From Participles To Gerunds

F: LA7608.tex / p.1 (46-124)

y

JB[v.20020404] Prn:24/11/2004; 14:41

Com

pan

From participles to gerunds*

ing

Io Manolessou

lish

El gerundio llegó a ser un tumor maligno en el cuerpo del idioma. . . Muñío Valverde (1995: 9)

Introduction

Joh

.

nB

enj

am

ins

Pub

This article examines the passage from the inflected Ancient Greek (AG) active participle to the uninflected Modern Greek (MG) active gerund, by offering a synchronic morphological and syntactic description period-by-period. It claims is that this evolution can be satisfactorily explained only if set in a wider context, taking into account (a) similar evolutions in other languages, namely, the passage from partiple to gerund in Romance, Slavic and Baltic and (b) other evolutions involving the participle within Greek, namely, the transformation of the Greek passive participle into a nominal (adjectival) category. The change is interpreted as the split of a “mixed” category (the participle), possessing both verbal and nominal features, to a purely verbal category (the active gerund) and a purely nominal one (the passive participle).

Un

cor

rec

ted

pro

ofs

-

The present paper investigates the transition from the complex AG participle system, which involves full nominal agreement (gender, number, case), multiple tenses and 3 voices (active, middle and passive), to the MG system, where the active voice possesses only an indeclinable gerund.1 It also proposes a general, cross-linguistic motivation for participle evolution, and sets down the implications of historical research for current theoretical approaches to Greek gerunds. The MG gerund is quite different from the English one,2 and therefore, the largest part of current research on the syntax of gerunds is irrelevant to it. Similarly, there is a gap in the research on the historical evolution of the Greek gerund as well,3 since most of the work done is now several decades old, not extensive and in some cases factually inaccurate or lacking a theoretical framework. The main claim in this paper is that gerund evolution in Greek can be properly interpreted only if viewed in a wider context, and not examined as an isolated phenomenon. What must also be taken into account is (a) similar evolutions in other languages and (b) other evolutions involving the participle within Greek.

JB[v.20020404] Prn:24/11/2004; 14:41

F: LA7608.tex / p.2 (124-180)

 Io Manolessou

. The data and the historical process of development

Com

pan

y

Section 2 gives a stage-by-stage account of the parallel evolutions of the gerund and the passive participle, and Section 3 a short exposition of similar developments in other languages. In Section 4 the various theories concerning the development of the MG gerund are evaluated. Finally, in Section 5 the theoretical implications of the data are set out.

enj

am

ins

Pub

lish

ing

What follows is an overview of the developments leading from the AG to the MG participial system in a series of stages/steps, from both a morphological and a syntactic viewpoint. This hopes to supplement the only available information until now, which comes either from a limited set of examples in standard Grammars or the at times inaccurate account by the only comprehensive article on the topic, Mirambel (1961). It should be noted that in order to acquire a secure picture of the linguistic situation in each period – especially of the earlier stages, where the origins of the development are to be sought – scanning of large textual samples would be desirable, for checking the distribution of the various alternative morphological and syntactic options. Unfortunately, such data are unavailable,4 and their collection goes much beyond the scope of this paper. Nevertheless, an effort was made to back up the description with substantial cross-checking from texts.

nB

. Stage 1 – Ancient Greek

ofs

-

Joh

Morphology AG has two types of participle: active (consonant-stem) and middle/passive (o-stem), both fully inflected for gender, number and case, and also indicating aspect and tense.5 The active participle shows a tendency to imitate the large sub-classes of consonantstem adjectives which have a common form for both masculine and feminine, and to exhibit the same usage. This is shown by Rupprecht (1926), Langholf (1977) and Petersmann (1979), who examine the appearance of masculine forms of the participle with feminine subjects in Classical and Post-Classical authors, (1): Khlo:rai de skiades brithontes green ptcl testers.nom.pl.fem abounding.nom.pl.masc.act anne:thou. dill ‘And leafy testers well-dressed with dill.’ (Theocr.Idyll.15.119) b. Hepta. . . pyra:n. . . seven. . . funeral.pyres.gen.pl.fem . . . telesthento:n. having.been consumed.gen.pl.masc.act ‘When the seven funeral pyres had been consumed.’ (Pind.Olymp.6.15)

Un

cor

rec

ted

pro

(1) a.

JB[v.20020404] Prn:24/11/2004; 14:41

F: LA7608.tex / p.3 (180-241)

From participles to gerunds 

pan

y

This usage is especially frequent in the corpus Hippocraticum (5th–4th c BC) and can easily be observed in the gynaecological treatises, where the subject is unambiguously a woman and where the masculine form is very often used for the active participle but never for the passive one (Langholf 1977: 300–302), (2): Myrsine:s phylla embalo:n. . . ito:. myrtle leaves having.placed.on.nom.sg.masc.act. . . go.imper.3sg pros ton andra. to the man ‘Having placed myrtle leaves (on herself). . . let her go to the man.’ (Hp.Mul.I.viii.168.11–12) b. Rhidzas anadzesa:s. . . kai roots having.boiled.nom.sg.masc.act and lu:samene:. . . ito: pros ton andra having.bathed.nom.sg.fem.mid. go.imper.3sg to the man ‘Having boiled roots. . . and bathed. . . let her go to the man.’ (Hp.Steril.VIII.434.7–10)

Pub

lish

ing

Com

(2) a.

am

ins

Thus, what previous accounts usually consider a “first evidence” of indeclinability is a feature of spoken language,6 already present in AG.

ofs

-

Joh

nB

enj

Syntax The participle, both active and passive, has three uses (cf. Smyth 1965: 454–479 and Schwyzer 1950: 387–403 for details): attributive (adjectival modifier), complement (of verbs), and adverbial (temporal, causal, manner, concessive, conditional, final etc). In all three cases it behaves like an adnominal modifier, agreeing in case, number and gender with its subject (cf. (1) and (2)). The participle is either “conjoined”, if its subject is co-referential with a constituent in the main clause, in which case the participle agrees with this constituent, or “absolute”, if its subject is unrelated with all constituents of the main clause, in which case both the participle and its subject appear in the genitive case. Circumstantial participles can be introduced by a variety of conjunctions: temporal (hama), causal (hate, hoion, ho:s), final (ho:s), concessive (kaiper, kai tauta) etc.

pro

. Stage 2 – Koine (NT & papyri)

Un

cor

rec

ted

Morphology The AG system is maintained in general, but papyri show frequent cases of masculine forms for feminine, and of occasional errors of formation (cf. Dieterich 1898: 207; Mayser 1934: 35, 194; Mandilaras 1973: 353–358; Gignac 1981: 131–132). Also, the papyri often display the nominative form of the participle in preference to cases, something which happens with other consonant stem adjectives as well (r-stems, s-stems etc.). However, several of the examples of lack of agreement proposed by Grammars

JB[v.20020404] Prn:24/11/2004; 14:41

F: LA7608.tex / p.4 (241-307)

