Adjectival periphrasis in Ancient Greek: the categorial status of the participle.

Bentein, Klaas

Introduction

The subject of this article is what has been called 'adjectival periphrasis' in studies on the Ancient Greek participle (see Bjorck 1940 for the term 'adjektivische Periphrase'). This term refers to combinations of the type [TEXT NOT REPRODUCIBLE IN ASCII] (Aristoph. Thesm. 77), 'he is alive', where [zeta][??]v may be said to have an 'adjectival' function, in that it refers to a property, similarly to a regular adjective. (1) This can be contrasted with cases such as [TEXT NOT REPRODUCIBLE IN ASCII] (Xen. Cyr. 8.1.21), 'he was their leader', and [TEXT NOT REPRODUCIBLE IN ASCII] (Soph. Aj. 1324), 'he was doing such things', where the participle denotes an identity and an action respectively. While the older studies (e.g. Bjorck 1940; Aerts 1965) have mostly concentrated on the present participle, it should be noted that the perfect and occasionally even the aorist participle (2) could have a similar 'adjectival' function, as in [TEXT NOT REPRODUCIBLE IN ASCII] (Xen. Cyr. 2.1.11), 'they are armed', and [TEXT NOT REPRODUCIBLE IN ASCII] (Soph. OT 90), 'I am fearing prematurely'.

Two questions which have been central in previous research on adjectival periphrasis are the following:

(a) Can the different types of adjectival periphrasis (i.e. with the present, perfect and aorist participle) be given a unified semantic description?; and

(b) What is the categorial status of the participle used in this type of construction? Is the participle 'adjectivised?'

Having dealt with the former topic in earlier work (Bentein 2013a), I now turn to the categorial status of the participle, concentrating again on Archaic and Classical Greek and adopting a Cognitively-inspired perspective (for the Cognitive Linguistic framework, see e.g. Langacker 1987, 2000, 2008). (3) In concluding this article, I will reconsider the relationship between the concepts of adjectival and verbal periphrasis, building on insights from an earlier article of mine (Bentein 2011). While adjectival and verbal periphrasis have generally been considered mutually exclusive concepts, I argue that their relationship is in fact more complex.

1. Conceptual integration

For reasons of space, I will not repeat my semantic analysis of adjectival periphrasis here, for which I refer to Bentein (2013a). Suffice it to say that --next to the fact that the participle has a property-referring function (see above)--adjectival periphrasis can be characterised in terms of low transitivity (with transitivity in the scalar sense of Hopper & Thompson 1980) and a low degree of conceptual integration (Langacker 2005, 2008) between the verb [TEXT NOT REPRODUCIBLE IN ASCII] and the (present/perfect/aorist) participle, whereby only a single, representative component state of ei^u is elaborated by the participle (as with regular adjectives; what is called the 'copulative' use), rather than all of its component states. (4)

It would seem that the notion of conceptual integration was used in the older works as well, though mostly implicitly, and in a somewhat different sense. The term 'adjektivische Periphrase' was coined by Bjorck (1940) specifically to distinguish this type of 'adjectival' construction from 'truly' periphrastic constructions, inter alia progressive periphrases of the type illustrated in (1), after which Bjorck named his work. In the former case, Bjorck (1940:28-29) is of the opinion that 'jetzt stellt sich auch klar heraus, dass der ganze Begriff Periphrase hier kunstlich ist und nur in der vorlaufigen Systematik eine Berechtigung hat.' (5)

(1) [TEXT NOT REPRODUCIBLE IN ASCII] (Lk. 5.17).

it happened on one of those days, that he was teaching (tr. WEB).

Initially, it is not entirely clear on what grounds Bjorck believes this to be the case: a typical adjectival periphrasis of the type [TEXT NOT REPRODUCIBLE IN ASCII], 'it is fitting', complies quite well with Bjorck's definition of periphrasis: 'jede solche Verbindung der Kopula mit einem pradikativen Partizip... die ohne Anderung des Tatsacheninhalts gegen eine einfache Verbalform vertauscht werden kann' (6) (Gildersleeve 1980 [1900]:81, for example, explicitly equates [TEXT NOT REPRODUCIBLE IN ASCII] with the synthetic form [pi][rho]o[sigma][??][kappa][epsilon][iota]). Bjorck somewhat clarifies his position by referring to what is nowadays called conceptual integration: he notes, with regard to adjectival periphrasis (referring to Rehdantz), that 'es wird ein Eigenschaftsword pradizierend zum Subjekt gestellt, nicht aber "das Pradikat in das Particip mit der Kopula aufgelost" (Rehdantz' Indices s.v. Participium).' (7) For Bjorck, however, this low degree of conceptual integration lies primarily with the fact that the adjectival participle is categorially an adjective ('adjektiviert,' adjectivised): (8) in his opinion an expression such as ouucpepovTec, [TEXT NOT REPRODUCIBLE IN ASCII], Dem. 24.24) does not mean 'diese Gesetze sind nutzend' (lit. 'these laws are helping') but rather 'diese Gesetze sind nutzlich' ('these laws are helpful') (Bjorck 1940:29). (9) Bjorck does make a distinction between so-called 'Daueradjektivierungen' (i.e. cognitively entrenched adjectivisations) such as [TEXT NOT REPRODUCIBLE IN ASCII], 'it would be necessary' (Pl. Leg. 800d) and 'Gelegenheitsadjektivierungen' (i.e. cognitively non-entrenched adjectivisations) such as [??][sigma][tau][iota]... [pi][rho]o[??][chi]ov[sigma]a, 'it is jutting out' (Thuc. 4.109.2) or [TEXT NOT REPRODUCIBLE IN ASCII], 'it injures bodies' (Pl. Leg. 933a), (10) but he considers the participle in both types 'adjectivised'. He stresses that they can only be distinguished on the basis of frequency (Bjorck 1940:23) and notes that 'ein prinzipieller Unterschied, so dass die Gelegenheitsperiphrasen etwa mehr "verbal" waren, kommt also nicht in Frage' (11) (Bjorck 1940:24; cf. also Bjorck's generalising discussions about 'Adjektivpartizipien' [p. 29] and the 'adjektiviertes Parti zip' [p. 34]).

More recent studies have similarly drawn attention to the adjectivisation of the participle, though mostly limiting themselves to what Bjorck calls Daueradjektivierungen. (12) Ceglia (1998:26), explicitly referring to Bjorck, notes that the participles of verbs such as [TEXT NOT REPRODUCIBLE IN ASCII], 'it is expedient', [TEXT NOT REPRODUCIBLE IN ASCII], 'it is fitting', and [TEXT NOT REPRODUCIBLE IN ASCII], 'it is necessary', as well as those of verbs such as [[TEXT NOT REPRODUCIBLE IN ASCII], 'I become', [TEXT NOT REPRODUCIBLE IN ASCII], 'I follow', [TEXT NOT REPRODUCIBLE IN ASCII], 'I have', and [TEXT NOT REPRODUCIBLE IN ASCII], 'I speak', 'non conservano traccia della natura verbale' (13) and should be considered lexicalised. Evans (2001:231-32) similarly mentions the existence of 'fully adjectivised' participles in the Greek Pentateuch, as for example the present participle [TEXT NOT REPRODUCIBLE IN ASCII], 'alive' (Gen. 9.3) and the perfect participles [TEXT NOT REPRODUCIBLE IN ASCII], 'stolen' (Gen. 30.33) and [TEXT NOT REPRODUCIBLE IN ASCII], 'fully shaped' (Exod. 21.23) (on the combination of [TEXT NOT REPRODUCIBLE IN ASCII] with an 'adjectivised' participle cf. also Aerts 1965:12-17; Amenta 2003:30-35). Critical voices have also been raised, however. The most important contribution is that of Porter (1989:454), according to whom the fact that a participle functions like an adjective does not necessarily mean that it categorially is an adjective, stressing that even in these cases the participle asserts its verbal aspect (Porter refers to Gildersleeve 1980 [1900]:122, who, while recognising the adjectival character of the present and perfect participle, still treats them under the heading of 'periphrastic tenses'). Kahn (1973:135) similarly notes that, as long as the participle is recognisably connected with finite verb forms from the same stem, it does not lose its verbal nature completely.

In what follows, I discuss the criteria used in the secondary literature to argue for adjectivisation of the participle (concentrating on the present participle, since the other types of adjectival periphrasis have received little or no attention so far), and show that what is argued for by Bjorck and others is far from evident ([section] 2.1). Moreover, it cannot account for adjectival periphrasis with the perfect and aorist participle ([section] 2.2). I thereafter propose a Cognitively-inspired alternative ([section] 2.3, [section] 2.4), which starts from the insight that the participle (both in general and specifically as used in adjectival periphrasis) is a non-prototypical category (along the lines of Pompei 2006). Finally, I reconsider the relationship between 'adjectival' and 'verbal' periphrasis, and the importance of the criterion of 'conceptual integration' for this discussion ([section] 3).

