin dat, since in Latin Gin contrast to English) more pragmatic factors seem to play & role inthe use of the SSC than mere opiclity or Focaity of the Sub- ject constituent. § 4.2 contains some remarks on the stylitc charac testis of the SSC. 4.1 Pragmatic characteristics On account of the large number of semantic and syntactic restre- tions (ce §2 and § 3.1, respectively) it seems reasonable to hypothe- size thatthe SSC will occur in a highly specific type of context. This is, in fac, the case. In accordance with its subjective nature, on the pragmatic level the SSC almost always funcions as background information’), either by itself or as part of a stretch of discourse with a background status, Whenever an SSC occurs, it appears that 1% Tor the notions Foreground nd Background compare e.g. Bokestein 11987). use te terms rather loosely o indicate the distinction between elements fon the main storyline of the text and elements that constitute all knds cf Sappontve material:Rerum Dictu The Latin Second Supine Constuction 223 the presentation of the foreground information (for example a chronological description of events) is interrupted. In genera, breaks inthe discourse may be ofthe following nature: (i) Breaks may contain an evaluation, reflection or comment of the speaker on the coatent or expression of some predication of the foreground level. Examples are (61) and (62): (61) quos homines, si, id quod facile facts fut et quod fieri debuit (Cie. Sert.'39) (62) haud tam facile dictw est faciendumne fueri: quam potuisse arbror fied ut... (Liv. 2,34, 12). (62) contains the comments of the author on a preceding speech ‘These comment sentences usually contain more elements that under- Tine the subjectivity ofthe expression involved: in this example arbi- rer Tn prose SSCs regularly appear in sentences that refer to the ‘moment of conception ofthe text, asin (62). It is not surprising that dd and memoratu occur predominantly in these structures. Poetry has a comparable mechanism for presenting background information, bt in general itis not the vision ofthe author at his desk through which the events are presented, but the vision of an eyewitness (which, however, may coincide with the author’). Thus, in comment containing sentences the poet may use vin instead of dcr (63) Quin jpsa arrects (ois minabile) in hastis praefigunt capita (Werg. Aen. 9,465-6) ‘The observation that SSC’s preferably occur in background contexts that contain the author's comment on what hat been said in the fore- {ground or on how it has been ssid, explains the non-specific features ‘of most of the supine predicates (see §2.2): the supine refers to a state of affairs already provided by the preceding context and there- fore need not be specific. In this respect the function of the $SC «an bbe compared to that of pronouns. In (64) faitu refers to indervivo: (@) urbanam expoliionem urge ille quidem et Philotimus et Cin- cus; sed eam ipse exebre intereiso; qued est facile fat ( ad Q fr 351,16) (ii) Breaks may supply additional information, i-e. information that plays no essential part on the foreground level. Compare: (65) Namque minum dict ut sit omnis Sermatarum virtus velut extra ipsos (Tac. Hist 1,79).24 Caroline Kroon In this example the chronological description of events is interrupted by a supine-concaining sentence providing secondary information about the cowardice of the Sarmatians (Gi) Breaks may serve to highlight the steucture of the text ie. they bring about the tansiton from one ciscourse theme to another. Introductory, summarizing and concluding sentences belong to this type. Actually, a large part of the SSC’s in the sample occurs in such transitional passages, which, therefore, seem tobe preferred environ- rents for subjecive and ether utterances that do not belong to the foreground information. Mlustrations are giten under (663) and (e6b) (66) a) Inter bellorum .. .curasintercessit res parca dic, sed quae (Liv. 34.1.1) by Delongitudine et latitudine haee sunt, quae digra menorats putem (Plin. NH 2,247) (G62) is the introduction to a new series of events. (666) is the intro- duction to a new discourse theme. ‘The above observations on the background status of the SSC hold good for most of the SSC’s with a copula. Moreover, most of the awtributive SSC’s viz. the non-restrictive ones, can be described in the same way, since the information in non-restrctive attributes is presented by the author as an independently valid message adéi- tional to that communicated in the matrix clause, In other words: nnon-restrictive attributes provide information that more or less resembles the kind of background information referred to above. ‘Compare (67) (67) Hine nata medicina. haec sola naturae placuerat esse remedia ‘parata vulgo, inventu faciba ac sine inpendio et quibus vivimas (lin, NE 24,4). Finally, it must be noted that only a small group of SSC’s does not fic in with this description, vir_ the restrictive attibutive SSC's. In §3.2.itis stated thattheserestietive attsbutes have in common that they function asa substitute for some ‘composite’ adjectival lexeme [As a conclusion of this section on pragmatic characteristics the hypothesis can be formulated that there is some sort of interaction beeween the syntactic characteristics of the SSC on the one hand and its pragmatic role in the discourse on the other: sentences that con- tsin background information will tend to be syntactically independ- cent from surrounding predications that are part of the foreground.Rarum Diet: The Latin Second Supine Consevetion 225; In other