 Io Manolessou

Joh

nB

enj

am

ins

Pub

lish

ing

Com

pan

y

(refs. above) are not true cases of lack of agreement, but instances of analogical reformations of endings. For example, the masculine nominative plural ending [-es] instead of the accusative [-as] in the participle (e.g. autu:s. . . tithentes BGU.1122, 14–5, 13 BC, tu:s philu:ntes P.Fay.119.26, 100 AD) is not a case of nominative being substituted for the accusative, but of the new analogical accusative ending, which had begun to spread already in the Hellenistic period.7 Mirambel (1961: 50) claims that the first stage of development is the disappearance of inflection, followed only later by confusions of gender, something that is inexact. As shown in 2.1, masculine forms in place of feminine ones are a feature of AG already, and, as will be described in 2.3, masculine forms instead of neuter ones (i.e., [-onta] instead of [-on]) are not a phenomenon of gender change, but of morphological innovation, i.e. a new neuter termination. Furthermore, some of his examples do not belong to this period at all (to:n riθen, ton sximatisθen quoted from Dieterich (1898: 207) come from documents of the 12th c. AD). So in this period, participle inflection seems to be rather well maintained.8 In order to acquire a clearer picture of the functioning of the participial system in the papyri, a closer look at the primary data is required, as the standard Grammars offer no quantitative information concerning the frequency of “incorrect” usage, which would give one an idea of the extent to which the change was beginning to spread. A search of two 3rd c. BC corpora, P.Cair.Zen I–IV and P.Hibeh I, which contain both official and private documents show that breaches of participial agreement are very rare: among the hundreds of instances of participles occurring in the large corpus of these documents, one can hardly find 4–5 examples. These include: use of the masculine instead of the feminine form, (3a), appearance of the masc. nom. plural ending [-es] instead of acc. [-as] as described above, (3b), use of the participle instead of the infinitive, (3c), and, once, use of the masc.nom.sing instead of the fem.acc.sing., (3d). To:n aigo:n tiktonto:n. the goats.gen.pl.fem giving.birth.gen.pl.masc.act (P.Cair.Zen.59338.2) b. Ginoske. . . upote eile:photes he:mas. know.2sg.imp never having.taken.nom.pl.masc.act we.acc.pl.masc ‘Know that we never took.’ (P.Cair.Zen.59343.7) c. Eneukhomai soi. . . apheis te:n gynaika 1sg.pres you.dat.sg. . . . leaving.nom.sg.masc.act the wife mu. my ‘I am ordering you. . . to leave my wife.’ (P.Cair.Zen.59482.5) d. Apestalka soi te:n gynaika send.1.sg.perf you.dat.sg the woman.acc.sg.fem phero:n soi te:n epistole:n. bringing.nom.sg.masc.act you.dat.sg the letteracc.sg ‘I have sent you the woman bringing you the letter.’ (P.Cair.Zen.59443.12)

Un

cor

rec

ted

pro

ofs

-

(3) a.

JB[v.20020404] Prn:24/11/2004; 14:41

F: LA7608.tex / p.5 (307-375)

From participles to gerunds 

Com

pan

y

On the contrary, the 2nd c. corpus of the Tebtunis papyri (P.Tebt.I and II) contains dozens of instances of failing agreement. A crucial point is that they do not concern only 3rd declensions participles, although, in the standard literature it is usually assumed that it is the difficult declension pattern of the 3rd declension (consonantstem), that has caused the eventual fossilization of the active participle into the gerund, (4): To hypomne:ma epidedomenon para Mestasytmios. . . the report delivered by Mestasytmis.gen.sg.masc. . . hypiskhnu:menos. promising.nom.sg.masc ‘The report delivered by Mestasytmis, promising. . . ’ (P.Tebt.58) b. Ge:s kle:rukhike:s syno:rismene:n. land.gen.sg.fem cleruchic bordered.acc.sg.fem.pass ‘Of cleruchic land, bordered. . . ’ (P.Tebt.82)

Pub

lish

ing

(4) a.

nB

enj

am

ins

A comparison with another 2nd c. corpus, UPZ I and I.II, shows that this high frequency is rather above the norm; in the UPZ corpus breaches of agreement occur only 5–6 times as well. Similarly, the papyrological specimens of the Christian era examined (Pap.Cair.Masp.I and III, BGU XV) contain very few instances of aberrant participial usage. The conclusion to be drawn from the above overview is that the image presented by the papyri depends mainly on whether the writer’s native idiom is Greek (as in most of the Zenon and UPZ papyri) or Egyptian (as in most of the Tebtunis corpus) or on the level of his education (as in the Christian era documents, which are mostly legal texts).

Un

cor

rec

ted

pro

ofs

-

Joh

Syntax Some uses of the participle are reduced: Thus, the predicative participle (i.e., in the function of a verbal complement) is radically restricted in the NT, confined almost entirely to Luke and Paul (BDF §414). Of the circumstantial meanings, some are retained (temporal, causal, concessive) and some are on their way to disappear (conditional, final), and the same is valid for the papyrological data as well (Mayser 1926: 348– 352, BDF §411; Jannaris 1897: 506). However, other uses are still quite strong, if not increased- extensive participle usage is a characteristic feature of the Koine (cf. Horrocks 1997: 46; Mayser 1934: 62). Most of the conjunctions accompanying participles disappear or are retained only when the participle is used absolutely (BDF §425, Mayser 1934: 64, 74). An important syntactic innovation is the increase of unclassical “absolute” participial constructions,9 in two different structures: use of genitive absolute participles although the participial subject is co-referent with a term in the clause (5a and b) and “hanging” nominative, a participle whose subject is not co-referent with any term in (5c):

JB[v.20020404] Prn:24/11/2004; 14:41

F: LA7608.tex / p.6 (375-430)

 Io Manolessou

Apestale:n eis ton hypo soi nomon. . . emu: send.1sg.aor.pass. to the under you nome. . . I-gen aite:samenu: having.petitioned.gen.sg.masc.act ‘I was sent to your district, having asked for it myself.’ (P.Giss.11.4 (118 AD)) b. Me: ekhontos autu: apodu:nai, ekeleusen neg having.gen.sg.masc.act he.gen.sg to-give back, ordered auton ho kyrios autu:. him.acc.sg the master his ‘He being unable to give it back, his master ordered him. . . ’ (Ev.Matt.18.25) c. Ekballu:sa hemas anekho:re:samen. kicking.out.nom.sg.fem.act us left.1pl ‘After she kicked us out, we left.’ (UPZ 18.17 (163 BC))

Pub

lish

ing

Com

pan

y

(5) a.

-

Joh

nB

enj

am

ins

This “breach” of the classical norm is not a Hellenistic evolution, as it already existed to a certain degree in AG – Spieker (1885) and Schwyzer (1942) provide several examples. The innovation lies in the heightened frequency of the phenomenon. An examination of the 4 Gospels of the NT, for example, shows that the norm is to use the absolute construction whenever the participle is not subject-oriented, and to employ the conjoined participle when it refers to the main clause subject (in which case it appears in the nominative) or to an infinitival clause subject (in which case it appears in the accusative). In the same vein, Whaley (1990), examining the choice of the absolute vs. the conjoined construction in the NT, establishes that absolute participles occur even when their subject is a constituent of the matrix clause when (a) the participial subject is co-referent with a non-primary term, such as an indirect object, a prepositional complement or noun raised out of a participial or subordinate clause or (b) the participle belongs to an unaccusative or unergative verb.

. Stage 3 – Late post-classical / early medieval (4th–6th c.)

pro

ofs

Morphology The first concrete signs of inflectional erosion are neuter nom./acc. singular forms ending in [-onta] instead of [-on], appearing ca. the 4th c. (Kapsomenakis 1938: 40), (6). Zoðion. . . katexonta lampaða. animal.acc.sg.neut. . . . holding.acc.sg.masc.act torch ‘An animal holding a torch.’ (PGM.II.36.179, 4th c. AD) b. To praγma to ðia Evðemonos lekθenta. the thing.acc.sg.neut the by Eudaimon said.acc.sg.masc.pass ‘The thing that was said by Eudaimon.’ (POxy.1348, late 3rd c. AD)

Un

cor

rec

ted

(6) a.

JB[v.20020404] Prn:24/11/2004; 14:41

F: LA7608.tex / p.7 (430-484)

From participles to gerunds 

ing

Com

pan

y

It is unclear whether this ending is an extension of the neuter plural nom./accusative or of the masculine singular accusative, as both end in [-onta]. The second alternative is much likelier, both because the masculine is more frequent in speech than the neuter, and also because of a possible analogy with the 2nd declension (o-stem) passive participles:10 in the 2nd declension, the masculine singular acc. ending is identical to the neuter singular nom. and acc. (being in all cases [-menon]), so this formal identity between masculine and neuter might have been carried over to the 3rd declension as well. In texts, active participles with the new ending frequently occur side by side with medio-passive ones, (7): Cektime ktima mite possess.1sg.perf possession.acc.sg.neut neither fθiromenon mite liγonta. decaying.acc.sg.neut.pass neither ending.acc.sg.masc.act ‘I possess a possession which neither decays nor ends.’ (Acta Thomae 136.15) b. Lutron mi ypoceomenon alla. . . bath.acc.sg.neut neg burning.under.acc.sg.neut.pass but parexonta. offering.masc.sg.acc.act ‘A bath not heated underneath, but offering. . . ’ (Malalas 178.65)

am

ins

Pub

lish

(7) a.

