Verb phrase

EnCyCLoPEdIA of AnCIEnt GrEEK LAnGuAGE And LInGuIStICS Volume 3 P–Z, Index General Editor Georgios K. Giannakis Associate Editors Vit Bubenik Emilio Crespo Chris Golston Alexandra Lianeri Silvia Luraghi Stephanos Matthaios LEIdEn • BoSton 2014 This is a digital offprint for restricted use only | © 2014 Koninklijke Brill NV table of Contents Volume one Introduction .................................................................................................................................................... List of Contributors ....................................................................................................................................... table of Contents ordered by thematic Category ............................................................................... transcription, Abbreviations, Bibliography ........................................................................................... List of Illustrations ......................................................................................................................................... Articles A–f ..................................................................................................................................................... vii xi xv xxi xxiii 1 Volume two transcription, Abbreviations, Bibliography ........................................................................................... Articles G–o .................................................................................................................................................... vii 1 Volume three transcription, Abbreviations, Bibliography ........................................................................................... Articles P–Z ...................................................................................................................................................... Index .................................................................................................................................................................. This is a digital offprint for restricted use only | © 2014 Koninklijke Brill NV vii 1 547 462 verb (rhêma), ancient theories of the thema, concerning which a certain degree of knowledge is shared by the speech participants (either deriving from the speech context or from the preceding discursive exchange, i.e., the cotext). The textual-discursive thema-rhema relationship is also referred to as a topic-comment relationship (→ Functional Grammar and Greek; → Topic). Bibliography Basset, Louis. 1994. “Platon et la distinction nom/verbe”. In: Les classes de mots. Traditions et perspectives, ed. by Louis Basset and Marcel Pérennec, 47–65. Lyon. Bertagna, Maria I. 2006. “All’origine del valore tecnico di rhēma: Platone”. In: Esegesi letteraria e riflessione sulla lingua nella cultura greca, ed. by Graziano Arrighetti and Mauro Tulli, 89–112. Pisa. Hoekstra, Marieke and Frank Scheppers. 2003. “Onoma, rhēma et logos dans le Cratyle et le Sophiste de Platon. Analyse du lexique et analyse du discours”, L’Antiquité Classique 72:55–73. Spina, Luigi. 2001. “Intorno a ‘rhema’ ”. In: Métalangage et terminologie linguistique. Actes du colloque international de Grenoble (Université Stendhal, Grenoble III, 14–16 mai 1998), ed. by Bernard Colombat and Marie Savelli, 253–264. Leuven – Paris. Swiggers, Pierre and Alfons Wouters. 2002. “Grammatical theory in Aristotle’s Poetics”. In: Grammatical theory and philosophy of language in antiquity, ed. by Pierre Swiggers and Alfons Wouters, 101–120. Leuven – Paris – Sterling. Tempesti, Anna M. 1982. “Da ‘rhēma’ a ‘verbum’. Contributo alla storia di una definizione”, Studi e ricerche 5:163–197. Alfons Wouters Pierre Swiggers Verb Phrase 1. Verb Phrase: A Definition At a syntactic level, a phrase can be defined as made up of one or more words corresponding to a single unit in the sentence and, consequently, forming a constituent (any node plus all the nodes that it dominates). The head of the phrase is the word which assigns the grammatical features characteristic of the phrase itself, whereas the other elements are its dependents, i.e., arguments (phrases which are required to occur with the head) and → adjuncts (optional phrases expressing information like instrument, location, manner and time). A verb phrase (henceforth, VP) is a phrase headed by a → verb. The traditional description of the VP in linguistics is based on English: following this description, illustrated here in simple terms, the VP may be constructed from a single verb, or it may contain: (i) the auxiliary which precedes the verbal head; (ii) arguments, optional specifiers and adjuncts, like prepositional phrases and adverbial phrases. From a theoretical point of view, the usefulness of the category of VP in Greek is disputed. In the following sections, we will examine some aspects of the Greek syntax which seem to provide evidence against or in favor of the syntactic relevance of the category of VP in this language. In other words, we will try to find syntactic clues of the fact that in Greek finite verbs and their objects, primarily, form a constituent functioning as a single syntactic unit. As to how nonfinite verbs (→ infinitives and → participles) can be interpreted in this respect, this is a complex issue which cannot be treated here (→ Argument Clause). 