2. The categorial status of the 'adjectival' participle

2.1 Complete adjectivisation?

In what follows, I briefly discuss the formal criteria that have been proposed in the secondary literature for adjectivisation of the present participle, paying particular attention to possible differences between Daueradjektivierungen and Gelegenheitsadjektivierungen (the former of which most scholars consider adjectivised). An overview of these criteria is given in Table I. Phonological criteria occupy a special position, because--though constituting a strong argument in favour of adjectivisation--evidence is scanty in Ancient Greek. Some exceptional examples, cited by Stahl (1907:681) are [TEXT NOT REPRODUCIBLE IN ASCII], 'glad', [TEXT NOT REPRODUCIBLE IN ASCII], 'readily', and [TEXT NOT REPRODUCIBLE IN ASCII], 'fair' (of a breeze). As such, I concentrate here on those criteria which analyse whether the present participle occurs in the same morpho-syntactic environments as 'true' adjectives.

Bjorck (1940:17) has drawn attention to the fact that the participles that occur in adjectival periphrasis can often be used in the comparative and superlative degrees, and/or as adverbs. (14) As Kahn (1973:136) notes, especially the latter seems a fruitful criterion. Kiihner & Blass (1983 [1892]:300) cite a large number of adverbs built on (present) participles, many of which belong to Bjorck's class of Daueradjektivierungen (cf. also the overview in Bjorck 1940:18-19). Examples are [TEXT NOT REPRODUCIBLE IN ASCII], 'abundantly', [TEXT NOT REPRODUCIBLE IN ASCII], 'differently', [TEXT NOT REPRODUCIBLE IN ASCII], 'profitably', [TEXT NOT REPRODUCIBLE IN ASCII], 'fitly', [TEXT NOT REPRODUCIBLE IN ASCII], 'suitably', and [TEXT NOT REPRODUCIBLE IN ASCII], 'expedient'. There are also some attested adverbs of participles which do not belong to this class (on the basis of frequency), e.g. [TEXT NOT REPRODUCIBLE IN ASCII], 'with admiration', [TEXT NOT REPRODUCIBLE IN ASCII], 'gladly', and [TEXT NOT REPRODUCIBLE IN ASCII], 'boldly', which are also based on lexically stative predicates. Perhaps unsurprisingly, there are no attested adverbs with lexically dynamic predicates (which could also be used for adjectival periphrasis, though less often).

Most criteria are of a syntactic nature. From a methodological point of view, such criteria are perhaps preferable, because they are more directly related to the use of the participle when combined with [TEXT NOT REPRODUCIBLE IN ASCII]. I have added 'frequent use in attributive position' and 'combination with adverbs of degree' in the overview, but one could argue that these are not valid criteria (at least not in the way they have been used in the literature). Frequent use in attributive position is mentioned by Aerts (1965:15-16) among others. Kahn (1973:136) criticises this criterion because, theoretically, every participle can be used in an attributive position, and frequency is a matter of degree. Indeed, it would seem that at best this criterion can give an indication of adjectivisation. The combination with adverbs of degree was proposed by Karleen (1980:120) (see Bjorck 1940:15 for the addition of adverbs in general), after the example of English. The rationale behind this criterion would be that adjectives are less complex in terms of defining sub-features (especially in comparison with nouns) and thus more suited for gradability (cf. Hamann 1991:659: 'adjectives ... are marked for gradability'; on complexity, cf. Givon 2001: 50ff.; but see Thesleff 1954 for nouns and verbs serving as 'concepts of value'). (15) His proposal faces two major problems: firstly, Karleen does not base himself on attested examples, but tests the grammaticality of examples by adding the adverbs [TEXT NOT REPRODUCIBLE IN ASCII], 'very' himself (on the use of the latter, cf. Thesleff 1954:92-111). (16) Secondly, it has been argued that even in English the validity of this test is questionable. As Borer (1990:97) notes, 'the ability to be modified by very ... has nothing to do with their adjectival (or non-adjectival) nature, since a similar contrast is attested for the verbs from which they are derived' (for [TEXT NOT REPRODUCIBLE IN ASCII] intensifying verbs in Ancient Greek, cf. Thesleff 1954:95-102).

Scholars take the co-ordination of a participle with one or more adjectives (mainly by means of the conjunction [[TEXT NOT REPRODUCIBLE IN ASCII], 'and', or simply a comma) as a clear sign of adjectivisation, under the assumption that conjuncts are of the same category (Boyer 1984:168; Amenta 2003:32). In (2) for example, the lexically stative participle [TEXT NOT REPRODUCIBLE IN ASCII], 'needing', 'in need of' is coordinated with the true adjective cpoPepd, 'formidable':

(2) [TEXT NOT REPRODUCIBLE IN ASCII] (Dem. 19.294).

yes, these are formidable offences, and in need of the utmost vigilance and precaution (tr. Vince & Vince, slightly modified).

If [TEXT NOT REPRODUCIBLE IN ASCII] is fully adjectival, however, it is difficult to explain why it retains its argument structure, taking the argument [TEXT NOT REPRODUCIBLE IN ASCII], 'much vigilance and precaution'. In my opinion, such co-ordination indicates functional likeness at best (cf. Nunez-Pertejo 2003:145; Huddleston & Pullum 2006:201), which is also apparent from examples where an aorist participle is co-ordinated with a true adjective (see e.g. Soph. OT 90, where [TEXT NOT REPRODUCIBLE IN ASCII], 'bold', is co-ordinated with [TEXT NOT REPRODUCIBLE IN ASCII], 'fearing prematurely').

Loss of argument structure is a good indication of whether a participle is still felt to be related to a content verb. (17) Contrary to what one might expect, the adjectival participle keeps its argument structure in most cases (compare Kahn 1973:136), as illustrated in (3):

(3) [TEXT NOT REPRODUCIBLE IN ASCII], (Isoc. 15.117).

what importance (lit. ability) do these things have? (my translation).

This observation not only concerns Gelegenheitsadjektivierungen, but also Daueradjektivierungen, even those which have an attested adverb. Bjorck (1940:20) admits that one could take this as an indication of the 'verbal' nature of the participle, but he refutes the argument on the basis of the expressions [TEXT NOT REPRODUCIBLE IN ASCII], 'having sense' and [TEXT NOT REPRODUCIBLE IN ASCII], 'having a logic'. These expressions, consisting of the participle [??][chi][omega]v and an accusative object, both have an attested adverb ([TEXT NOT REPRODUCIBLE IN ASCII], 'sensibly' and [TEXT NOT REPRODUCIBLE IN ASCII], 'reasonably'), showing that the presence of an object does not prevent the adjectivisation/lexicalisation of the participle (Aerts 1965:13). While this is certainly a valid observation, such expressions are quite rare. To generalise on this basis that all present participles used for adjectival periphrasis are adjectivised, is contestable, to say the least.

I conclude with some observations on word order, which is often mentioned in research on other languages as a formal criterion (e.g. prenominal position in English). No scholar has used word order to argue for the adjectivisation of the present participle in Ancient Greek (neither will I, for that matter), as Ancient Greek has 'free' word order. Interestingly, however, both Ceglia (1998:29) and Amenta (2003:33) note that in the case of what they call verbal periphrasis the participle tends to be placed after the finite verb, while this is not (or much less) the case with adjectival periphrasis (in Classical Greek at least). While this observation does seem to have some validity (though only under certain circumstances, for example not in poetry or when the participle is co-ordinated with true adjectives), adjectival (and verbal) constructions of both syntactic types do occur (even with one and the same participle). Admittedly, there is a noticeable difference between Daueradjektivierungen and Gelegenheitsadjektivierungen: the former occur more often in pre-finite position than the latter (for further discussion, see [section] 2.3).

A second element concerns the position of the negation o[??]([sigma]) or [mu][??]. (18) With both Daueradjektivierungen and Gelegenheitsadjektivierungen, the negation can come before the participle and split the component parts of the construction, as illustrated in (4) (for an example with the Gelegenheitsadjektivierung [TEXT NOT REPRODUCIBLE IN ASCII], 'accepting'; cf. Pl. Hp. mai. 289e):

(4) [TEXT NOT REPRODUCIBLE IN ASCII](Dem. 4.38).

but if smooth words that are not fitting prove a curse in practice (tr. Vince).

This placement, which occurs predominantly with adjectival periphrasis, (19) may be taken as an indication of a lesser degree of conceptual integration (Ceglia 1998:26; cf. Moorhouse 1959:138-40 for the possibility of this position when [TEXT NOT REPRODUCIBLE IN ASCII] is combined with a true adjective). It is unclear, however, to what extent this also proves the supposed adjectivisation of the participle. In my view, a low degree of conceptual integration between the component parts of the periphrastic construction does not necessarily entail adjectivisation of the participle. In any case, this position of the negation is found only in a minority of the examples.

As may be clear, the above-mentioned criteria do not make a convincing case for the complete adjectivisation of the present participle (when combined with [TEXT NOT REPRODUCIBLE IN ASCII]). They do allow us to draw the following three conclusions:

(a) the present participle used in adjectival periphrasis (both Daueradjektivierungen and Gelegenheitsadjektivierungen) occurs in a number of typically 'adjectival' morpho-syntactic environments, e.g. with the negation between the component parts of the construction or co-ordinated with a true adjective;

(b) one can hardly generalise, however, that this participle is completely adjectivised--not even the Daueradjektivierungen--as indicated by examples where the participle retains its argument structure; and

(c) in general, constructions with an adjectival present participle do not form a homogeneous category. There are some notable differences (along the lines of the distinction made by Bjorck), for example, concerning adverbial formation or the position vis-a-vis the finite verb.