Un

cor

rec

ted

pro

ofs

-

Joh

nB

enj

In early Medieval Greek (MedG) texts, such neuter forms occur occasionally.11 There are no quantitative studies on the extent of the phenomenon, and indeed it is almost impossible for them to exist or to form a solid basis for conclusions, due to the textual tradition of these linguistic monuments: the evidence for spoken language for the period between the 6th and the 12th c. is very meager, and in most cases preserved in manuscript copies several centuries posterior to the date of composition. The different manuscripts preserving a text of the period may present several variant readings in the passages in question, some maintaining the “classical” form and others showing the innovative one. To give some examples: the text of the critical edition of the Life of St. John the Almsgiver, 6th c. (Gelzer 1893) prints 6 cases of the neuter participle with the new -onta ending (cf. the editor’s Grammatisches Verzeichnis, entry Participia), and there are alternative readings in -onta in 3 more cases, to be spotted only by checking the apparatus (at 50.6, 87.22 and 97.15). However, the manuscript tradition (the 6 mss., ABCDEF, used in the edition) is unanimous in none of these eight cases: three appear only in A, two only in C, one only in E, one in ACEF and one in ABCE. It is thus impossible to guess which and how many of those stood in the original text, and which are readings introduced by a later copyist. On the other hand, the text of Malalas, a main source for the low-register language of the period, is preserved in only one manuscript, of the 12th c., and thus the lack of corroborating manuscript evidence makes it another kind of insecure textual witness.

JB[v.20020404] Prn:24/11/2004; 14:41

F: LA7608.tex / p.8 (484-569)

 Io Manolessou

Table 1. Century

Attributive

Complement

Malalas Leontios of Neapolis Chronicon Paschale Vita Epiphanii Apocalypses Apocryphae Funerary inscriptions Digenis E Chronicle of Morea (6000vv.) War of Troy (4000vv.) Velthandros Livistros V Machairas (40pp)

6th 6th 5th 6th 2nd–6th 5th–7th 12th 14th 14th 14th 14th–15th 15th

15 4 3 3 5 formulaic 0 0 0 0 0 2

2 0 0 0 1 – 2 9 2 3 0 0

Adverbial 4 2 0 0 0 – 5 45 17 7 2 49

Pub

lish

ing

Com

pan

y

Text

-

Joh

nB

enj

am

ins

Nevertheless, considering that the new neuter -onta ending appears in almost all the near-vernacular texts we possess from the period, even if not in unanimous textual tradition, and even if less frequently than the classical ending, it should be considered as established in the spoken language by this time. The inscriptional evidence from the same period (cf. Section 2.4 below) corroborates this assumption. A further development in the active participle is the obsolescence of the Perfect form (in -o:s, -yia, -os). It occurs very rarely in early MedG texts, often conjoined with Aorist forms (e.g. elθo:n. . . ce eorako:s Mal.221.62, pyisas ce. . . katesxiko:s Mal.372.15), showing that the feeling for a special meaning of the Perfect had been lost. This semantic development in the Perfect (cf. Hatzidakis 1892: 204–205; Wolf 1911: 65–66, and esp. Moser 1988 for details), which led to the loss of monolectic Perfect forms and the evolution of periphrastic ones, left also the Passive Perfect Participle without paradigm support. Gradually, one of its main characteristics, initial reduplication, was lost, and this marked a step away from verbal properties.

Etiçise to ðoras, xorion fortify.1sg.aor the Doras.acc.sg.neut, village onta. being.acc.sg.masc.act ‘He fortified Doras, which is a village. . . .’

cor

rec

(8) a.

ted

an attributive modifier, equivalent to a relative clause, (8):

Un

–

pro

ofs

Syntax The new neuter form is still an agreeing participle, employed mostly with attributive meaning. This can be gauged from the Table 1 which lists the usage of -onta/-ontas forms in earlier vs. later MedG texts. All [-onta] forms in earlier texts are neuter nom. or acc. singular, while those in the later texts are of any gender/number. “Attributive usage” includes cases where the participle is used as:

(Mal.326.54)

JB[v.20020404] Prn:24/11/2004; 14:41

F: LA7608.tex / p.9 (569-637)

From participles to gerunds 

–

Com

pan

y

b. To paraksenon lutron tu xrisokastrou/ to the beautiful bath.acc.sg.neut. of the golden-castle/ the jemonta tas iðonas. filled.acc.sg.masc.act the pleasures ‘The beautiful bath of the Golden Castle / filled with every kind of pleasure.’ (Kallimachos 1720) a substantivised adjective, (9):

–

Pub

lish

ing

(9) Meli to pote apoθanonta ce going.to.3sg the.acc.sg.neut once dead.acc.sg.masc.act and anastanta. . . apelefsesθe. having.risen.acc.sg.masc.act. . . leave.inf ‘That which has died and risen again is going to leave.’ (Vit.Epiph.89A) a predicative modifier (i.e. a small clause), (10):

am

ins

(10) Peðariu teleftisantos, zonta apeðoce child.gen.sg.neut having died, living.acc.sg.masc.act give.3sg.aor ti mitri. to the mother ‘A little child having died, he gave it back to its mother alive.’ (Chron.Pasch.181)

enj

“Complement” includes cases where the participle is the complement of a verb, (11): θeorunton to θirion ekfevγonta. seeing.pl.gen the beast.neut.sg.acc escaping.masc.sg.acc.pres.act ‘Seeing the beast escape.’ (Romanus Melodus 72.κδ.2) b. Fenome pipraskonta ton emon ambelona. appear.1sg.pres selling.acc.sg.masc.act the my vineyard ‘I declare that I am selling my vineyard.’ (Guillou, Messina 5, 1135 AD)

-

Joh

nB

(11) a.

ofs

“Adverbial usage” includes cases where the participle is used as an adverbial, expressing manner, cause, concession, etc., (12): ðeçete afton jyneon greet.1sg.pres him woman.nom.sg.neut prospiptonta ce leγonta. falling.down.acc.sg.masc.act and saying.acc.sg.masc.act ‘A woman greets him by falling on her knees and saying.’ (Leontios 64.1) b. Kleonta ce oðiromenos cite crying.acc.sg.masc.act and wailing.nom.sg.masc.pass lie.3sg.pres is to klinarin. at the bed ‘He lies on the bed, crying and wailing.’ (Digenis E 393)

Un

cor

rec

ted

pro

(12) a.

JB[v.20020404] Prn:24/11/2004; 14:41

F: LA7608.tex / p.10 (637-708)

 Io Manolessou

Com

pan

y

The old forms of the participle (masculine, feminine, other cases of the neuter), perfective and imperfective stem, are still in full use and extremely frequent. In early MedG the unclassical usage of absolute participles observed in the previous period continues and even increases (cf. Jannaris 1897: 500; Weierholt 1963: 70– 75). At the same time, a more clear-cut indication of the breakdown of agreement between the matrix clause and the participial clause appears: the participle often seems to agree with the element immediately adjacent to it instead of with its subject (cf. also Wolf 1912: 27–28), (13). Emine to jenos tu Perseos remained the kin.nom.sg.neut of Perseus.gen.sg vasilevontos [for vasilevon]. ruling.gen.sg.neut.act ‘The kin of Perseus remained ruling (in power).’ (Mal.28.24) b. Apelθonta ce to riγonti having-left.acc.sg.masc.act and the shivering.dat.sg.masc.act synantisanti [for synantisanta]. met.dat.sg.masc.act ‘When he left and he met the shivering one.’ (Leontios 48.8) c. Eroso my cyrie mu pater, be.healthy.2sg.imp. me.dat, lord my father, eftyxunti my [for eftyxon]. being.happy.dat.sg.masc.act me.dat ‘Be healthy, my lord father, being happy for my sake.’ (Rev.Eg.1919.201)

nB

enj

am

ins

Pub

lish

ing

(13) a.