2. Syntactic Constituents: Word Order and Discontinuity The first and, in some respects, main problematic issue with regard to the identification of the category of VP in Greek is represented by → word order: as is well-known, this language is characterized by freedom of word order. In particular, the verb has been described as the element of the clause which shows “la plus grande mobilité ” (Humbert 1960:92), with the consequence that “il ne semble point avoir une place où il se complaise. Il en résulte qu’il peut être précédé aussi bien que suivi de son objet, précédé aussi bien que suivi par les participes qui expriment des circonstances accessoires de l’action verbale” (Humbert 1960:96). In other words, Greek exhibits every possible order of the verb and its complement(s): for instance, focusing on the direct object, both OV and VO are attested (studies based on statistical analysis have reached different results with regard to which of the two orders has to be considered as predominant: cf. Taylor 1994). However, what is more relevant to our topic is the fact that the verb and its direct object may be separated by other elements appearing in the same clause. This has been considered as the main obstacle to admitting the syntactic relevance of the VP in Greek. An example is the following, where the noun functioning as a subject (Períandros) is located between the direct object and the verb (from Taylor 1994:7): This is a digital offprint for restricted use only | © 2014 Koninklijke Brill NV verb phrase (1) epeíte gàr tḕn heōutoû gunaîka Mélissan Períandros apékteine ‘For after Periander had killed his own wife Melissa’ (Hdt. 3.50) In other words, if we admit that the category of VP is relevant to syntactic processes in Greek, we should also recognize that permutation of constituents is allowed, and try to explain why this type of phrase may be discontinuous. This approach is followed, for instance, in Morrel (1989): on the basis of X-bar theory, she assumed that “the basic structure of Greek consists of an Imax node that will accommodate a subject noun phrase followed by a Vmax projection that is headed by a verb form” (Morrel 1989:14), where Xmax represents the maximal projection of the head and the highest node of a given phrase. In other words, the VP is described as consisting of the verb, which is the head, whereas the complement of this first node is represented by the object position. In Morrel’s opinion (1989:14–30), what confirms this are statistical surveys showing a clear preference for Subject-Verb (SV) over Verb-Subject (VS) ordering, and references to word order by ancient commentators who were native speakers, which sustain the idea that the underlying structure of Greek was fundamentally SV(O), although this structure may greatly diverge from the surface structure (Morrel 1989:30). Processes of movement during the delta phase should be invoked in order to explain the freedom of word order, which is only apparent (Morrel 1989:116). An interesting example is the following, where the head of the VP is in sentence-initial position, in order to express emphasis, and precedes both the subject noun phrase and the object noun phrase, marked as a → genitive (from Morrel 1989:138; cf. also examples in section 5.a.): 463 tion occur far less frequently” (1989:139). In her opinion, this happens in order to avoid syntactic discontinuity, and is particularly interesting in light of the fact that discontinuous phrases are allowed in Greek, since this is consistent with its rich inflectional system. However, “the fact that elements move far more often in manners that preserve syntactical continuity than in ways that create discontinuity is strong evidence for the premise that constituency and hierarchical structure play a fundamental role in the language” (Morrel 1989:139). According to this analysis, movements to the third and fourth specifier positions, which create highly discontinuous syntactic structures, are comparatively much less frequent (1989:149), in order to preserve constituency and hierarchical structuring. As mentioned above, some scholars do not share this type of approach. Instead, they claim that freedom of word order and discontinuity are those features which unequivocally suggest that Greek lacks a VP, as will be shown in the next section. 