2.2. What about the perfect and aorist participle?

One of the weaknesses of previous research on the categorial status of the 'adjectival' participle is that it does not take into account either the perfect or the aorist participle (which, as mentioned above, could also be used for adjectival periphrasis). Few scholars would argue that the aorist participle is adjectivised. With regard to the perfect participle, it has been suggested that it is 'strongly adjectival' (Aerts 1965:13), but in general it is not considered adjectivised. (20) It is important to note, though, that the present and the perfect participle have a lot in common. From a formal point of view, the perfect participle shares a number of characteristics with the present participle. Adverbial formation, for example, is attested with the perfect participle. Some examples are: [TEXT NOT REPRODUCIBLE IN ASCII], 'timorously', [TEXT NOT REPRODUCIBLE IN ASCII], 'with due caution', [TEXT NOT REPRODUCIBLE IN ASCII], 'tumultuously', and [TEXT NOT REPRODUCIBLE IN ASCII], 'in orderly manner' (Kiihner & Blass 1983 [1892]:300). Furthermore, the perfect participle is also found co-ordinated with adjectives, as in (5) (cf. Gildersleeve 1980 [1900]:124):

(5) [TEXT NOT REPRODUCIBLE IN ASCII] (Thuc. 6.49.1).

while they (the Syracusans) are still unprepared, and very much panic-stricken (my translation).

Concerning word order, we may note that the perfect participle shows a tendency to occur in pre-finite position (cf. Gildersleeve 1980 [1900]: 123). There are also examples where the negation comes before the participle and splits the component parts of the construction, as in the Homeric example (6):

(6) [TEXT NOT REPRODUCIBLE IN ASCII] (Hom. Od. 2.61).

we will be found wretched and knowing nothing of valour (my translation).

As it would seem, the main reason why the perfect participle has not been considered adjectivised is that constructions with the perfect participle are to a large extent paradigmatically integrated, (21) as a consequence of which they should be qualified as instances of 'verbal periphrasis' (or, to be more precise, better instances; see Bentein 2011). As previous research has correlated the adjectivisation of the participle with non-periphrasticity, it would be difficult to consider the perfect participle adjectivised.

I will return to the relationship between what is called 'adjectival' and 'verbal' periphrasis in [section] 3. In the next section, I would like to propose an alternative account of the categorial status of the participle, which takes into account the three main conclusions made at the end of [section] 2.1, and at the same time is able to account for constructions with the aorist and the perfect participle.

2.3. An alternative account: the participle as a non-prototypical category

The alternative account presented in this section is based on recent functional-typological research in the field of linguistic categorisation, parts of speech in particular. As Pustet (2003:17) notes, at present there are two main approaches to the issue, namely the models of Givon (1979:320-23; 1984: ch. 3; 2001: ch. 2), setting out a 'time-stability' hypothesis, and Croft (1991; 2001), which is based on markedness theory. Combined, these two approaches form a powerful theoretical framework, providing an alternative to the long-standing problem of the hybrid nature of the participle (cf. Hopper & Thompson 1984; Pustet 2003:21-22). Connecting both approaches is the insight that categories are prototypically organised, which is one of the main tenets of the school of Cognitive Linguistics. In what Langacker (1987:16-17) calls the 'prototype model for categorization,' it is acknowledged that a given category may have central, 'prototypical' members, and more peripheral ones, and that the boundaries between categories are not always clear-cut.

As already mentioned, Croft's model is centred around the notion of morpho-syntactic markedness. Croft proposes an innovative matrix which separates the semantic categories of objects, properties and actions, from the pragmatic functions of argument, attribute and predicate, as shown in Table 2. Croft predicts that in any given language unmarked nouns will be used for reference to an object, unmarked adjectives for modification by a property and unmarked verbs for predication of an action. As such, he proposes to consider 'noun,' 'adjective' and 'verb' functional/typological prototypes. This means that the participle, whether used in the field of reference, modification or predication, constitutes a marked 'non-prototypical' category. (22)

Croft's model puts the parts-of-speech issue in a different perspective, by showing that participles are a non-prototypical category, and furthermore that one cannot simply assume that present participles and adjectives should be considered one and the same part of speech. Of course, this does not mean that a participle can never be 'adjectivised' in a conceptual sense (profiling a simplex, rather than a complex relationship, i.e. one which consists of distinct component states), (23) nor does it mean that we cannot further analyse the status of the different types of participle used for adjectival periphrasis. Concentrating on the former point in [section] 2.4, I would like to argue here that the participle, as a non-prototypical category, is itself prototypically organised, presenting as it were an inner continuum from noun-like to verb-like (cf. Pompei 2006:363). (24)

One of the main semantic factors structuring not only this inner continuum, but also driving 'the pronounced inclination of grammatical systems to produce structural patterns that comply with the scale NOUNS > ADJECTIVES > VERBS' (Pustet 2003:17), is Givon's time-stability parameter. More specifically, Givon argues that the semantic difference between the parts of speech 'noun,' 'verb' and 'adjective' can be analysed in terms of a single parameter, called time-stability (Givon 1979:320-23; 1984:51-56). (25) According to Givon, (prototypical) nouns are most time-stable, while (prototypical) verbs are least so. In this framework, adjectives are of intermediate time-stability. This gives us the scale represented in

Figure I:

Givon also recognises intermediate stages, both with regard to the morphologically unmarked expressions (contrast for example English 'rock' with 'child,' and 'hit' with 'know'), and the marked ones. In the context of the latter, Ross (1972:316) has similarly posited the following scale (what he calls a 'linear squish') for English: verb > present participle > perfect participle > passive participle > adjective > preposition > adjectival noun > noun.

In her recent cross-linguistic work on copularisation, Pustet (2003: ch. 3) has proposed a revision of Givon's single-parameter model by a four-parameter model, which I will follow here (for criticism of Givon's time-stability concept, see e.g. Hopper & Thompson 1984:705-06; Pustet 2003: 21). More specifically, Pustet recognises the following four parameters: (a) 'dynamicity,' (b) 'transience,' (c) 'transitivity' and (d) 'dependency.'

Let me give a brief description of these four concepts before I apply them to the Ancient Greek participle, as used in adjectival periphrasis. The dynamicity parameter corresponds to the semantic distinction between state vs. change of state, i.e. whether the component states of the process profiled by the participle are or are not homogeneous. (26) The transience parameter largely coincides with Givon's parameter of timestability. Pustet also refers to Langacker's concept of 'boundedness in time', which not only refers to the absence or presence of an endpoint, but also to that of a starting point. (27) In her opinion, this 'adds refinement to the time-stability hypothesis', but she notes that defining boundedness in time solely in terms of presence/absence of an endpoint yields slightly better predictions (Pustet 2003:111). She therefore calls this version of bounding in time 'transience'. The transitivity parameter relates to semantic valency (cf. also Croft 1991:63ff., distinguishing between valency value 1, 2 or 3). Contrary to traditional approaches, Pustet (2003:114) does not take into account the coding strategy for a secondary argument (English 'worth' and 'devoid of' are thus both considered transitive). The fourth and final parameter, dependency, refers to whether a given concept is conceptually autonomous, i.e. whether its existence or presence 'requires the existence or presence of another entity' (Croft 1991:62, defining valency as 'inherent relationality') or whether it does not ('house', for example, does not require the presence of another entity, but 'yellow' does, i.e. something which is yellow). Since it can be assumed that all participles under analysis are [+dependent], I will not further discuss this criterion here. (28)

In what follows, I will apply Pustet's parameters of 'dynamicity,' 'transitivity' and 'dependency' to the Ancient Greek participle (as used in adjectival periphrasis). The discussion will mainly concentrate on the perfect and the present participle, as there are virtually no examples attested with the aorist participle.

The dynamicity parameter

With regard to dynamicity, the adjectival present participle is most complex. Most present participles are formed with non-dynamic, stative predicates such as [TEXT NOT REPRODUCIBLE IN ASCII], 'I have' (e.g. Isoc. 15.117), [TEXT NOT REPRODUCIBLE IN ASCII], 'I live' (e.g. Aristoph. Thesm. 77) and [TEXT NOT REPRODUCIBLE IN ASCII], 'I am able to' (e.g. Dem. 10.3), but dynamic predicates can also be employed. Some examples are dvaSexo^ai, 'I take up' (e.g. Dem. 19.36), [TEXT NOT REPRODUCIBLE IN ASCII], 'I go' (Pl. Phd. 82a) and [TEXT NOT REPRODUCIBLE IN ASCII], 'I do' (e.g. Pl. Resp. 441e). Adjectival perfect participles are always based on highly dynamic ('telic') predicates, such as [TEXT NOT REPRODUCIBLE IN ASCII], 'I destroy' (e.g. Dem. 35.36), [TEXT NOT REPRODUCIBLE IN ASCII], 'I bring about' (e.g. Aesch. Pers. 260), and [TEXT NOT REPRODUCIBLE IN ASCII], 'I die' (e.g. Dem. 43.64), but here focus is on the final 'resultant' state, so there can be no question of dynamicity. As for the aorist participle, it is based either on dynamic predicates (e.g. with [TEXT NOT REPRODUCIBLE IN ASCII], 'I become' in Pl. Leg. 711c) or, if not, participial morphology explicitly indicates a change of state (e.g. with [TEXT NOT REPRODUCIBLE IN ASCII], 'I fear prematurely' in Soph. OT 90), this boundary providing a measure of heterogeneity.