Joh

Another potential symptom of a purely “verbal” re-interpretation of the participle is the tendency to equate it with a finite verb, able to be coordinated with it (cf. Jannaris 1897: §2168b; Wolf 1911: 56; Frisk 1928; Schwyzer 1950: 407; Mandilaras 1973: 372; Kavˇciˇc 2001, and esp. Cheila-Markopoulou 2003), (14).

ofs

-

(14) γrapsas ðe o Zakçeos. . . ce leγi. having.written.nom.sg.masc.act and the Zacchaeus. . . and say.3sg.pres ‘Zacchaeus has written. . . and says.’ (Evang.Thomae B 7.1)

pro

. Stage 4 – Middle Byzantine

Un

cor

rec

ted

Morphology The next step in the evolution is the spread of the [-onta] ending to masculine and feminine forms. Early examples occur in inscriptions from Asia Minor (Klaffenbach 1933). A search in inscriptional corpora12 shows that early Christian funerary inscriptions standardly contain the formula mnimion/mnimorion/cymitirion ðiaferonta tu X (‘grave.neut belonging to X’, cf. Bees 1910). There are abundant attestations from Corinth, Thessaly, Macedonia, Attica etc., evidence of the spread of the new neuter ending. In the inscriptions from Korykos (5th and 6th c., Keil & Wilhelm 1931), the fu-

JB[v.20020404] Prn:24/11/2004; 14:41

F: LA7608.tex / p.11 (708-775)

From participles to gerunds

Com

pan

y

nerary formula is feminine: θici/somatoθici ðiaferusa tu X (‘coffin.fem belonging.fem to X’). In many cases, it has changed to θici ðiaferonta. These are the earliest consistent13 attestations of the spread of [-onta] to the feminine. The earliest securely dated (9th c.) occurrence of the [-onta] suffix in an adverbial function with a masculine noun occurs in the Proto-Bulgarian inscriptions (Beshevliev 1963), (15):

ing

(15) Is tis Plskas ton kampon menonta at the Plska.gen the plain remaining.acc.sg.masc.act, epyisen avlin. make.3sg.aor court ‘Remaining in the plain of Plska, he built a court.’ (56.5–7, 822 AD)

am

ins

Pub

lish

Syntax Unfortunately, texts close to spoken language are insufficient in this period, and thus do not allow reliable conclusions. The -onta form seems to be still mainly attributive; however, the adverbial usage of the gerund form has already appeared. The AG participle form is still extensively used, mainly in temporal and causal function. The passive participle is used mainly attributively, and often as a verbal complement, being very frequent in perfect periphrases with the verbs “to be” and “to have” (cf. Browning 1983: 32–34; Aerts 1965; Moser 1988 for an overview).

enj

. Stage 5 – Later Byzantine (12th–15th c.)

ofs

-

Joh

nB

Morphology The [-onta] gerund becomes established for all genders and cases. From the 14th c. onwards, a final -s is added to the ending, giving it the standard MG form [-ontas].14 This is in all likelihood an adverbial suffix, appearing in other adverbs as well, e.g. totes ‘then’, potes ‘never’, tipotas ‘nothing’ (Hatzidakis 1934; Horrocks 1997: 229), and not the [-s] suffix of the nominative (as in Schwyzer 1950: 410). The forms with and without [-s] coexist in texts of the period, but distributional data are lacking. This period sees an important evolution in the passive participle, the extension of passive perfect forms to active morphology verbs with stative, inchoative, unaccusative or unergative meaning, something that is maintained in MG (Tzartzanos 1989: 330– 331; Moser 1988: 145–152; Alexiadou & Anagnostopoulou 1999: 25), (16):

rec

ted

pro

(16) pinao → pinazmenos I am hungry troo → faγomenos I eat nistazo → nistaγmenos I am sleepy γonatizo → γonatizmenos I kneel kitrinizo → kitrinizmenos I grow yellow

Un

cor

This was not the case in AG, except for a couple of isolated instances; it is a Medieval evolution (details in Hatzidakis 1924, 1927) but there is no information on when these



JB[v.20020404] Prn:24/11/2004; 14:41

F: LA7608.tex / p.12 (775-850)

 Io Manolessou

y

participles first appeared. Examples do occur in the first vernacular medieval texts (12th–13th c.), (17).

Com

pan

(17) Ipate, kinijisete opu iste maθimeni. go.2sg.imp hunt.2sg.imp where are.2pl learned.nom.pl.masc.pass ‘Go hunt where you used to.’ (Digenis E 1303)

ing

It also seems that passive participles with active meanings are a Balkan Sprachbund phenomenon, appearing in some Balkan languages as well (cf. Lindstedt 2002); this would point to a medieval datation of this evolution as well.15

Pub

lish

Syntax The main function of the active participle is adverbial – complement usage is rare, and attributive almost non-existent. The active form can now be properly called a gerund, as it is not an acceptable nominal modifier any more, (18): Tafta ipon estrafice jelonta that saying.nom.sg.masc.act turned.3sg laughing.acc.sg.masc.act pros ecinin. towards her ‘Having said that, he turned to her laughing.’ (Velthandros 861) b. Utos exonta jinekan, to ðiceon orizi. he having.acc.sg.masc.act wife, the law dictate.3sg ‘If he has a wife, the law dictates. . . ’ (Assizai.146.15–6)

enj

am

ins

(18) a.

nB

However, the complement usage is not yet extinct, (19): Tote na iðes arxondises. . . kratunta ta then sub see.2sg.aor noblewomen holding.acc.sg.masc.act the peðia tus. children-their ‘Then you could have seen. . . noblewomen holding their children.’ (War of Troy 1101) b. Fenome pipraskonta ton emon ambelona. appear.1sg.pres selling.acc.sg.masc.act the my vineyard ‘I declare that I am selling my vineyard.’ (Guillou, Messina 5, 1135 AD)

pro

ofs

-

Joh

(19) a.

Un

cor

rec

ted

The only attributive usages appear with the neuter (cf. the formula xorion to onta ce ðiacimenon ‘the village being and located’ in S. Italian documents.16 The passive participle is mainly used adjectivally (details in Moser 1988: 229–235). The AG forms are still used side by side with the gerundival ones, their frequency depending on the register of the text. With the loss of case from the gerund, the ancient type of “absolute” participles (in the genitive or hanging nominative) disappear. However, vestiges of absolute participles in the genitive survive in conservative MG dialects, such as Cypriot and Cretan,

JB[v.20020404] Prn:24/11/2004; 14:41

F: LA7608.tex / p.13 (850-900)

From participles to gerunds 

y

in the form of gerunds having genitive subjects (Menardos 1925b: 64; Giakoumaki 1993) (20): Poθanonta tu pappu mu, epiramen to dying.acc.sg.masc.act the grandfather.gen.sg. my, get.1pl.aor the spitin. house ‘Upon my grandfather’s dying, we got the house.’ b. Kateontas mu inda muflusis ine, ðe du knowing I.gen.sg. what bankrupt is, not he.gen.sg ðuðo ðanika. give.1sg.pres loan ‘Knowing what kind of a loser he is, I’m not giving him a loan.’

lish

ing

Com

pan

(20) a.

am

ins

Pub

Alternatively, these examples of deviant dialectal syntax, unacceptable in standard MG, could be viewed not as archaic survivals but as new independent evolutions; in this connection, the parallel with the structures evolved in languages such as English or Hebrew, where genitive subjects for indeclinable gerunds are the norm is quite striking. However, it has to be noted that these constructions date at least to late MedG, as they can be found in works of the period, (21): I rijena i Alis. . . pijenonta the queen.nom.sg.fem the Alice going.acc.sg.masc.act tis is ton Maçeran. . . eθelise na bi. she.gen.sg to the Machairas want.3sg.aor sub enter.3sg.aor ‘Queen Alice, upon her going to Machairas. . . wanted to enter.’ (Machairas 54.18) b. Strefonta tus i mandatofori. . . turning.acc.sg.masc.act they.gen.sg the envoys . . . eðokan tas γrafas. give.3pl.aor the letters ‘Upon their returning, the envoys gave the letters.’ (Machairas 94.36)

-

Joh

nB

enj

(21) a.

ofs

Apart from these dialectal constructions with the genitive, the gerund does have “absolute” uses, in that its subject is not co-referential with that of the main clause, (22): ðiavonta γar enas ceros, having.passed.acc.sg.masc.act a time.nom.sg.masc ejirisen ecinos. return.3sg.aor that.nom.sg ‘Some time having passed, he returned.’ (Chr.M.1048) b. Ce lalonta ton loγon, ekaθarisen i and speaking.acc.sg.masc.act the word, clear.3sg.aor the

Un

cor

rec

ted

pro

(22) a.