3. Greek as a Non-Configurational Language? In generative grammar, a distinction has been proposed between configurational and nonconfigurational languages: “the term nonconfigurational implies that the language has a rather flat (as opposed to hierarchical) phrase structure. In the most highly nonconfigurational languages, there is little or no evidence for the verb phrase as a syntactic constituent, and there are no subject-object asymmetries that require a structural explanation” (Devine & Stephens 1999:142). Apart from lacking a VP, non-configurational languages may have the following properties, among others: (2) thaumázō dè égōge tês tólmēs toû adelphoû kaì tês dianoías ‘I am astounded at the shameless spirit shown by my brother’ (Antiph. 1.28) – – – – – Extremely free word order Syntactically discontinuous elements Complex case system Lack of expletives Null anaphora Morrel (1989) also notes that within the corpus on which her analysis is based, “in comparison with the number of fronting processes that move a single minimal or maximal projection to the first specifier position, fronting processes that move elements beyond the first specifier posi- It must be said that the distinction between these two types of languages is not unanimously accepted nowadays, not even within the generative framework. Moreover, it has been noted that languages which are clearly configurational may share some properties with non-configurational This is a digital offprint for restricted use only | © 2014 Koninklijke Brill NV 464 verb phrase languages, whereas languages which apparently are non-configurational may show syntactic traces of a VP. As already pointed out in section 2, Greek has a rich case system and a free word order; moreover, apart from the existence of different possible word order patterns between verb, direct object and adjuncts (i.e., the hypothetical constituents of the VP), we find cases of discontinuous noun phrases (→ Noun Phrase) and adpositional phrases (→ Adpositional Phrase). Furthermore, a special instance of discontinuity in Greek is represented by → hyperbaton (cf. section 3.a.), which, from a cross-linguistic point of view, appears to be related to the other syntactic features proper to non-configurational languages (Devine & Stephens 1999:142). Mainly on the basis of these considerations, Greek has been described as a non-configurational language: Taylor (1988) was the first to assume that Greek has gradually changed from non-configurational to configurational. Devine and Stephens (1999:142–153) also assumed that certain features of Greek syntax, well attested in particular in Homeric Greek and presumably of Indo-European origin (→ IndoEuropean Linguistic Background), document the non-configurational nature of the language at an earlier stage. Apart from free word order and discontinuity, associated with hyperbaton, these features are: the absence of articles, the particular use of prepositions (i.e., their occurring as adverbs, disjoined by their complements) and the paratactic nature of Homeric syntax, shared by Vedic and Germanic poetry, which resembles the fact that non-configurational languages prefer phrasal juxtaposition for conjunction and disjunction (Devine & Stephens 1999:147–148). More precisely, Devine and Stephens (1999:151) hypothesized that “what we can reconstruct from survivals in Homer is probably a not so extreme version of nonconfigurationality, in which some arguments are admitted into the nuclear phrase and others are adjuncts, and some modifiers are integrated and others are paratactic”. 3.a. Hyperbaton By definition, hyperbaton takes place when the subconstituents of a noun phrase are not continuous to one another, but are interspersed with other elements of the clause. In the normal type of hyperbaton (Y₁ hyperbaton), one can note the extraction of an adjective referred to the object taken by the verb. As pointed out in the preceding section, following Devine and Stephens (1999) the phenomenon of the hyperbaton, as is documented in Homeric Greek, represents a trace of the prehistoric (Indo-European) nonconfigurational syntax. As they state, in Homer “a single noun could easily form a phrase with the verb, but a more complex structure like noun plus adjective or noun plus noun (→ coordination) would run into greater resistance. One way of handling a modified lexical argument in a single sentence would be to