words, the syrtactic independence corresponds to 4 prag- matic independence”) 42 Solistc characteristics [As a result ofits highly specific semantic, syntactic and pragmatic characteristics, the Latin SSC ~ in contrast with the English Tough Construction ~ already at an early stage tended to become a stereo- typical expression. The sample contains several indications for this tendency: () First, Plautus uses the SSC prefersbly in proverbial and forms Iaic expressions, and in the exymological figure. (68) . . . quorum opera mihi facta factu facta haee sunt, qui volui ‘fori (Pl. Por. 761), (i) Secondly, the way in which Cicero uses the SSC is very stereoty- Pieal, as appears from table 1 in $2.3: in almost all cases of a per- onal SSC the Subject isa neuter pronoun. There are only 6 excep- tions, three of which are listed under (69)-(69¢): (69) a) Aut docest ... fortirem [me] in patiendo dolore si eum [dolorem] asperum et dificilem pergersu et contra naturam esse quam si malum dixero (Cic, Fin. 4,72) 1) (about dolere) Sed eat tamen reiiciendum: Curt: Asperum est, contra naturam, difficile perpes, triste, durum (Cis, Tas. 2,28) cunt ili [Stoiei]asperum esse dolere, molestum, odiosun, contca naturam, difficile tolerate (Cie. Fin 4,52). In these examples the SSC seems to be part of a set phrase. Other cecamples of supines occurring in set phrases are (703)-(760) and (71)-71b): ° °") The remarks inthis paragraph on the bickground status of SSC contain- ing sentences are more or lst accordance withthe cbservaions of Wehr (984 12-5). [na dacosson of foregrounding strategies in Romance languages ‘She states that ne cf these erates i the Pétion of an artftal pause jst before some inportant or suprising foregrourdinformatin. This pause may be filed up by seme linge clement, provided cha hese cements do ot coo tun any new information that is essential on the Foreground level. According 10 ‘Wehr one of these explicit arf paoes in Latin isthe pareathess Of the ‘ramples thatthe cles as instances of other foregroune-signaling srategin However, Wehr doe: not take ary notice of the occattence ofthe SSC’.26 Carole Kroon (70) a) horrendum et dictu video mizbile monstrum (Verg. Aer 3,28) ) hie vero subisum ac dict mirabile monstrum aspiciuat (erg. Georg. 4,354-5) «) «eum subitem dictuque oritor mirehile monstrur (Verg. Aen. 2,680). (71) a) mune hoc mihi factu est optima, ut... (Plt. Ane 542) bj mune adeo hoc factust optimum ut. (PR. Ax 185) (com pare also St. 83). (ii) A thie indication thatthe SSC isa stereotypical expression is the fact that the material consiss of a significant number of the standard oppositions visu - audi, dic ~ andits and dictu ~ vise “The wie of the SSC in later Latin is almost confined to these stan- éard oppositions"). Moreover, in clasical Latin they appear in the works of authors who, apart from these stereotypical cases, do not use the SSC, e.g, Juvenal: (72) nil dic foedim visugue hace limina tangat / intra quie pater cent Quv. 14,44). “The stereotypical character of these standard oppositions also may explain deviant cases of SSC’s like (73 (73) ipse arduus, altaque pulsat / sidera ... / nec vin fails nee ic adfabils lls erg. Aen. 3,621). which can be paraphrssed:'It is not exsy to bring yourself to look ‘upon Polyphemus, and nobody would dare to speak to hin.’ This ‘example is deviant in two respects: first the supine dctuis never used ‘ith the meaning ‘to speak to’. And secondly, the element dicts is redundant: adfabilis contains the element dicere as well as the poten- tial value, 5. Conclusion For several reasons the Latin SSC, although basically productive, never developed into 2 common expression type comparable to, for example, the English Tough Construction. From an investigation of 4 great number of SSC’s it turns out that the construction is restricied both as regards the semantic festures of its constituent parts and with respect to its syntactic behaviour. 1) An exception is Claedian.228 Caroline Kroon Jansen, HLH, (1974): Le Carateritche della lingua Poetica Romans (tans fon the Dutch, 1941)-a: A Lanell (ed) La Lingus Poetca Latin, Bologna Pion Editor. une, K.8CC.Stegmann (1912): Ausfubdiche Grammank der Lawiniscien Sprache Stslehre, vol. I. Hanover: Habniche Buchhandlung, ‘eumann, Mt, (1999). Nleine Schiften.Zarch: Artemis-Vera. Lindsty, WH. (1895) A Short Historia Latin Grammar. Osfon Maroazeau J. (1949): Qudques Aspects dela Formation du Lain Lévae Pats: Kincseck ‘osten, J an (1983)The Nature of Subjects, Topics and Agen A Cognitive Explanation. Indians Univesity Linguists Clb, Pinker H (1972) Om Latin Adverin, Arter: Novi Holland ~ (198): Lateinsche Syntax und Semantik. Tubingen: Franck. Postal P.M. (1974): On Raising: One Role of English Grammar and ts Tho: ‘etal Implications. Cambridge Mass: MIT Press, Richter, E (1856-1862) De Supine Ltinae Linguae. Kenigsberg. Rissada,R (1984): Coordination ard Juntapostion of Adjectives ia che Latin INP? Glee 62, 202-231 Rosée, H. (1981): Studies in the Systax of the Vertal Noun in Early Latin, ‘Menches: Wibelen Fink Verlag Sjesuand, N. (191) De Vi et Usu Supini Secundi Lainorum. Landae Sraniyr, A. (1968) Laviniche Syntax und Stlisik, Munchen: Beck ‘Ver, ©. (1961): Subject Asignment inthe Impersonal Construcion ie Fresch Tn: Bothestein otal eds). Prediction and Expresin in Functional Gromer London/New York: Academie Pes. ‘Wer. B. (1984: DsharsStateien im Romanischen. Tubingen: Gt Vela, Zebian, G. (1965): The Relative Use of the Ablatves of Quality and Respect in atin Literature? American Jonna of Philology 6, 40-257. nber Nare