The transience parameter

Concerning transience, a basic distinction must be made between the present and perfect participle on the one hand, which do not inherently (i.e. grammatically) profile any boundaries, and the aorist participle, which profiles the boundaries of the event. As for the present participle, we can again distinguish between those cases where a lexically stative predicate is used, and those where a lexically dynamic one is used. In the overview given below, I present an analysis (29) of the properties expressed by the resultative perfect participle and the present participle of lexically stative predicates (as attested in my corpus) (30) in terms of transience:31

[-transient]

* abstract state: [TEXT NOT REPRODUCIBLE IN ASCII], 'established'; [TEXT NOT REPRODUCIBLE IN ASCII], 'accomplished'; [TEXT NOT REPRODUCIBLE IN ASCII], 'ordained by law'; [TEXT NOT REPRODUCIBLE IN ASCII], 'fated'; [TEXT NOT REPRODUCIBLE IN ASCII], 'fulfilled'; [TEXT NOT REPRODUCIBLE IN ASCII], 'dictated'; [TEXT NOT REPRODUCIBLE IN ASCII], 'imperfectly done'; [TEXT NOT REPRODUCIBLE IN ASCII], 'determined'.

* body feature/bodily state: [TEXT NOT REPRODUCIBLE IN ASCII], 'destroyed'; [TEXT NOT REPRODUCIBLE IN ASCII], 'seeing'; [TEXT NOT REPRODUCIBLE IN ASCII], 'destroyed'; (32) [TEXT NOT REPRODUCIBLE IN ASCII], 'alive'; [TEXT NOT REPRODUCIBLE IN ASCII], 'hearing'; [TEXT NOT REPRODUCIBLE IN ASCII], 'dead'; [TEXT NOT REPRODUCIBLE IN ASCII] 'existent'. (33)

* colour: [TEXT NOT REPRODUCIBLE IN ASCII], 'coloured'. (34)

* (physical/metaphorical) distance: [TEXT NOT REPRODUCIBLE IN ASCII], 'distant'; [TEXT NOT REPRODUCIBLE IN ASCII], 'separated from'; [TEXT NOT REPRODUCIBLE IN ASCII], 'separated from'; [TEXT NOT REPRODUCIBLE IN ASCII], 'bordering on'; [TEXT NOT REPRODUCIBLE IN ASCII], 'removed from'; [TEXT NOT REPRODUCIBLE IN ASCII], 'jutting out'; [TEXT NOT REPRODUCIBLE IN ASCII], '(being) above, pre-eminent'.

* mental property: [TEXT NOT REPRODUCIBLE IN ASCII], 'knowing'; [TEXT NOT REPRODUCIBLE IN ASCII], 'knowing'; [TEXT NOT REPRODUCIBLE IN ASCII], 'sane'; [TEXT NOT REPRODUCIBLE IN ASCII], 'mad'; [TEXT NOT REPRODUCIBLE IN ASCII], 'wise'; [TEXT NOT REPRODUCIBLE IN ASCII], 'over-proud'.

* personality feature: [TEXT NOT REPRODUCIBLE IN ASCII], 'able'; [TEXT NOT REPRODUCIBLE IN ASCII], 'accustomed'; [TEXT NOT REPRODUCIBLE IN ASCII], 'puffed up'; [TEXT NOT REPRODUCIBLE IN ASCII], 'educated'; [TEXT NOT REPRODUCIBLE IN ASCII], 'by nature so or so'; [TEXT NOT REPRODUCIBLE IN ASCII], 'devoted to'. (35)

* physical property: [TEXT NOT REPRODUCIBLE IN ASCII], 'written (in/on)'; [TEXT NOT REPRODUCIBLE IN ASCII], 'split'; [TEXT NOT REPRODUCIBLE IN ASCII], 'founded'; [TEXT NOT REPRODUCIBLE IN ASCII], 'broken'; [TEXT NOT REPRODUCIBLE IN ASCII], 'constructed'; [TEXT NOT REPRODUCIBLE IN ASCII], 'dug', [TEXT NOT REPRODUCIBLE IN ASCII], 'perforated';[TEXT NOT REPRODUCIBLE IN ASCII], 'constructed' (of a road); [TEXT NOT REPRODUCIBLE IN ASCII], 'built'.

* social relation: [TEXT NOT REPRODUCIBLE IN ASCII], 'wedded'; [TEXT NOT REPRODUCIBLE IN ASCII], 'descendant'; [TEXT NOT REPRODUCIBLE IN ASCII], 'left in orphanhood'.

[[+ or -] transient]

* (physical/metaphorical) absence/presence: [TEXT NOT REPRODUCIBLE IN ASCII], 'absent'; [TEXT NOT REPRODUCIBLE IN ASCII], 'lacking'; [TEXT NOT REPRODUCIBLE IN ASCII], 'present in'; [TEXT NOT REPRODUCIBLE IN ASCII], 'lacking'; [TEXT NOT REPRODUCIBLE IN ASCII], 'being an object of thought'; [TEXT NOT REPRODUCIBLE IN ASCII], 'present'; [TEXT NOT REPRODUCIBLE IN ASCII], 'left over'; [TEXT NOT REPRODUCIBLE IN ASCII], '(being) at hand'; [TEXT NOT REPRODUCIBLE IN ASCII], 'left remaining'.

* bodily state: [TEXT NOT REPRODUCIBLE IN ASCII], 'set free'; [TEXT NOT REPRODUCIBLE IN ASCII], 'seized'; [TEXT NOT REPRODUCIBLE IN ASCII], 'captured'; [TEXT NOT REPRODUCIBLE IN ASCII], 'deserted'; [TEXT NOT REPRODUCIBLE IN ASCII], 'captured'.

* emotional state: [TEXT NOT REPRODUCIBLE IN ASCII], 'persuaded'; [TEXT NOT REPRODUCIBLE IN ASCII], 'supine'; [TEXT NOT REPRODUCIBLE IN ASCII], 'deceived'; [TEXT NOT REPRODUCIBLE IN ASCII], 'wanting'; [TEXT NOT REPRODUCIBLE IN ASCII], 'hating'.

[??] evaluative: [TEXT NOT REPRODUCIBLE IN ASCII], 'wrong'; [TEXT NOT REPRODUCIBLE IN ASCII], 'pleasing'; [TEXT NOT REPRODUCIBLE IN ASCII], 'sufficient'; [TEXT NOT REPRODUCIBLE IN ASCII], 'befitting'; [TEXT NOT REPRODUCIBLE IN ASCII], 'needing'; [TEXT NOT REPRODUCIBLE IN ASCII], 'necessary';[TEXT NOT REPRODUCIBLE IN ASCII], 'amazing'; [TEXT NOT REPRODUCIBLE IN ASCII], 'sufficient'; [TEXT NOT REPRODUCIBLE IN ASCII], 'erroneous'; [TEXT NOT REPRODUCIBLE IN ASCII], 'dear to'; [TEXT NOT REPRODUCIBLE IN ASCII], 'profitable'; [TEXT NOT REPRODUCIBLE IN ASCII], 'due'; [TEXT NOT REPRODUCIBLE IN ASCII], 'suiting'; [TEXT NOT REPRODUCIBLE IN ASCII], 'fitting'; [TEXT NOT REPRODUCIBLE IN ASCII], 'expedient'; [TEXT NOT REPRODUCIBLE IN ASCII], 'surpassing', 'incredible'; [TEXT NOT REPRODUCIBLE IN ASCII], 'necessary'.

* (physical/metaphorical) position: [TEXT NOT REPRODUCIBLE IN ASCII], 'leading', 'dominant'; [TEXT NOT REPRODUCIBLE IN ASCII], 'having primacy'; [TEXT NOT REPRODUCIBLE IN ASCII], 'appointed'; [TEXT NOT REPRODUCIBLE IN ASCII], 'poor'.

* possession: [TEXT NOT REPRODUCIBLE IN ASCII], 'in possession of; [TEXT NOT REPRODUCIBLE IN ASCII], 'equipped (so or so)'; [TEXT NOT REPRODUCIBLE IN ASCII], 'equipped (so or so)'; [??][chi][omega]v, 'having'; [TEXT NOT REPRODUCIBLE IN ASCII], 'occupying'; [TEXT NOT REPRODUCIBLE IN ASCII], 'containing'.

* resemblance: [TEXT NOT REPRODUCIBLE IN ASCII], 'not in accordance with', 'inconsistent'; [TEXT NOT REPRODUCIBLE IN ASCII], 'different from'; [TEXT NOT REPRODUCIBLE IN ASCII], 'similar to'; [TEXT NOT REPRODUCIBLE IN ASCII], 'in accordance with'; [TEXT NOT REPRODUCIBLE IN ASCII], 'in accordance with'; (36) [TEXT NOT REPRODUCIBLE IN ASCII], 'in accordance with'.