allow the noun to form a phrase with the verb and leave the paratactic modifier in adjunct position” (Devine & Stephens 1999:151). As compared to Homeric Greek, Classical Greek presents a partly different state of affairs: whereas in poetry Y₁ hyperbaton is still attested in a non-configurational form, its occurrences in prose document an intermediate stage from a non-configurational to a configurational syntax (Devine & Stephens 1999:203). More precisely, in Classical Greek, the superficial discontinuity caused by hyperbaton implies that the VP is unordered, but not necessarily that it does not exist: indeed, although syntactic discontinuity given to premodifier hyperbaton “may at first sight appear to be a particularly strong indication of flat unstructured serial word order” (Devine & Stephens 1999:3), a deeper investigation of the Greek data reveals “consistent cross-categorial patterning for premodifiers in both discontinuous and continuous phrases, which clearly calls for a phrase structural account” (1999:3). In particular, following Devine and Stephens (1999), Y₁ hyperbaton performs a pragmatic function in terms of focus marking: the modifier is placed in a left position, different from its usual position in the noun phrase, since it has strong focus, whereas the noun referred to represents tail material. Compare the following occurrences of the (object) noun phrase pâsan tḕn pólin ‘all the city’ (from Devine & Stephens 1999:13): (3) idṑn d’ ēdikēkóta . . . pâsan tḕn pólin ‘When I discovered that he had defrauded all the city’ (Dem. Or. 24.8) (4) ou gár esti díkaion tḕn mèn khárin, hḕ pâsan éblapte tḕn pólin, toîs tóte theîsin hupárkhein ‘It is not fair that those legislators should enjoy a popularity that injured all the city’ (Dem. Or. 3.13) This is a digital offprint for restricted use only | © 2014 Koninklijke Brill NV verb phrase In the first example, the verb precedes the adjective and the object noun phrase, in the second the adjective (pâsan) is focused on by means of hyperbaton, and precedes both the verb (éblapte) and the direct object (tḕn pólin). There is another type of hyperbaton (Y₂ hyperbaton), where it is the direct object which precedes both the adjective referred to it and the verb. Also Y₂ hyperbaton has a specific pragmatic function: the noun can be a weak focus or a topic, whereas the adjective normally represents a weak focus. In both types of hyperbaton, an additional element or adjunct can be placed internally or externally to the structure. I quote an example illustrating an Y₂ hyperbaton where the adjunct (es tò Árgos ‘to Argos’) appears inside the structure (from Devine & Stephens 1999:89): (5) . . . aphikómenoi es Tegéan lógous proúpempon es tò Árgos xumbatēríous ‘[The Lacedaemonians] arriving at Tegea sent on to Argos proposals of accommodation’ (Thuc. 5.76) The evolution from a non-configurational (Homeric) syntax to a configurational syntax can be considered as completely achieved in the Greek of the → New Testament (Devine & Stephens 1999:203; cf. also Palmer 1995). 3.b. Lack of Expletives and Null Anaphora Greek lacks expletives in syntactic contexts where they typically appear in configurational languages, as with impersonal verbs (for instance, with weather verbs and impersonal passives; → Impersonal Verbs/Constructions). Moreover, it extensively shows a feature generally attributed to non-configurational languages, the so-called → null anaphora, i.e., omission of pronominal arguments. In principle, the phenomenon of null objects in Greek may be regarded as evidence against the existence of a VP. Indeed, the → direct object of transitive verbs (→ Transitivity) can be omitted also when it refers to a definite antecedent, which is not necessarily a direct object. Luraghi (2003) has examined this phenomenon in depth, by focusing on null objects which are definite and referential (henceforth NOs). This is illustrated as follows (from Luraghi 2003:167): 465 (6) toîsi dè dexiòn hêken erōidiòn engùs hodoîo Pallàs Athēnaíē: toì d’ouk ídon ophthalmoîsi núkta di’ orphnaíēn, allà klánxantos ákousan. ‘Athena sent them a heron to the right of their route: they could not see it in the dark night, but heard it screaming.’ (Hom. Il. 10.274–276) In example (6), the direct object (erōidión) is shared by the verbs (hêken, ídon and ákousan) of the three subsequent clauses, but is mentioned only in the first. Indeed, in the other two clauses we find an instance of definite NOs (as observed by Luraghi 2003, pronominal objects are necessary in English in order to make the translation correct from a grammatical point of view). In Greek, the omission of the definite direct object may be discourse conditioned – i.e., the direct object is omitted if it is recoverable from the preceding mentions of the referent or, more generally, from the textual context – or it may be syntactically conditioned. By focusing on the last case, there are three contexts in which the omission of the direct object is conditioned by the syntactic context, with the result of its being obligatory: i. Conjunction participles governed by a verb form with which they share the same subject and the same object: both are expressed only once. ii. Coordination: in two or more coordinated clauses linked by coordinating conjunctions (especially by kaí), the shared direct object is expressed only once, i.e., in the first clause. Some examples may simply be interpreted as cases of VP coordination, characterized by the reduction of a part of the VP itself, as may happen in English. In other examples, however, two distinct coordinated clauses are found (from Luraghi 2003:179): (7) kaí min Athēnaîoi dēmosíēi te éthapsan autoû têi per épese kaì etímēsan megálōs ‘The Athenians buried him at public expense on the spot where he fell and gave him much honor’ (Hdt. 1.30) In a case like (7), we could not say that the VP is reduced: indeed, the omission of the direct object (min) would be impossible in a language like English. This is a digital offprint for restricted use only | © 2014 Koninklijke Brill NV 466 verb phrase iii. Yes/no questions: as a rule, the object which is coreferential with the direct object mentioned in the question is omitted in the answer. As a consequence, often this consists of the sole verb. I will come back to this point in section 5.c. In the next section we shall briefly examine some constituency tests for identifying a VP in Greek. 4. What is Evidence for the Existence of a VP in Greek? Constituency tests can be defined as a diagnostic means for identifying the constituent structures of a sentence, i.e., phrases, in a given language. In the relevant literature, various constituency tests have been proposed, mainly on the basis of English. Some of them may be applied to Greek, and they seem to provide evidence for the syntactic relevance of the VP. The passivization test, for instance, is easily applicable to Greek, where, as is well-known, the object of an active verb may be changed into the subject of the corresponding passive verb (→ Passive (Syntax), → Passive (Morphology)): consequently, on the basis of this test, the verb and the direct object may form a constituent. In this respect, Greek differs from non-configurational languages, which, in principle, lack a transformational passive. Significant results may be obtained also by applying the coordination test, which is based on the assumption that only constituents can be coordinated. As pointed out by Morrel (1989:40–55), in particular, the conjunction te functions as a delimiting element for conjoined phrase structures (→ Coordination (includes Asyndeton)) and can, thus, be interpreted as an “indicator of constituency” (1989:55). In the following example (from Morrel 1989:50), for instance, te delimits the VP the head of which is mēkhanôntai, and separates it from the second VP (paraskeuázousin): Although the phenomena examined so far may be recognized as formal signs of non-configurationality in Greek, they do not provide incontrovertible evidence of its lacking a VP. To sum up, firstly, the idea that discontinuous phrases were not constituents in Greek has no explanatory force: indeed, “free word order could arise as a result of pragmatically driven movement from an underlying configurational structure” (Devine & Stephens 1999:143), which is true also for discontinuity (cf. sections 2 and 3). Second, even if we admit that Homeric Greek preserves traces of a non-configurational syntax, things are partly different in Classical Greek (cf. section 3.a.). Third, there are other phenomena in Greek which, on the contrary, seem to prove the syntactic relevance of the VP. According to Morrel (1989), for instance, interesting results can be reached by looking at intrusions, i.e., parenthetical expressions (including the vocative) which interrupt the expected syntactical hierarchy of a sentence. By definition, intrusions “show a distinct preference for appearing at ‘major’ structural boundaries. Intrusions tend to come at the boundary of a ‘major’ maximal projection” (1989:56). The data from Morrel’s corpus show that intrusions in Greek often mark the boundaries of verb phrases as distinct from noun phrases with the function of subject. An example is the following, where the intrusive apostrophe (ô ándres) appears between the subject and the VP (from Morrel 1989:58): (8) Hē dè pallakḕ toû Philóneō tḕn spondḕn háma enkhéousa ekeínois eukhoménois há ouk émelle teleîsthai, ô ándres, enékhei tò phármakon ‘But Philoneos’ mistress, who poured the wine for the libation, while the men offered their prayers – prayers never to be answered, gentlemen – poured in the poison with it’ (Antiph. 