JB[v.20020404] Prn:24/11/2004; 14:41

F: LA7608.tex / p.14 (900-965)

(Machairas 68.21)

pan

γlosa tis. tongue.nom.sg hers ‘And as she spoke the word, her tongue was freed.’

y

 Io Manolessou

Com

This continues into the next period as well, and is maintained up to MG (details of this usage in MG literature in Nakas 1985), cf. two examples from the 17th c., (23): Ontas aftos peðion akomi, ton being.acc.sg.masc.act he.nom.sg child still, he.acc.sg estilen o pateras tu. send.3sg.aor the father he.gen.sg ‘When he was still a child, his father sent him. . . ’ (BGV III, 362.21) b. Angaliazontas ena ðendro. . . me ektipusan embracing.acc.sg.masc.act a tree. . . I.acc.sg hit.3pl.imperf e petre. the rocks.nom.pl ‘As I was holding on to a tree. . . the rocks were hitting me.’ (BGV I, 331)

Pub

lish

ing

(23) a.

ins

. Stage 6 – Post-Byzantine Greek

enj

am

Morphology The next stage in the evolution of the gerund is the loss of tense. In the previous period active participles could be formed both from aorist and present stems, (24):17 Staθonta kata anatolas anaγnose ce ipe having.stood towards east read.2sg.imp and tell.2sg.imp ton. he.acc.sg ‘First stand facing east, and then read and tell him.’ (War of Troy 561) b. Iðontas tuto o ajios. having.seen that the holy ‘The holy father, having seen that. . . ’ (Chr.M.18)

-

Joh

nB

(24) a.

pro

ofs

After the end of the MedG period, gerunds can only be formed from present stems, except for the Greek dialects of S. Italy. In some of these dialects, aorist gerunds are used only with the verb “to be” in perfect periphrases and the present stem is employed in all other functions (Katsoyannou 1995); however, in others the Aorist gerund is still a living category (Karanastasis (1997: 144), (25).

rec

ted

(25) Vletsonta o cˇ ero, en ixa panta Luppiu. having.seen the weather, not have.1sg gone Lecce ‘If I had seen the weather, I wouldn’t have gone to Lecce.’

Un

cor

As far as the passive participle is concerned, the Present form of the participle (-omenos) becomes gradually reduced in use, and in fact ceases to be an integrated

JB[v.20020404] Prn:24/11/2004; 14:41

F: LA7608.tex / p.15 (965-1030)

From participles to gerunds 

Com

pan

y

part of the participial system. Apart from a few analogically created popular forms in -amenos and -umenos, (e.g. trexamenos ‘running’, petumenos ‘flying’), which have almost exclusively attributive function,18 the present passive participle in MG is a formation re-introduced in the language via the kathareuousa (see below 2.7). So the system of the participle, as presented in the first Grammar of MG, Nikolaos Sofianos (ca.1544), is as follows (Papadopoulos 1977: 76):

lish

ing

all the participles of the old Greeks are analysed with the finite verb of the tense which the participle would be in, plus [the relative complementiser] opu. . . In all active tenses. . . there is only one non-finite participle, γrafontas, kratontas. . . As for the passive verbs, the perfect participle has been maintained up to our time, and it is inflected in the three genders, o γramenos, i γrameni, to γrameno. . .

am

ins

Pub

Around the 16th c., therefore, the active voice possessed only the active gerund in [-ontas], while in the passive voice only the perfect participle survived. The situation described by Sofianos is easily exemplified by a recent study on the transposition of AG participles in 17th c. vernacular translations of the lives of Aesop (Karla 2002). In this case, the large majority of participles of the original, post-classical, text have been replaced by finite clauses or adjectives, and only 11% of them have been retained (either as archaic declinable participles or as gerunds).

. Stage 7 – Modern Greek

nB

enj

Syntax The MG situation, with the active gerund having predominantly adverbial usage and the passive participle predominantly adjectival, has been reached.

Un

cor

rec

ted

pro

ofs

-

Joh

Morphology Standard MG sees a new regularisation/filling out of the participial and gerundival paradigm. Side-by-side with the “present” gerund, a periphrastic perfect form is created, using the gerund of the auxiliary verb “to have” along with a perfective infinitive form (e.g. exontas γrapsi). There are no data concerning the date of appearance and the spread of this new formation (Nakas 1991: 178), but older Grammars (Triantafyllidis 1941: 373; Tzartzanos 1989: 339) recognise its innovative, “artificial”, status, and statistical analyses of participial usage in MG (Iordanidou 1985; Rydå 1988) show that it is very rare; some Grammars (including school grammars) do not mention it at all. In the passive domain, the present participle is re-introduced. This is generally accepted to be a feature of kathareuousa, and not a direct inheritance from AG (Nakas 1991: 182–187). Earlier Grammars explicitly state that it is an element foreign to “demotic”, and statistical investigations also show that its use is limited. Furthermore, it is far from productive; thus, verbs of modern/demotic origin rarely if ever have present participles (e.g. *lavonomenos, *viðonomenos, *leronomenos, *spazomenos) and verbs which have both a kathareuousa and a demotic variant possess a present participle only for the first (liomenos vs. *linomenos, enðiomenos vs. *dinomenos, feromenos vs.

JB[v.20020404] Prn:24/11/2004; 14:41

F: LA7608.tex / p.16 (1030-1095)

 Io Manolessou

lish

ing

Com

pan

y

*fernomenos etc.). The passive present participle cannot therefore be considered as fully integrated into the verbal paradigm, on a par with any other verbal form. Another evolution belonging to the MG period is the appearance of a periphrastic passive gerund, formed by the gerund of the verb to have and a perfective passive infinitive (e.g. exontas γrafθi). This is yet another new and not extensively used form, frequently absent from Grammars of MG because of its rarity. The evolution of this form is one more indication of the non-verbal character of the monolectic passive participle: it is this new periphrastic formation, and not the inherited “perfect participle”, that unambiguously expresses the categories of “perfect tense” and/or “perfect aspect”, as well as of “passive voice”. But it is a gerundival, and not a participial form, showing once more that verbal and nominal features are incompatible in MG participles.

am

ins

Pub

Syntax This topic is extensively treated in Tsoulas (1996), Tsimpli (2000), Sitaridou and Haidou (2002), Moser (2002) and Tsokoglou and Kleidi (2002) and will therefore not be developed in any detail here. The MG gerund has only adverbial meaning and cannot appear in argument positions.19 Its temporal interpretation is dependent on the meaning of the verb and the general context (cf. Tsimpli 2000; Moser 2002 for details). Contrast (26a) with (26b): Efije klinontas ðinata tin porta. leave.3sg.aor shutting loudly the door ‘He left, banging the door loudly.’ (anteriority) b. Klinontas tin porta, iðe tin efimeriða sto katofli (simultaneity) shutting the door, saw the newspaper at the threshold “As he was shutting the door, he saw the newspaper on the threshold.”

nB

enj

(26) a.

ofs

-

Joh

Furthermore, the gerund is often considered exclusively subject oriented, although it is possible to find gerund subject which are: (a) null and non-co-referential with the main clause subject in the case of impersonal or arbitrary/generic reference (27a) (b) co-referential with another term in the clause (27b and c) “nominative absolute” subjects different to that of the main clause, which must always appear after the gerund (27c): Troγontas erçete i oreksi. eating come.3sg.pres the appetite.nom.sg ‘The appetite comes with eating.’ b. Telionontas to Panepistimio, ton pirane sto strato. finishing the University, he.acc.sg take.3pl.aor to the army ‘When he finished the University, he was drafted in the Army.’ c. Fevγontas i ðaskala, jelasan ta peðjia. leaving the teacher.nom.sg, laugh.3pl.aor the children ‘As the teacher left, the children laughed.’