[??] social behaviour: [TEXT NOT REPRODUCIBLE IN ASCII], 'having regard for'; [TEXT NOT REPRODUCIBLE IN ASCII], 'having mercy'; [TEXT NOT REPRODUCIBLE IN ASCII], '(being) of one mind'; [TEXT NOT REPRODUCIBLE IN ASCII], 'putting first in honour'.

[+transient]

* bodily state: [TEXT NOT REPRODUCIBLE IN ASCII], 'being at one's height'; [TEXT NOT REPRODUCIBLE IN ASCII], 'positioned against'; [TEXT NOT REPRODUCIBLE IN ASCII], 'cut off from'; [TEXT NOT REPRODUCIBLE IN ASCII], 'bound'; [TEXT NOT REPRODUCIBLE IN ASCII], 'mixed up'; [TEXT NOT REPRODUCIBLE IN ASCII], '(thoroughly) trained'; [TEXT NOT REPRODUCIBLE IN ASCII], 'strong'; [TEXT NOT REPRODUCIBLE IN ASCII], 'gathered'; [TEXT NOT REPRODUCIBLE IN ASCII], 'assembled'; [TEXT NOT REPRODUCIBLE IN ASCII], 'hobbled'; [sigma][pi]a[rho][gamma][??]v, 'bursting'; [TEXT NOT REPRODUCIBLE IN ASCII], 'gathered together'; [TEXT NOT REPRODUCIBLE IN ASCII], 'gathered together'; [TEXT NOT REPRODUCIBLE IN ASCII], 'vigorous'; [TEXT NOT REPRODUCIBLE IN ASCII], 'posted'.

* emotional state: [TEXT NOT REPRODUCIBLE IN ASCII], 'utterly grieved'; [TEXT NOT REPRODUCIBLE IN ASCII], 'desperate'; [TEXT NOT REPRODUCIBLE IN ASCII], 'relaxed'; [TEXT NOT REPRODUCIBLE IN ASCII], 'panic-stricken'; [TEXT NOT REPRODUCIBLE IN ASCII], 'eager'; [TEXT NOT REPRODUCIBLE IN ASCII], 'exalted'; [TEXT NOT REPRODUCIBLE IN ASCII], 'angry'; [TEXT NOT REPRODUCIBLE IN ASCII], 'thrown in disorder'; [TEXT NOT REPRODUCIBLE IN ASCII], 'steeled (for war)'.

* physical property: [TEXT NOT REPRODUCIBLE IN ASCII], 'opened'; [TEXT NOT REPRODUCIBLE IN ASCII], 'full'; [TEXT NOT REPRODUCIBLE IN ASCII], 'painted'; [TEXT NOT REPRODUCIBLE IN ASCII], 'closed'; [TEXT NOT REPRODUCIBLE IN ASCII], 'hidden'; [TEXT NOT REPRODUCIBLE IN ASCII], 'visual'; [TEXT NOT REPRODUCIBLE IN ASCII].

On the basis of this overview we can draw the following conclusions:

(a) in general, the perfect/present participle is hardly ever used for prototypical adjectival properties, such as 'size', 'colour', 'auditory qualities', 'shape', 'taste' or 'tactility' (Givon 2001:82; cf. also Dixon 1977:63 and Stassen 1997:168-69);

(b) the perfect participle is most diverse: it has most 'types' (while both categories have a more or less equal number of tokens);

(c) both perfect and present participles can express properties that are [-transient], [[+ or -] transient] or [+transient]. However, it would seem that the perfect participle is especially typical for [-transient] and to some extent [+transient] properties, while the present participle is most frequent in the area of [[+ or -] transient].

The transitivity parameter

Generally speaking, adjectival participles resemble 'true' adjectives in that they have valency value 1, i.e. they express a property of the subject (as there are only a few examples with the aorist participle, it is hard to draw any conclusion; there is one example where the participle has valency value 2). This is especially true for adjectival perfect participles, which in combination with etui denote a resultant state (cf. Haspelmath & Muller-Bardey 2004, who characterise the (object-oriented) resultative as a valency-decreasing, i.e. agent-removing, category). As we have already seen above, the present participle is more complex in that a variety of examples are attested where the participle has an argument structure, i.e. it takes an argument next to the subject. This concerns both cases with the participle formed on the basis of a lexically dynamic predicate and those with a lexically stative one. With regard to the former type, verbs with valency value 2 are often used, e.g. [TEXT NOT REPRODUCIBLE IN ASCII], 'I increase' (Dem. 3.33), and occasionally even a verb with valency value 3 (which is rare; cf. Pustet 2003:114), namely [TEXT NOT REPRODUCIBLE IN ASCII], 'I give' (as in Eur. IT 721-22). As for the latter type, particular attention must be paid to verbs expressing possession, especially the bivalent verb [??][chi][omega], 'I have,' which is used with an exceptionally high number of different accusative objects, such as ataxuvnv, 'shame' (Eur. Suppl. 767), [TEXT NOT REPRODUCIBLE IN ASCII], 'a weak spot' (Pl. Soph. 267e), [TEXT NOT REPRODUCIBLE IN ASCII], 'an ability' (Isoc. 15.117), [TEXT NOT REPRODUCIBLE IN ASCII], 'an appearance' (Eur. Bacch. 471), [TEXT NOT REPRODUCIBLE IN ASCII], 'a rationale, weight' (Dem. 20.8), [TEXT NOT REPRODUCIBLE IN ASCII], 'a copy' (Pl. Leg. 713b), [TEXT NOT REPRODUCIBLE IN ASCII], 'a nature' (Dem. 2.26) etc. Verbs denoting what I call an 'emotional' state also often take an accusative object, e.g. [TEXT NOT REPRODUCIBLE IN ASCII], 'hating our city' (Dem. 19.312).

An overall summary of my findings is given in Table 3:

In this table we can see that the perfect participle is non-dynamic, grammatically non-transient (though I have provided a more complex analysis of transience), and intransitive. The present participle is more complex, in that it can be either dynamic or non-dynamic and both transitive and intransitive. Grammatically, it is non-transient, but again I have suggested a more complex analysis of transience, comparing the adjectival present and perfect participles. It is difficult to draw firm conclusions with regard to the aorist participle, since there are very few examples, but as far as one can tell, it is dynamic, grammatically transient and possibly transitive.

These findings, in turn, allow us to locate the adjectival participle on a noun-to-verb continuum, as shown in Figure 2: on the left noun-like side, we find the perfect participle, and on the right verb-like side, the aorist participle. In between these two categories, I locate the present participle, whereby a further distinction can be made between cases where the participle is formed on the basis of lexically stative content verbs, versus those where it is formed on the basis of lexically dynamic ones. We need not assume a strict dividing line between these three groups: especially between the perfect and the present participle, there seems to be considerable overlap (compare Ceglia 1998:38-39, who notes competition between the two in Post-classical Greek).

As for the present participle, there might be reason to believe that its internal organisation is not bifold, that is, divided between constructions with the participle based on a lexically stative predicate versus those based on a lexically dynamic one, but threefold. Long ago, Alexander (1883:23940), discussing [TEXT NOT REPRODUCIBLE IN ASCII] with ('adjectival') present participle in Classical Greek, made the following proposal:

   There are, in the first place, certain verbs in Greek expressing
   the manifestation of a quality while at the same time no adjective
   exists for the expression of the quality itself ... But, in the
   second place, besides these cases, the language often felt the need
   of embodying the conception of a verb as a quality, especially when
   the signification of the verb naturally lends itself to an
   adjective use ... Thirdly, we group together such participles as
   having nothing in their signification which would lead them to be
   used as adjectives, but are nevertheless occasionally so used in
   these periphrases.

This comes down to a threefold distinction between present participles based on (a) verbs with lexicalised predication of qualities, e.g. [TEXT NOT REPRODUCIBLE IN ASCII], 'I am wise' (= 'group 1'); (b) other stative predicates, e.g. [TEXT NOT REPRODUCIBLE IN ASCII], 'I have' (= 'group 2'); and (c) lexically dynamic predicates, e.g. [TEXT NOT REPRODUCIBLE IN ASCII], 'I give' (= 'group 3'). Alexander (1883:303) furthermore noted a difference in word order between participles of the first group and the second group: the former occur much more often before the finite verb than the latter (cf. our earlier observations in [section] 2.1). I have applied Alexander's division to the entire corpus, and analysed the word order of each group. The results of this analysis can be found in Table 4

As can be seen, there indeed seems to be a quite noticeable difference in word order between these three groups. Obviously, this is not the place for an extensive discussion of the multiple factors influencing word order, but I believe that one of the main reasons why participles of the first group are more often used in pre-finite position, is that participles of this type are less complex or, in other words, that the adjectival property they denote is more salient; generalising, it could be said that the preposed participle is focal (in other words, we have constituent focus), while the postposed participle forms part of a broad focus domain together with [TEXT NOT REPRODUCIBLE IN ASCII].