1.19) 5. Constituency Tests for identifying a VP (9) ou gàr dḗpou martúrōn g’ enantíon hoi epibouleúontes toùs thanátous toîsi pélas mēkhanôntaí te kaì paraskeuázousin ‘Those who plot the death of their neighbors do not, I believe, form their plans and make their preparations in front of witnesses’ (Antiph. 1.28) Three other tests are particularly relevant to our case: (a) topicalization; (b) pro-form substitution; (c) question test. I will briefly examine each of them in turn. This is a digital offprint for restricted use only | © 2014 Koninklijke Brill NV verb phrase 5.a. Topicalization If in a given language a sequence of words may be moved to the front of the sentence in order to be topicalized, this may be considered as a clue that such words form a constituent in that language. In Greek the verb and its complement(s) may be topicalized: in this case the subject occupies the (marked) final position. Some examples are quoted here belonging to different chronological stages of the language (from Taylor 1994:9; on Mycenaean (→ Mycenaean Script and Language), cf. Panagl 1999): (10) all’ ei mèn dṓsousi géras megáthumoi Akhaioí ‘But if the great-hearted Achaians would give me a prize’ (Hom. Il. 1.135) (11) épempsan gàr dḕ pentakosías mnéas arguríou hoi Kurēnaîoi ‘The Cyreneans sent 500 minas of silver’ (Hdt. 3.13) (12) hōs dè ḗkousan toùs lógous toútous hó te stratēgòs toû hieroû kaì hoi arkhiereîs ‘When the captain of the temple and the chief priests heard these words’ (Acts 5.24) 5.b. Pro-Form Substitution: do so-Test To apply the pro-form substitution test implies to replace the (presupposed) constituent by means of pronominalization, i.e., with a proform like a pronoun. A special instance of it is represented by the do so-test: if the verb and its complement(s) may be substituted by a proform like do so, one can infer that the verb and such complement(s) form a constituent, i.e., a VP. This test may be indirectly applied to Greek, by examining the occurrences of do so-phrases in the second clause, in order to avoid the repetition of the verb attested in the first clause (this use is more frequent in prose than in poetry). The application of this test shows that a do sophrase substitutes the whole VP, but cannot be referred to the whole preceding clause. A clear example is the following, where toûto poiḗsō substitutes the VP erṓtēson autoús, with a change in the subject and in the grammatical features of tense, aspect and mood (→ Mood and Modality, → Tense/Aspect): (13) ei d’apisteîs, erṓtēson autoús, mâllon d’egṑ toûth’ hupèr soû poiḗsō 467 ‘If you doubt my word, ask them; or rather I will do it instead of you’ (Dem. Or. 18.52) 5.c. Question Test The question test is usually applied in order to verify the constituency of a VP, by testing the ability of a series of words to stand alone in the answer to a given question. As already mentioned in section 3.b., in the answers to yes/no questions, the direct object is generally omitted if it is coreferential with the object mentioned before. I quote an example illustrating how the answer may be represented by the verb alone (from Luraghi 2003:183): (14) – thômen oûn boúlei, éphē, dúo eídē tôn óntōn, tò mèn horatón, tò dè aidés? – thômen, éphē. – ‘Now, shall we assume two kinds of existence – said he –, one visible, the other invisible?’ – ‘Let us assume them – said (Cebes).’ (Pl. Phd. 79a) Following Luraghi (2003:183–184), omission of the direct object can be explained as due to syntactic constraints only in part: indeed, it rather can be interpreted as a strategy linked to a relevance principle, since it allows one to make the answer shorter and to limit the repetition of the question to the most relevant part of the information. In this respect, it is interesting to note that the direct object is not the only syntactic element which may be omitted. Any part of the VP can be deleted, including the verb (from Luraghi 2003:184): (15) – – – – taûta ḕ tí eroûmen? taûta nḕ Día, ô Sṓkrates. ‘Shall we say that, or what?’ ‘That is what we shall say, by Zeus, Socrates.’ (Pl. Crit. 50c) In (15), where we have a disjunctive question, the answer is made up of the only direct object, without any verb. Similarly, in (16), only the indirect object required by the unexpressed verb (dokeî) is mentioned: (16) – kaí moi lége; dokeî tí soi eînai híppou érgon? – émoige. This is a digital offprint for restricted use only | © 2014 Koninklijke Brill NV 468 verb phrase – ‘Tell me then: would you say that a horse has a specific work or function?’ – ‘I would.’ (Pl. Resp. 352d8–e1) It is worth noting that the omission of syntactic elements may take place in replies to all kinds of questions, as illustrated below: (17) – – – – tí toûto légeis? empeirían égōgé tina. ‘What thing do you mean?’ ‘I mean a certain habitude.’ (Pl. Grg. 462c2–3) In (17), only the subject and the direct object are mentioned in the answer, whereas the verb is omitted. In conclusion, the question test does not give positive results in the search for syntactic evidence of the existence of the VP in Greek. However, by means of a deep investigation of the answers to yes/no questions it is possible to see that “often they contain a constituent only, or a part of it, which has the highest communicative dynamism in the question, and conveys the most relevant part of the information questioned” (Luraghi 2003:183). In other words, the deletion of the object as well as of the verb or of other syntactic elements in this special type of sentence does not allow us to identify a VP in Greek, but it does not prove its non-existence, since it can be interpreted as pragmatically determined. 6. Subject/Object Asymmetries in Greek Apart from the tests examined so far, there is an interesting aspect of Greek morpho-syntax which could be relevant to our topic. It is usually assumed that in languages lacking a VP there is no structural divergence between subject and object; on the contrary, in configurational languages, one expects to find subject-object asymmetries. By focusing on Greek, traces of such subject-object asymmetries may be found in some verbal compounds. First of all, an interesting case is represented by verbs in -éō, in which a nominal form precedes the predicate, as in androktonéō ‘to slay men’, karpologéō ‘to gather fruit’, oinometréō ‘to measure out wine’, paidopoiéō ‘to beget children (of men); to bear children (of women)’. In these forms, based on a process which resembles syntactic incorporation (cf. Pompei 2006), the first element generally corresponds to the direct object taken by the verb, but cannot correspond to the subject. This is consistent with the hypothesis that only the object, as opposed to the subject, is governed by the verb, since it is part of a VP. Considering that also adjuncts may be part of the VP, it is equally consistent that, among verbs in -éō, there are compounds in which the first element corresponds to a complement expressing instrument (kērodoméō ‘to build with wax’), location (hulomakhéō ‘to fight in the woods’), or having a comitative value (androkoitéō ‘to sleep with a man’). A similar phenomenon takes place with verbal adjectives in -to-: in those which are compounds, the first nominal element cannot correspond to the subject. It rather corresponds to a direct object (aigibótos ‘feeding goats’, odunḗphatos ‘killing pain’) or to a different complement (douriktētós ‘won by the spear’); it may also be an adverb referred to the action (eupoíētos ‘wellmade’, polúplanktos ‘much-wandering’). In conclusion, these kinds of subject/object asymmetries involving the process of composition seem to offer indirect, although significant, support to the existence of a VP in Greek. Bibliography Crespo, Emilio. 1997. “Sintaxis de los elementos de relación en griego clásico”. In: Actas del IX congreso español de estudios clásicos, 3–42. Madrid. Devine, Andrew M. and Laurence D. Stephens. 1999. Discontinuous syntax. Hyperbaton in Greek. Oxford. Hewson, John and Vit Bubenik. 2006. From case to adposition. The development of configurational syntax in IndoEuropean languages. Amsterdam – Philadelphia. Humbert, Jean. 1960. Syntaxe grecque. Paris. Kiparsky, Paul. 1997. “The rise of positional licensing”. In: Parameters of morphosyntactic change, ed. by Ans van Kemenade and Nigel Vincent, 460–494. Cambridge. Luraghi, Silvia. 2003. “Definite referential null objects in Ancient Greek”, IF 108:169–196. Morrel, Kenneth S. 1989. Studies on the phrase structure of early Attic prose. Ann Arbor. Palmer, Michael. 1995. Levels of constituent structure in New Testament Greek. New York – Bern. Panagl, Oswald. 1999. “Beobachtungen zur mykenischen Syntax”. In: Floreant Studia Mycenaea. Akten des X. Internationalen mykenologischen Colloquiums in Salzburg vom 1.-5. Mai 1995, ed. by Sigrid Deger-Jalkotzy, Stefan Hiller and Oswald Panagl, 487–494. Vienna. Pompei, Anna. 2006. “Tracce di incorporazione in greco antico”. In: Fonologia e tipologia lessicale nella storia della lingua greca, ed. by Pierluigi Cuzzolin and Maria Napoli, 216–237. Milan. Taylor, Ann. 1988. “From non-configurational to configurational: a study of syntactic change in Greek”. In: The Penn This is a digital offprint for restricted use only | © 2014 Koninklijke Brill NV verb phrase Review of Linguistics. 