Un

cor

rec

ted

pro

(27) a.

JB[v.20020404] Prn:24/11/2004; 14:41

F: LA7608.tex / p.17 (1095-1174)

From participles to gerunds 

pan

y

As for the passive participle, there is a relative consensus that it has a number of nominal properties which make its distinction from a simple adjective difficult. These include (Moser 1988: 167–168; Anagnostopoulou (2001: 2–4): it can occur in pre-nominal position as noun modifiers it can have comparative and superlative it can be conjoined (with ce (‘and’)) with true adjectives it can occur as predicate with verbs like fenome (‘look’), mnjiazo (look like’) etc. the agent can be incorporated, e.g. iliomavrismenos (with suntan) it can be intensified with para-(‘too’) , e.g. paramavrismenos (with too much suntan) g. it can form adverbials with -a, e.g. θlimenos > θlimena (‘sad’ > ’sadly’) h. its derivation is irregular, i.e. several active morphology verbs form it, and several passive form verbs do not.

lish

ing

Com

a. b. c. d. e. f.

ins

Pub

The passive participial clauses quoted in Tsimpli (2000: 159) are no different from other adjective small clauses, which attribute a “temporary” property to the modified noun. Thus, compare the sets of sentences in (28a, b) and (29a, b): Me ðemena ta çerjia ðen borusa na with tied the hands neg can.1sg.imperf sub ksisto. scratch.1sg.aor.pass ‘With my hands tied, I couldn’t scratch myself.’ b. Me elefθera pja ta çerjia, boresa na with free at.last the hands can.1sg.aor sub ksisto. scratch.1sg.aor.pass ‘With my hands free at last, I managed to scratch myself.’

Joh

nB

enj

am

(28) a.

Kinijimenos / ðjoγmenos o Jianis, anangastike na hunted / kicked.out the John, force.3sg.aor.pass sub fiji. leave.3sg.aor ‘John, hunted / kicked out, was forced to leave.’ b. Etimos jia ola o Jianis, arniθice na fiji. ready for everything the John, refuse.3sg.aor sub leave.3sg.aor ‘John, ready for anything, refused to flee.’

pro

ofs

-

(29) a.

ted

So the syntactic roles of the passive participle are identical to those of the adjective.20

rec

. Summary of evolution

Un

cor

On the basis of the above, the evolution of the Greel participial system can be summarised as follows:

JB[v.20020404] Prn:24/11/2004; 14:41

F: LA7608.tex / p.18 (1174-1221)

 Io Manolessou

nB

enj

am

ins

Pub

lish

ing

Com

pan

y

1. The active participle already showed diminished gender agreement from AG times, but maintained its functionality until the first centuries AD. The first sign of evolution is syntactic, namely the increase in absolute constructions in the Koine period, an indication that the agreement mechanism of the participle with the matrix clause is breaking down. Around the 4th c. the first morphological effects of change become evident, with the reduction in the inflectional system represented by the innovative neuter ending -onta (arising probably from collocations with the accusative of the passive participle). This spreads to masculine and feminine nouns during the 6th–8th c., as testified by inscriptional evidence. The next step in the evolution, after the loss of agreement, is the loss of tense (in reality of the imperfective/perfective distinction), occurring around the 14th c. The addition of the adverbial -s suffix marks the final and definitive reanalysis of the gerund as an adverb. The MG gerund has neither Agreement, nor Tense – however, it is in the process of acquiring a new aspectual distinction. 2. The passive participle maintained its full nominal agreement system throughout the history of Greek. The development went rather in the way of loss of verbal characteristics. Thus, Koine times see a reduction in the types of participle available (loss of final, concessive, causal etc. participles), as well as of the subordinating conjunctions/complementisers introducing them. Also, the Tenses become reduced, with the loss first of the Future and the Aorist, and later of the Present. In later stages, the uses of the participle become further reduced, until it is reserved only for attributive functions. At some point in the Medieval Period, the category of Voice is lost as well, since Passive participles begin to be formed also from active morphology verbs with unaccusative or unergative meaning.

pro

ofs

-

Joh

Thus it is crucial misconception that “the active participle changed and disappeared”, while “the passive participle did not change and was maintained”. In fact, the passive participle has changed; an indication of this are the numerous studies examining the similarities and differences between “passive participles” and adjectives, and proposing criteria of differentiation, e.g. Laskaratou and Philippaki (1984), Setatos (1985), Sklavounou (2000), Anagnostopoulou (2001). The evolution examined here is not a change affecting one part of the participle system while leaving the other intact, but a restructuring of the whole system, affecting both parts in different ways.

. Parallel evolution in other languages

rec

ted

The change from participle to gerund is a phenomenon not unique to Greek, but crosslinguistically well attested. In the words of Haspelmath (1995: 17):

Un

cor

Converbs [=nonfinite verb forms functioning as adverbial modifiers] seem to arise from two main types of sources: (a) . . . verbal nouns which have become independent from their original paradigm; and (b) . . . participles which lost their capability for agreement.

JB[v.20020404] Prn:24/11/2004; 14:41

F: LA7608.tex / p.19 (1221-1283)

From participles to gerunds 

pan

y

As will be shown below, process (b) appears in Slavic, Baltic and Romance, where, independently, the active present participle has become a gerund, as in Greek.

. Romance

ins

Pub

lish

ing

Com

The historical syntax of the Romance languages displays a pattern very similar to Greek: the indeclinable gerund acquires all the verbal functions of the active participle, and the latter becomes a pure adjective, outside the verbal system (Harris 1978: 199– 203). More specifically: Classical Latin possessed a triple opposition between (a) an agreeing passive gerundive, (b) a non-agreeing active gerund, similar to a verbal noun and (c) an agreeing, active present participle. The gerundive disappeared completely, and so did three out of the four cases of the noun-like gerund (genitive, dative and accusative). The ablative case of the gerund, however, originally denoting only instrumental notions such as manner, accompaniment etc., gained ground widely, and rivalled the present participle as an adverbial modifier to the main verb. Already in Classical Latin it was possible to find present participles co-coordinated with ablative gerunds (Kühner-Stegman 1966: 753), (30).

nB

enj

am

(30) incendium. . . in edita assurgens et rursus fire. . . in high climbing.neut.nom.sg.pres.act and then inferiora populando lower devastating.ger.abl ‘The fire. . . climbing first to the high places and then devastating the lower ones.’ (Tacitus 15.38)

-

Joh

Later Latin sees: (a) the reduction of the active present participle to adjectival uses (b) the extension of absolute constructions, even when the subject is a term of the clause (c) the use of the participle as a finite verb (three developments identical to Greek) and (d) the passing over of the adverbial uses of the participle to the ablative of the gerund.21 For example, in Medieval Spanish translations of Latin texts, the participle, especially in absolute usages, is often glossed by the gerund (Muñío Valverde 1995: 12).

ofs

ignorans: glosed as non sapiendo ignoring-masc.sg.nom.pres.act: glossed as not knowing-gerund.abl

pro

In French, the gerund and the participle coalesced morphologically, to give one form, ending in -ant, which is used either indeclinably as a gerund or declinably as an adjectival participle, (31):

rec

ted

(31) Je viens en chantant (‘I come singing)’ La femme chantante vient (‘The singing woman comes’)

Un

cor

In the other Western Romance languages, Spanish and Portuguese, the two forms are still morphologically separate, but (a) the gerund (-ndo) has both retained its original gerundival functions (as a manner adverbial) and taken over all the verbal functions of the present participle as well, including all occurrences with periphrastic verbal

JB[v.20020404] Prn:24/11/2004; 14:41

F: LA7608.tex / p.20 (1283-1340)

 Io Manolessou

pan

y

paradigms; and (b) the still agreeing present participle (-nte) is wholly adjectival, lying outside the verbal system proper and unable to express any circumstantial notion.