Perhaps it could be argued that Alexander's proposal partly goes against the predictions of Pustet's second, 'transience' parameter, as a participle such as [TEXT NOT REPRODUCIBLE IN ASCII], 'angry' (which would belong to the first group) is more transient than a participle such as [TEXT NOT REPRODUCIBLE IN ASCII], 'having' (which would belong to the second), and hence would have to be considered more verb-like. However, we must keep two elements in mind:

(a) there does not seem to be any reason why these three parameters themselves could not be prototypically organized (cf. Winters 1990:299: 'the features of a prototypical syntactic construction can be viewed as forming a radial category of their own'), whereby transience, or at least the proposed classification of less and more transient adjectival present participles, is of lesser importance than we have thus far assumed (in the case of [TEXT NOT REPRODUCIBLE IN ASCII], the fact that this predicate takes an accusative object makes it inherently verb-like);

(b) perhaps more importantly, we must also take into account frequency. To quote Bybee (2010:79): 'graded category membership can come about in an exemplar model by the interaction of two categorization dimensions--similarity and frequency ... it seems that frequency of occurrence might significantly influence categorization in language.' Indeed, many participles of the first group occur much more frequently than those of the second group and third group, particularly the (mostly neuter) participles of the impersonal verbs [TEXT NOT REPRODUCIBLE IN ASCII]. Significantly, these participles occur even more often in pre-finite position, as shown in Table 5 (especially note the case of [TEXT NOT REPRODUCIBLE IN ASCII]:

2.4 The usage-based model: analysability

In conclusion of [section] 2, I would like to note that the alternative proposal I have presented does not exclude the possibility that some participles conceptually become adjective-like, i.e. come to profile a simplex relationship (one which does not consist of multiple component states). In Langacker's (2000) usage-based model such a process is characterised in terms of a decrease in 'analysability,' which is defined as 'the extent to which speakers are cognizant of the presence and the semantic contribution of component symbolic elements', which in our case would be the verbal stem and the participial morphology. As Pustet (2003:158) notes, this is a process that is well attested in the Indo-European languages: the English forms 'astonished,' 'scared' and 'worried' would be currently in the 'transition zone' (and, as some colleagues of mine have noted, the same would be true for a number of Modern Greek participles).

Especially for those expressions at the left boundary of the continuum, namely present participles of lexically stative predicates (especially predicates with lexicalised predication of qualities), and resultative perfect participles, which are most adjective-like, a decrease in analysability, whereby the participle acquires 'unit status' (cf. Bybee 2010 for 'chunking'), would not come as a surprise, especially in the case of frequently used participles such as npenov. Adverbial formation of present and perfect participles indicates that this is a real phenomenon, though it does not mean that such participles lose their verbal (processual) nature in all cases. As Bybee (2010:47) notes:

   When a speaker or listener processes a morphologically complex
   word, the extent to which the component parts are activated can
   vary. On one extreme, the complex word could be based directly on
   its component morphemes, especially if it is unfamiliar, thereby
   activating the parts completely. Or it might be possible to access
   the complex word directly as a single unit, while still activating
   the morphemes that make it up. On the other extreme, the complex
   word could be accessed without activating the component morphemes
   at all, which would be the case if analysability has been lost for
   that word. Given that activation is gradient and associations among
   parts of words in a network model are also gradient, there are many
   degrees of activation that are possible.

As a result, it is often unclear how a particular example should be interpreted, and what semantic/pragmatic differences this brings about. Consider examples (7) and (8): to what extent are the component parts activated in the forms [TEXT NOT REPRODUCIBLE IN ASCII], 'seeing' and [TEXT NOT REPRODUCIBLE IN ASCII], 'plain- fashioned' (lit. 'tight-drawn')?

(7) [TEXT NOT REPRODUCIBLE IN ASCII] (Soph. OT 747).

I have grievous misgivings that the prophet may have sight (my translation).

(8) [TEXT NOT REPRODUCIBLE IN ASCII] (Pl. Grg. 511d).

and at the same time it (the art of piloting) is plain-fashioned and orderly (my translation).

It is very difficult to answer this question. In both cases one could argue that this is not the case: in (7) [TEXT NOT REPRODUCIBLE IN ASCII] is explicitly opposed to [TEXT NOT REPRODUCIBLE IN ASCII], 'blind', while in (8) [TEXT NOT REPRODUCIBLE IN ASCII], has an extended, metaphorical meaning. Any argument, however, necessarily remains tentative (compare it to a linguist who has to decide whether modern-day printer is still felt to be analysable into its component parts print and -er; cf. Langacker 2000:47). Moreover, there is a real possibility that this was not clear to the author or audience of these writings either; in any case, the language user does not have to make a choice between these options, they are not mutually exclusive. As such, I believe Porter (1989:454) is right in criticising the communis opinio that a participle such as [pi][rho][??][pi]iv is necessarily adjectivised in every single instance.

3. Conclusion: the relationship between 'adjectival' and 'verbal' periphrasis

Up until now, the relationship between 'adjectival' and 'verbal' periphrasis has generally been considered in terms of a binary opposition, on the basis of the fact that with the former the participle would be categorially an adjective and as such there would be nothing 'verbal' about the participle and its combination with the verb [TEXT NOT REPRODUCIBLE IN ASCII]. As I have shown, this view entails a number of difficulties, most importantly (a) that the criteria adduced for this statement are either theoretically problematic or insufficient, and (b) that it does not account for the perfect and aorist participle (especially the latter can hardly be considered 'adjectivised').

I have proposed an alternative account which is centred around the insight that the participle is a non-prototypical category, locating the adjectival participle on an intracategorial continuum with an adjective-like and a verb-like side (including the perfect and aorist participle). (37) I concluded that the present participle is most complex in terms of this continuum, and that the perfect and aorist participles should be located towards the left 'noun-like' and the right 'verb-like' side of the continuum respectively. Connecting these findings to what I have discussed in my earlier work (Bentein 2013a), I argue that while all types of adjectival periphrasis resemble each other in that only a single, represen-tative component state of [TEXT NOT REPRODUCIBLE IN ASCII] is elaborated, and as such all involve a low degree of conceptual integration, this conceptual integration will be the lowest with the participles that are located on the left side of the noun-to-verb continuum, that is, resultative perfect participles and present participles of lexically stative verbs.

As I have recently discussed the identification of verbal periphrasis in Ancient Greek (Bentein 2011), it is worth reconsidering the interrelationship between the two concepts here. In brief, I have argued that verbal periphrasis can be characterised as a prototypically organised category. After analysing six criteria proposed in the secondary literature, I conclude that the parameters of 'conceptual integration', 'syntactic contiguity' and 'paradigmatic integration' are the most suitable to distinguish between constructions with regard to their periphrasticity. On this basis, I distinguish between four main groups of 'periphrastic' constructions, which can be ordered from more prototypical to more peripheral (or, from a diachronic point of view, from most grammaticalised to least so). In comparison with this view, it would seem that adjectival periphrasis can equally be considered a prototypically organised category, in that it is possible to locate the adjectival participle on a noun-to-verb continuum and that we can distinguish between different degrees of conceptual integration. Overall, however, I would argue that the category of adjectival periphrasis is much less complex than that of verbal periphrasis, in that it mainly involves semantics (with adjectival periphrasis, for example, paradigmatic criteria play no role whatsoever). As a result, one and the same construction can have an 'adjectival' character while at the same time constitute a good example of verbal periphrasis: comparing for example [TEXT NOT REPRODUCIBLE IN ASCII] with present and perfect participle (as used in adjectival periphrasis), it becomes clear that while both score low on the parameter of conceptual integration (with the perfect participle arguably the lowest), the latter is much more often syntactically contiguous (72% of the cases versus 36% with the present participle), next to the fact that it is paradigmatically integrated, that is, filling a gap in the paradigm (see footnote 21). As a result, [TEXT NOT REPRODUCIBLE IN ASCII] with perfect participle constitutes a more prototypical example of verbal periphrasis than [TEXT NOT REPRODUCIBLE IN ASCII] with present participle.

Bibliography

Aerts, W.J. 1965. Periphrastica. Amsterdam: Hakkert (diss. Amsterdam).

Alexander, WJ. 1883. 'Participial periphrases in Attic prose.' AJPh 6:291-308.

Amenta, L. 2003. Perifrasi aspettuali in greco e in latino. Milano: Franco Angeli.

Barbelenet, D. 1913. De la phrase a verbe ettre dans lionien d' Herodote. Paris: H. Champion.

Bentein, K. 2011. 'Towards the identification of verbal periphrasis in Ancient Greek: a prototype analysis.' A Class 54:1-25.

Bentein, K. 2013a. 'Adjectival periphrasis in Ancient Greek: a cognitive account.' Onomazein 27:15-34.

Bentein, K. 2013b. 'Transitivity, ecology and the emergence of verbal periphrasis in Ancient Greek.' CPh 108:286-313.

Bjorck, G. 1940. [TEXT NOT REPRODUCIBLE IN ASCII]. Die periphrastischen Konstruktionen im Griechischen. Uppsala: Almqvist & Wiksell (diss. Uppsala).

Borer, H. 1990. 'V + ing: it walks like an adjective, it talks like an adjective.' Linguistic Inquiry 21:95-103.