12th Penn Linguistics Colloquium, February 1988, 1–15. ——. 1994. “The change from SOV to SVO in Ancient Greek”, Language Variation and Change 6:1–37. Maria Napoli Verba Dicendi Verba dicendi (‘verbs of speaking’) are plentiful in the literature of Ancient Greece, a tribute to the high esteem in which speech was held in the predominately oral Indo-European culture. Speech divided the human and animal world, and, to some extent, the human and divine. Verbs of speaking regularly developed from roots referring to mental processes: ‘plan’, ‘reason’, ‘judge’, ‘think’, etc. (e.g. *men- ‘think’ > Hittite memaḫḫi ‘I speak’), and “come later to be used for the oral expression of these processes” (Buck 1915:137). Many have cognates in sister languages, allowing a large number of verbs referring to speech to be reconstructed for the mother language. Some thirty-five verbs of speaking occur with notable frequency in Homer, the most common being phēmí ‘say, declare’ (Linear B pa-si ‘says, affirms’ (Hooker 1980:61); cognates Latin fāma, likely Sanskrit bhāṣate ‘says’, Armenian bay ‘says’). Preverbs may add nuances to basic verbs: katáphēmi ‘say yes, affirm’. Formulaic phrases involving speaking verbs (e.g. phasí ‘they say, it is said’) are common. Other frequently occurring verbs in Homer include eîpon ‘I spoke, said’ (directly comparable to Vedic ávocam ‘I spoke’), with cognates Armenian gočem ‘I call’, Tocharian AB weñ ‘will speak, say’, Avestan vac‘say’; audáō ‘speak’ (cognate with Sanskrit vad‘speak’); peíthō ‘persuade’ (with accusative of the one persuaded + infinitive); phōnéō ‘speak aloud’; agoreúō ‘harangue, speak’ (with an aspectual nuance: ‘speak’ rather than ‘say’, and referring to public address as does agoráomai ‘hold assembly, speak’); eírō ‘say’ (cognate with Hittite wer(i)ye ‘call, summon’); eúkhomai ‘pray, vow, declare’ (cf. Linear B middle e-u-ke-to = eúkhetoi (Hooker 1980:53)), with athematic aorists in Greek and Avestan suggesting antiquity; kaléō ‘call, summon’ (cognate with Hittite kalless‘lure’); muthéomai ‘relate, tell’ (one of several Homeric speaking verbs later replaced by légō); aráomai ‘pray, wish, curse’ (cognate with Hittite ariya ‘pray’, Vedic āryati ‘praises’). See Owen and Goodspeed (1969:4–22) for a full list. 469 Because of their important status and plentiful occurrence, verba dicendi are especially valuable in tracking changes in lexicon, semantics and syntax. For instance, laléō, used originally to express the inarticulate sounds of animals (dogs and monkeys in Plutarch), had, by Hellenistic times, lost this special sense and become the normal verb ‘to speak’ (Buck 1915:13). The verb légō originally referred to counting and sorting, and still has that sense in Homer (Il. 23.239), only later turning toward ‘say’ (in Hesiod), from then on maintaining its role for some 2000 years as a common verb meaning ‘say’ (Buck 1915:7). Appearing regularly in reported speech situations, verba dicendi allow us to track changes in subordinate structures which follow them. Greek shared with Latin the accusative + infinitive construction, possibly as an inheritance from the mother language: kaí té me (accusative) phēsì mákhēi Trṓessin arḗgein (infinitive) (lit: ‘and she says me to help the Trojans in battle’) ‘. . . and she says that I am helping the Trojans in battle’ (Hom. Il. 1.521). Both Latin and Greek developed alternative subordinating strategies to the accusative + infinitive, including ‘that’ clauses with hóti or hōs (always an option in Greek (Moore 1957:140)), with Latin steadily moving toward subordinate clauses with quod. Following syntax may affect word meaning: for instance, eîpon, followed by a subordinate clause with hóti or hōs, generally means ‘said’; followed by the infinitive construction, however, the meaning is ‘commanded’; eîpon, followed by the infinitive with the meaning ‘said’ is rare but “occurs in good Attic prose” (Smyth 1920:4.45.2017c). Bibliography Buck, Carl D. 1915. “Words of speaking and saying in the Indo-European languages”, American Journal of Philology 36:1–19, 125–154. Hooker, J. T. 1980. Linear B: an introduction. Bristol. Moore, R. W. 1957. Comparative Greek and Latin syntax. London. Owen, William B. and Edgar J. Goodspeed. 1969. Homeric vocabularies. Oklahoma. Smyth, Herbert W. 1920. A Greek Grammar for Colleges, accessed from Perseus Digital Library Project. Ed. Gregory R. Crane. Tufts University. http://www.perseus. tufts.edu (Updated April 10, 2011) This is a digital offprint for restricted use only | © 2014 Koninklijke Brill NV Sarah Rose