. Slavic and Baltic

ofs

-

Joh

nB

enj

am

ins

Pub

lish

ing

Com

The Russian gerund (deeprichastiye) evolved out of the active present participle, and is, in the modern language a tenseless and nonfinite form (Babby & Franks 1998). First traces of this development appear already in Old Church Slavonic translations of the Gospels, and become more frequent in later texts (Vaillant 1964: 252–253). Examining the Bulgarian participle, Hult (1991: 100) offers an overview of the participle evolution in the Slavic languages. According to him, the gerund of Russian, Czech and Slovenian goes back to the Old Church Slavonic masculine/neuter nominative form of the present active participle, while the gerund of Ukrainian, Byelorussian, Polish, Slovak, Serbo-Croatian and Bulgarian to the oblique case stem of the same form. Taube (1981: 128–129) describes the environments where the breach of agreement between subject and participle, is most frequent in Old Russian texts (chronicles of the 15th c. AD. According to him, it is exceptional when the participle has adjectival (attributive) function, or is used as a predicate with a copula or an auxiliary verb; it is occasional when the participle is absolute and it is regular when the participle has an adverbial function. Although the Lithuanian participial system is more complex than that of other examined languages, the general evolution is similar: a language which did not originally possess gerund forms develops them out of active participles with adverbial meaning.22 As in Greek, the change starts from active neuter forms in the accusative, later spreading to masculine and feminine nouns and pronouns (Ambrazas 1990: 253). So the developments in the participle of other languages bear a hitherto unnoticed, but striking similarity to Greek: in the environments where the participle has a more strongly “verbal” meaning (i.e. imparts a temporal, causal etc. sense) it tends to lose its nominal characteristics and become a pure indeclinable adverbial, whereas in cases where its attributive function is strong it verges towards total loss of verbal features.23

ted

. Introduction

pro

. The evolution from participle to gerund

Un

cor

rec

The evolution described in Sections 2 and 3 constitutes a potentially fruitful testing ground for theories of syntactic change. It involves a radical change in a specific syntactic category – a change which takes place over a long period of time, appears in several languages, has not up to now been sufficiently investigated, and does not fall under the heading of grammaticalisation, the kind of change most commonly exam-

JB[v.20020404] Prn:24/11/2004; 14:41

F: LA7608.tex / p.21 (1340-1401)

From participles to gerunds 

Pub

lish

ing

Com

pan

y

ined in diachronic syntax. However, the restricted amount of interest it has generated is, to a certain extent, a limiting factor for its in-depth investigation. More specifically, as discussed in 2.1, there are no studies offering syntactic accounts of the complex AG participial system or the quantitative data necessary for them, and even the comparatively simpler MG gerund is an object of controversy (see Section 5); thus, neither the beginning nor the end of the process are at all securely established. Furthermore, none of the well-studied modern languages possesses a participial system similar to the AG one, and so the analyses covering the participles of other languages cannot apply to AG without considerable further elaboration. Moreover, current analyses of participles tend to focus on the question either of internal participial structure or of participles in auxiliary verb structures; thus they have very little to say on topics that are crucial for the evolution under investigation, which involves mainly participles in adverbial function.

. Previous accounts

Un

cor

rec

ted

pro

ofs

-

Joh

nB

enj

am

ins

The standard interpretation of the evolution of the Greek active participle is based on morphological considerations. According to this view (cf. Jannaris 1897: 206; Dieterich 1898: 206; Horrocks 1997: 122–124), the active participle, which belongs to the complex AG 3rd (consonant-stem) declension and shows widely differing endings according to person and tense, has limited learnability, in contrast to the passive participle, which belongs the much simpler 2nd declension (o-stem). The “unwieldy” active form, after going through a period of instability, as attested by numerous errors of agreement in papyri, ends up in total indeclinability, whereas the passive form is maintained throughout the history of Greek. There are a number of problems with this account. First, as shown in 2.2, the papyrological data are not conclusive: errors of agreement occur also with 2nd declension forms, and are much more common when the writer is not a native speaker of Greek – so the picture of declensional instability presented in Grammars of papyri cannot be relied upon without systematic textual investigations. However, the main difficulty with this account is that “indeclinability” is not an option for Greek, whose overall system depends on overt case differentiation; in cases of “difficult” declension patterns, the evolution is metaplasm, i.e. analogical re-formation according to a simpler inflectional pattern. Thus, as Hatzidakis (1928: 635) notes, the corresponding 3rd declension nt-stem nouns such as drako:n, gero:n, Kharo:n, have had a totally different evolution from the participle: they have not lost inflection, but have been reformed to 1st or 2nd declension ones (ðrakos, γeros/γerontas, Xaros), and the same goes for substantivised participles as well (arkho:n, patho:n > arxontas, paθos). And generally, the whole of the AG 3rd, consonant-stem, noun declension, has changed over to a simpler vowel-stem paradigm in MG, apart from certain subclasses (cf. Seiler 1958). Similarly, 3rd declension adjectives with comparable inflection patterns (e.g. s- or r-stems) also show break-down of agreement (cf. Gignac 1981: 138–141) in the papyri, without ever ending up in indeclinability. These too are re-formed analogically, fol-

JB[v.20020404] Prn:24/11/2004; 14:41

F: LA7608.tex / p.22 (1401-1466)

 Io Manolessou

lish

ing

Com

pan

y

lowing the vowel stem declensions (Browning 1983: 78), e.g. melas > melanos, ale:the:s > aliθinos etc. In other words, the process of indeclinability affects only participles, and not nouns and adjectives of similar inflectional patterns. The morphology-based interpretation of the participle > gerund evolution has the added disadvantage that it cannot account for two main claims made in this paper, namely that: (i) this evolution is not an exclusively Greek phenomenon, but, as demonstrated in Section 2, is to be found in several languages, thus precluding a Greek-specific causation and (ii) the parallel developments in the domain of the passive participle, i.e. the evolution participle > adjective (crucially, also not an exclusively Greek phenomenon) show that we are dealing with a larger change, which affects the whole participial and agreement system.

Pub

. Origins of the change

enj

All participles must have a subject of their own, due to theta-role considerations. This subject may be a full lexical noun phrase (in the case, e.g., of absolute participles) but is usually a null phrase (in the case of conjoined adverbial participles). When the subject of the participle is null, it is standardly assumed to belong to the category PRO (cf. Kester 1994; Alexandrova 1995), i.e. it is a null pronoun appearing as the subject of non-finite forms, which acquires reference through a control mechanism. Participles agree with their subjects. Gender and number features are transmitted from the subject to the participle. Case, however, depends on the syntactic function of the participle.

-

–

Joh

nB

– –

am

ins

If one rejects the standard view of gerund evolution, an alternative interpretation, not involving a direct morphological trigger and taking into consideration the crosslinguistic extent of the phenomenon is desirable. In this section, we will pursue a potential alternative interpretation, based on the syntactic mechanism of control in participial subjects. Some initial assumptions on the analysis of participles must be made:

ofs

On the basis of the above, the following analysis of AG participles can be attempted: Participles have three main functions, as described in 2.1:

ted

pro

I. Attributive modifiers, functioning as adjectives or relative clauses. II. Complements of verbs, functioning as non-finite complement clauses. III. Adverbial modifiers, functioning as clausal adjuncts.

Un

cor

rec

In case (I) participial syntax is relatively simple, as participles appear in the same syntactic positions as adjectives. Presumably, adjectival participles are located in some functional specifier within the noun phrase (DP). They therefore acquire their φfeatures in the same way as adjectives do, from spec-head agreement with the head noun. Attributive participles have a PRO subject, which is always co-indexed with the head of the nominal projection DP where the participle occurs. So: adnominal partici-

JB[v.20020404] Prn:24/11/2004; 14:41

F: LA7608.tex / p.23 (1466-1533)

From participles to gerunds 

Com

pan

y

ples obtain number, gender and case from the head of the nominal projection within which they are. In case (II), things are straightforward again: participial complements function like infinitival ones, with the added proviso that they agree with their subject. Presumably, participles are complements of VP, and acquire case from the main verb, (32). De:los e:mi erkhomenos. obvious be.1sg coming.nom.sg.masc.pass ‘I am seen to be coming.’ b. Eidon auton erkhomenon. see.1sg.aor he.acc.sg coming.acc.sg.masc.pass ‘I saw him coming.’ c. E:kusa autu: erkhomenu:. hear.1sg.aor he.gen.sg coming.gen.sg.masc.pass ‘I heard him coming.’

Pub

lish

ing

(32) a.