Boyer, J.L. 1984. 'The classification of participles: a statistical study.' GTJ 5:163-179.

Bybee, J.L. 2010. Language, Usage and Cognition. Cambridge: CUP.

Ceglia, L. 1998. 'L'evoluzione della costruzione perifrastica verbale nel greco del Nuovo Testamento.' AGI 83:20-44.

Crespo, E.; Conti, L. & Maquieira, H. 2003. Sintaxis del Griego Cldsico. Madrid: Gredos.

Croft, W. 1991. Syntactic Categories and Grammatical Relations. Chicago: University of Chicago Press.

Croft, W. 2001. Radical Construction Grammar. Oxford: OUP.

Dietrich, W. 1973. Der periphrastische Verbalaspekt in den romanischen Sprachen. Tubingen: Niemeyer.

Dixon, R.M. 1977. 'Where have all the adjectives gone?' Studies in Language 1:19-80.

Evans, T. 2001. Verbal Syntax in the Greek Pentateuch. Oxford: OUP.

Gildersleeve, B.L. 1980 [1900]. Syntax of Classical Greek. Groningen: Bouma.

Givon, T. 1979. On Understanding Grammar. New York: Academic Press.

Givon, T. 1984. Syntax. Vol. 1. Amsterdam: Benjamins.

Givon, T. 2001. Syntax. Vol. 1. Amsterdam: Benjamins.

Hamann, C. 1991. 'Adjectival semantics.' In A. von Stechow et al. (edd.), Semantics: An International Handbook of Contemporary Research, 657-673. Berlin: de Gruyter.

Haspelmath, M. & Muller-Bardey, T. 2004. 'Valency change.' In G. Booij et al. (edd.), Morphology: A Handbook on Inflection and Word Formation, 2:1130-1145. Berlin: de Gruyter.

Hopper, P.J. & Thompson, S.A. 1980. 'Transitivity in grammar and discourse.' Language 56:251-299.

Hopper, P.J. & Thompson, S.A. 1984. 'The discourse basis for lexical categories in universal grammar.' Language 60:703-752.

Huddleston, R. & Pullum, G.K. 2006. 'Coordination and subordination.' In B. Aarts (ed.), The Handbook of English Linguistics, 198-219. Oxford: Basil Blackwell.

Kahn, C.H. 1973. The Verb Be and Its Synonyms. Dordrecht: Reidel.

Karleen, P.S. 1980. The Syntax of the Participle in the Greek New Testament. Ph.D. thesis, University of Pennsylvania.

Kuhner, R. & Blass, F. 1983 [1892] (3rd edition). Ausfuhrliche Grammatik der griechischen Sprache. Erster Teil: Elementar--und Formenlehre. Hannover: Verlag Hahnsche Buchhandlung.

Langacker, R.W. 1987. Foundations of Cognitive Grammar. Stanford: Stanford University Press.

Langacker, R.W. 2000. 'A dynamic usage-based model.' In M. Barlow & S. Kemmer (edd.), Usage-based Models of Language, 1-63. Stanford: CSLI Publications.

Langacker, R.W. 2005. 'Integration, grammaticization, and constructional meaning.' In M. Fried & H.C. Boas (edd.), Grammatical Constructions: Back to the Roots, 157-190. Amsterdam: Benjamins.

Langacker, R.W. 2008. Cognitive Grammar: A Basic Introduction. Oxford: OUP.

Liddell, H.G. & Scott, R. 1968 (9th edition). A Greek-English Lexicon. Oxford: Clarendon Press.

Moorhouse, A.C. 1959. Studies in the Greek Negatives. Cardiff: University of Wales Press.

Nunez-Pertejo, P. 2003. 'Adjectival participles or present participles? On the classification of some dubious examples from the Helsinki Corpus.' In J.L. Bueno Alonso et al (edd.), Nothing but Papers, My Lord. Studies in Early Modern English Language and Literature (SEDERI 13), 141-153. Vigo: Universidade de Vigo.

Pompei, A. 2006. 'Participles as a non prototypical word class.' In E. Crespo et al. (edd.), Word Classes and Related Topics in Ancient Greek, 361-388. Louvain-la-Neuve: Peeters.

Porter, S.E. 1989. Verbal Aspect in the Greek of the New Testament. New York: Peter Lang (diss. Sheffield).

Pustet, R. 2003. Copulas: Universals in the Categorization of the Lexicon. Oxford: OUP.

Rosen, H.B. 1957. 'Die "zweiten" Tempora des Griechischen: zum Pradikatsausdruck beim griechischen Verbum.' MH 14, 133-154.

Ross, J.R. 1972. 'The category squish: endstation Hauptwort.' In Papers from the Eighth Regional Meeting of the Chicago Linguistic Society, 316-328. Chicago: University of Chicago Press.

Stahl, J.M. 1907. Kritisch-historische Syntax des griechischen Verbums der klassischen Zeit. Heidelberg: Winter.

Stassen, L. 1997. Intransitive Predication. Oxford: OUP.

Thesleff, H. 1954. Studies on Intensification in Early and Classical Greek. Ph.D. thesis: University of Helsinki.

Winters, M.E. 1990. 'Toward a theory of syntactic prototypes.' In S.L. Tsohatzidis (ed.), Meanings and Prototypes, 285-230. London: Routledge.

Klaas Bentein

Ghent University

Klaas.Bentein@UGent.Be

* I would like to thank Rutger Allan, Wolfgang de Melo, Mark Janse, as well as two anonymous referees of Acta Classica, for their helpful comments on earlier versions of this article. Parts of the article were presented at the Sixth Cambridge Postgraduate Conference in Language Research (Cambridge, 7-8 December 2010), Hellenistendag (Nijmegen, 28 January 2011) and the international conference Linguistics and Classical Languages (Rome, 17-19 February 2011). My work was funded by the Special Research Fund of Ghent University (grant no. 01D23409).

(1) This is not to say that the participle has entirely lost its verbal properties, as I will be arguing below.

(2) Constructions of [TEXT NOT REPRODUCIBLE IN ASCII] with aorist participle occur infrequently in Archaic and Classical Greek. The verb [??][chi][omega] is much more frequently combined with the aorist participle, as in [TEXT NOT REPRODUCIBLE IN ASCII], 'I have announced' (Soph. Ant. 192), though in these constructions the participle cannot be considered 'adjectival', i.e. it does not refer to a property.

(3) My research is based on an extensive survey of the specialised literature, most notably Alexander 1883; Barbelenet 1913; Bjorck 1940; Rosen 1957; Aerts 1965 and Dietrich 1973. Taken together, the evidence collected from these studies comprises a large part of Ancient Greek literature, both prose and poetry. They contain a total number of 418 examples for the present participle and 397 for the perfect participle; as we will see, there are almost no examples with the aorist participle.

(4) In Cognitive Grammar (see e.g. Langacker 1987; 2008), it is said that verbs (both finite verbs and participles) are processual, as they focus on the evolution of a process through time. They highlight multiple component states. Typically, when a verb such as [TEXT NOT REPRODUCIBLE IN ASCII] integrates with a participle, as in [TEXT NOT REPRODUCIBLE IN ASCII], 'he was teaching' in (1), there is a high degree of conceptual integration, in that the component states highlighted by [TEXT NOT REPRODUCIBLE IN ASCII] are mapped onto each other, component state by component state. Adjectives, on the other hand, are said to be non-processual, and highlight only a single component state (or more correctly, a simplex relationship). As such, when [TEXT NOT REPRODUCIBLE IN ASCII] integrates with an adjective, it suffices that a single, representative component state of [TEXT NOT REPRODUCIBLE IN ASCII] is elaborated. In Bentein 2013a, I argue that the integration of [TEXT NOT REPRODUCIBLE IN ASCII] with an 'adjectival' participle resembles that of [TEXT NOT REPRODUCIBLE IN ASCII] with a regular adjective, without, however, being identical to it.

(5) 'Now it becomes clear that the notion of periphrasis is artificial here and only has legitimacy in the provisional systematics.'

(6) 'Each combination of the copula with a predicative participle which is such that it can be replaced by a synthetic verbal form without a change in meaning.'

(7) 'A word denoting a property is related predicatively to the subject, rather than that "the predicate in the participle is integrated conceptually with the copula" (Rehdantz' Indices s.v. Participium).'

(8) Cf. Aerts 1965:12: 'the term adjectival periphrasis implies the combination of a copula with a participle that has been completely adjectivized.'

(9) Conceptually, the participle would not profile a complex relationship, i.e. one consisting of multiple component states, but rather a simplex one, similarly to a regular adjective. Compare note 4.

(10) Note that in this example a lexically dynamic verb is used [TEXT NOT REPRODUCIBLE IN ASCII], 'to do evil', 'to injure'). Rather than denoting an ongoing action, however, the construction characterises the subject.

(11) 'A fundamental distinction, so that less entrenched periphrases ('Gelegenheit-speriphrasen') would be somewhat more "verbal", is out of the question.'

(12) Aerts 1965:16-17 also seems to take into account Bjorck's Gelegenheitsadjektivierungen, but it is not clear to me from his discussion whether he also considers them adjectivised, which would be a logical conclusion from the fact that he defines adjectival periphrasis in terms of the complete adjectivisation of the participle (p. 12), and that he considers Gelegenheitsadjektivierungen instances of adjectival periphrasis (p. 17).

(13) 'They do not conserve a trace of their verbal nature.'

(14) While these morphological contexts may be considered typical for adjectives, it should be noted that not all 'true' adjectives can be used as such, i.e. there are semantic restrictions. Adjectives of colour, for example, are felicitous with the superlative degree only in certain contexts, e.g. when comparing different shades of a given colour. Dixon 1977:39 remarks with regard to adverb formation that 'adjectives differ as to whether or not they form adverbs.'

(15) Cf. Thesleff 1954:14: 'I shall use the term concept of value for any concept which is subject to the aspect of "weighing", i.e. any concept of which intensive degrees can be (or are) felt.'

(16) Other candidates would be [TEXT NOT REPRODUCIBLE IN ASCII], both meaning 'very', occurring for example in Soph. Phil. 420 and Hdt. 1.8.1 (the latter of which is ambiguous).

(17) One could argue that there are true adjectives with a bivalent argument structure; see e.g. the adjectives listed in Crespo, Conti & Maquieira 2003:36-37. However, this usually concerns the genitive or dative case, not the accusative case.

(18) I focus here on the negation, but essentially the same argument could be made by referring to adverbs. Cf. e.g. the position of [TEXT NOT REPRODUCIBLE IN ASCII] in a sentence such as [TEXT NOT REPRODUCIBLE IN ASCII], 'or is there any other way in which neglect occurs?' (tr. Bury) (Pl. Leg. 901c).

(19) Bjorck 1940:62 notes that the periphrastic progressive almost never co-occurs with the negation.

(20) Aerts 1965, however, does characterise participles of verbs with a perfect form but a present meaning (e.g. [TEXT NOT REPRODUCIBLE IN ASCII], 'resembling') as 'adjectival' instead of 'situation-fixing', meaning that they are not periphrastic, hence adjectivised. Stahl 1907:681 enumerates some 'adjektivische' (adjectivised) participles, among which [TEXT NOT REPRODUCIBLE IN ASCII], 'sorrowful', [TEXT NOT REPRODUCIBLE IN ASCII], 'wise', [TEXT NOT REPRODUCIBLE IN ASCII], 'fulfilled'/'fulfillable' in Homer, [TEXT NOT REPRODUCIBLE IN ASCII], 'bursting' in Pindar, and [TEXT NOT REPRODUCIBLE IN ASCII], 'in need of' in Aeschylus. These participles all have an idiomatic meaning, diverging from the content verb on which they were built. Remarkably, in the standard lexicon of Liddell & Scott (1968), none of these have a separate entry.

(21) More specifically, they are suppletive in the third person plural of the medio-passive indicative perfect and pluperfect of verbs with consonant-final roots verbs, and the medio-passive subjunctive and optative perfect.

(22) In this context, it is worth noting that the term 'participium' (attested already in Varro [1 BC]; compare also Ancient Greek [TEXT NOT REPRODUCIBLE IN ASCII], and Dutch 'deelwoord') came about because participles form an intermediate category, sharing in the properties of both verbs and nomina (nouns and adjectives).

(23) On the relationship between (compatibility of) Croft's typological analysis and Langacker's conceptual one, see Croft 2001:104 and Langacker 2008:96-98.

(24) Pompei 2006 has similarly shown that the conjunct participle can be located on a noun-to-verb continuum, with the so-called 'appositive' use most adjective-like and the co-subordinative use most verb-like.

(25) In more recent work, Givon 2001:50 recognises three more features ('complexity', 'concreteness' and 'spatial compactness'), though he still considers time-stability to be the primus inter pares, 'giving coherence to the cluster as a whole.'

(26) Pustet 2003:97, in fact, makes a threefold distinction between [+dynamic] lexemes, [[+ or -] dynamic] lexemes and [-dynamic] lexemes, but this will not concern us here.

(27) Compare with Stassen 1997:162, who splits the concept of time-stability into two subordinate principles, namely the 'ingressive parameter' and the 'permanency parameter'.

(28) Note, however, that the same cannot be said of the participle in general: [TEXT NOT REPRODUCIBLE IN ASCII], 'ruler'/'ruling' could be called [[+ or -]dependent], a possibility which Pustet does not seem to recognise.

(29) To be more specific, I analyse whether the property expressed by a particile under normal circumstances can be taken to hold for the (remaining) lifetime of the entity in question (= -transient), is likely not to hold (= +transient), or is rather indeterminate towards this distinction (= [+ or -] transient).

(30) For reasons of space, the overview is not exhaustive, though semantically it covers nearly all types of adjectival present/perfect participle.

(31) This overview must be considered approximative: both the category labels and the classification of the different participles under one of these labels are, to some extent, open for discussion. My main purpose, however, is to give a rough overview of properties. The category labels are largely borrowed from the cross-linguistically based studies of Dixon 1977, Givon 2001:81-84 and Pustet 2003.

(32) Note that [TEXT NOT REPRODUCIBLE IN ASCII] could also be listed under 'mental property' (corrupted in a moral sense) or 'physical property' (in reference to an inanimate entity).

(33) I list [TEXT NOT REPRODUCIBLE IN ASCII], 'existent' under 'bodily state', though it should be noted that this participle is mostly used in a philosophical context (particularly in Plato, see e.g. Pl. Prm. 162a).

(34) [TEXT NOT REPRODUCIBLE IN ASCII], 'coloured' is listed under -transient, contrary to [TEXT NOT REPRODUCIBLE IN ASCII], 'painted' (of a wall), as it is used in reference to haloes (Aristot. Mete. 374a).

(35) Note that [pi][epsilon][phi]u[kappa][??][zeta] can also be used in reference to an inanimate entity.

(36) Note that [TEXT NOT REPRODUCIBLE IN ASCII] can also be used with the meaning of 'acknowledged'.

(37) One of the referees notes that this insight also has its diachronic relevance: 'the fact that the uses of participles with [TEXT NOT REPRODUCIBLE IN ASCII] range from verbal to adjectival on a continuum reflects also the history of many such constructions.' In Bentein 2013b, I argue that the history of verbal periphrasis with [TEXT NOT REPRODUCIBLE IN ASCII] in Archaic and Classical Greek can be captured in terms of transitivisation (with transitivity in the scalar sense of Hopper & Thompson 1980), whereby not only the different periphrastic constructions but also periphrasis in general become increasingly more transitive, i.e. move towards the right side of the continuum.

Table 1:
Formal criteria proposed for adjectivisation of the (present)
participle

Phonological criteria    Morphological         Syntactic criteria
                         Criteria

Phonological reduction   Adverbial formation   Combination with
  of the participle        (Bjorck 1940:17)      adverbs of degree
  (Amenta 2003:32)                               (Karleen 1980:120)
Verbal stem no longer    Formation of          Frequent use in
  recognisable             comparative and       attributive
  (Amenta 2003:32)         superlative           position (Aerts
                           degrees               1965:17)
                           (Bjorck 1940:17)    Syntactic
                                                 coordination with
                                                 true adjectives
                                                 (Boyer 1984:168)
                                               Loss of argument
                                                 structure (Kahn
                                                 1973:136)

Table 2:
Parts of speech as typologically unmarked combinations
(after Croft 2001:88)

             Reference   Modification   Predication

Objects      Unmarked
               nouns
Properties               Unmarked
                           adjectives
Actions                                 Unmarked
                                          verbs

Table 3:
Summary of scores with regard to the three parameters

               Perfect       Present participle    Aorist participle
               participle

Dynamicity     -dynamic      [+ or -] dynamic      +dynamic
Transience     -transient    -transient            +transient
Transitivity   -transitive   [+ or -] transitive   [+ or -] transitive

Table 4:
The position of the participle vis-a-vis [TEXT NOT REPRODUCIBLE IN
ASCII]

                  Group 1     Group 2    Group 3

[TEXT NOT         117 (47%)   55 (62%)   62 (74%)
REPRODUCIBLE IN
ASCII]-
participle

participle--      131 (53%)   34 (38%)   25 (26%)
[TEXT NOT
REPRODUCIBLE IN
ASCII]

Total             248         89         84

Table 5:
The position of [TEXT NOT REPRODUCIBLE IN ASCII]

               TEXT NOT      [TEXT NOT      [TEXT NOT
               EPRODUCIBLE   REPRODUCIBLE   REPRODUCIBLE
               N ASCII]      IN ASCII]      IN ASCII]

[TEXT NOT      1(26%)        10 (42%)       6 (43%)
REPRODUCIBLE
IN ASCII]-
participle

participle--   2 (74%)       14 (58%)       8 (57%)
[TEXT NOT
REPRODUCIBLE
IN ASCII]

Total           43             24             14

Figure 1:
The noun-to-verb continuum (after Givon 2001:54)

most time-stable                    least
                                    time-stable

tree   green   sad    know   work   shoot
noun   adj.    adj.   verb   verb   verb

Figure 2:
Three types of 'adjectival' participle on the
noun-to-verb continuum

N                                          V

    Perfect      Present      Aorist
    participle   participle   participle