-

Joh

nB

enj

am

ins

Gender and number on the participle do not originate from the main verb, but from the participial subject. Case on the participial subject can be interpreted as originating either from the participle, once it has acquired it from the matrix verb, or as in regular raising/ECM constructions, where the participial subject gets case from the main verb as well. So, in case II, participles acquire two f-features from their subject (gender and number) and one from the verbal projection of which they are complements (case). If the participial subject is PRO, it can be controlled and co-indexed since it occurs in a c-commanded position. On the other hand, construction (III), namely, participles functioning as adverbial modifiers presents a number of difficulties. Most analyses agree that adverbial participles and gerunds are clausal adjuncts (see e.g. Kester 1994; Pires to appear). This follows from their free position in the clause, but also from semantic considerations: what they modify by their adverbial meaning (cause, consequence, aim, concession etc.) is a verbal action and not a noun. This becomes obvious in sentences containing sequences of participles, which all refer to the same subject, but which modify different verbal actions, (33): Eteteleute:kei pharmakon pio:n die.3sg.pluperf medicine having.drunk.nom.sg.masc pyresso:n. having.fever.nom.sg.masc ‘He died after having drunk a medicine, because he had a fever.’ (X.Anab.6.4.11) b. Hois pa:si khro:menoi krea which.dat.pl all.dat.pl using.nom.pl.masc meats.acc.pl epsontes e:sthion. roasting.nom.pl.masc eat.3pl.imperf ‘Using all the said wood, roasting meat, they ate.’ (X.Anab.2.1.6)

Un

cor

rec

ted

pro

ofs

(33) a.

JB[v.20020404] Prn:24/11/2004; 14:41

F: LA7608.tex / p.24 (1533-1590)

 Io Manolessou

Un

cor

rec

ted

pro

ofs

-

Joh

nB

enj

am

ins

Pub

lish

ing

Com

pan

y

Under this analysis, adverbial participles occupy a peripheral position within the verbal projection of the main clause, the same as adverbs do. This would be either an external adjunct to CP, VP or IP, or a special functional projection hosting the relevant adverbial function (cause, manner etc.) – cf. 4.3 below for discussion. Such a peripheral position, however, entails difficulties for the establishment of agreement and control between the participle and the matrix term to which it refers.24 “Conjoined” participles have a PRO subject, co-referential with a term in the matrix clause, usually the subject but not always. However, relationship between PRO and this term must work on a very loose basis: the structural position of PRO is such that it does not allow any formal mechanism of control or licensing – as argued above, the adverbial participle will be in the specifier of projection functioning as an adverbial adjunct: it is neither within a nominal projection, nor in a complement position. Furthermore, since the participle displays case, gender and number features, which are in fact the overt indications of its co-reference with a matrix term, one would have to assume that they have somehow been transmitted from this matrix term to the participial subject, and thence on to the participle. There is no other source for them, as adverbial adjuncts are not case-marked positions and as gender and number have to be identical with those of the matrix term. Now, assuming that participial subject is PRO, which by definition carries null case, and invariable gender and number features (3rd sing. masc. in the case of arbitrary reference) the agreement mechanism of adverbial participles becomes even more complex.25 On the other hand, absolute adverbial participles do not present any problems: they obtain gender and number from their subjects, which are full lexical noun phrases. And they acquire case (genitive) by default, so their position within the matrix clause is immaterial. No relationship with the matrix clause is established; the absolute participle is completely independent and its agreement mechanisms are local. From the perspective of agreement, therefore, the conjoined adverbial participle is a marked and complex construction, possessing a much simpler equivalent in the absolute participle. Crucially, as discussed in Sections 1 and 2, the history of both Greek and the Romance languages displays a growing preference for the absolute over the conjoined construction. This can be interpreted, following Roberts and Roussou (2002), as a normal result of the make-up of the human language learning device, which is conservative and has an inherent tendency for unmarked representations (marked in the sense of less economical, requiring additional movement or realisation of features). In the presence of two equivalent constructions, one marked26 and one unmarked, diachronic change will go in the direction of the unmarked option, and in the presence of less and less evidence to the contrary, the simpler structure will be attributed to all instantiations of the construction. In this particular case, the radical increase in absolute participles will in the end limit the input evidence for agreement between the participle and the matrix clause to such a degree that this will be in the end unlearnable to next generations of language learners.

JB[v.20020404] Prn:24/11/2004; 14:41

F: LA7608.tex / p.25 (1590-1656)

From participles to gerunds 

Com

pan

y

So, in the case of the participle > gerund evolution, language change can be seen to be inherently connected with language acquisition (for this cf. Kroch 2001). The change will not affect passive participles in the same way, as, due to their resultative meaning (Haspelmath 1994), they will much more frequently modify nouns than verbs, and will thus occur much more rarely in adverbial function.

. From participle to gerund

nB

enj

am

ins

Pub

lish

ing

In more detail, the evolution from participle to gerund can be viewed as a three-stage process. In the first, initial, stage, adverbial participles have an agreement requirement with a term in the main clause, usually the subject but not necessarily so. The alternative of the absolute participle exists in AG, even when the participial subject is co-referential to a term in the clause; so a “more economical” alternative was already in place in the language. In step II, the agreement requirement with the matrix clause is dropped, something which is mirrored in the radical increase of absolute participles. In this step, the participial clause is independent, and thus equivalent to an infinitival or subordinate clause. In this connection, one should mention the view of Mandilaras (1973: 352– 373), according to which the indeclinability of the active participle is caused by its confusion with the infinitive. He provides a considerable amount of evidence for an interchange of functions between infinitive and participle in the non-literary papyri. Examples include prepositional participles instead of prepositional infinitives, (34a), participles as complements of verbs which in AG were construed with the infinitive, (34b), and infinitive as complement of verbs which in AG were construed with the participle, (34c): Dia to eme metrio:s ekhonta. for the.acc.sg.neut i.acc.sg mediocrely having.masc.sg.acc.act ‘Because I am in mediocre health.’ (P. Lips.108.5–6, 3rd c. AD) b. Sy ikanos e: you.nom.sg capable.nom.sg.masc be.2sg dioiko:n. managing.masc.sg.nom.act ‘You are capable of managing.’ (P. Cair.Zen.59060.11, 257 BC) c. Tynkhaneis ekhein. happen.2sg have.inf.pres ‘You happen to have.’ (P. Grenf.ii.57.8, 168 AD)

ted

pro

ofs

-

Joh

(34) a.

Un

cor

rec

Of course, none of the “interchanged” constructions appearing in the papyri are found later, i.e. articular participles are very rare,27 and there are no verbs which prefer the participial over the AG infinitival syntax; however, there are several verbs which prefer the infinitival over the AG participial syntax (cf. Jannaris 1897: 493), something which goes hand-in-hand with the general reduction in the uses of the participle. What is

JB[v.20020404] Prn:24/11/2004; 14:41

F: LA7608.tex / p.26 (1656-1722)

 Io Manolessou

-

Joh

nB

enj

am

ins

Pub

lish

ing

Com

pan

y

more, the infinitive was also undergoing change and weakening in the same period. Especially the declarative infinitive (corresponding to oti (‘that’) complement clauses), had all but disappeared by the start of the medieval period, when the first instances of the gerund appear (cf. Jannaris 1897: 572–574; Mackridge 1997). Thus the confusion of participle and infinitive cannot have been the cause, but rather the result of the evolution of the former towards the gerund. An additional indication of the break-down of agreement between the matrix and the participial clause are the multiple examples of faulty agreement of the participle with the constituent immediately adjacent to it, and an indication of the more clearly verbal status of the participle is its use in conjunction with finite form (for both phenomena cf. Section 2.3). Step III. If the participle is a non-finite verbal element independent from the matrix clause, it need have no agreement at all – therefore, its form becomes fixed, in the form most frequently used. It is here that morphological considerations can play a role: not to motivate the freezing of a form, but to explain which particular form it will take. In this respect, one should bear in mind the evolutions described in Section 3 for Romance and Slavic: in each case, it was a different case-form of the present active participle that constituted the origin of the gerund; in some cases it was the nominative, in others the oblique form. The -onta ending had by then become the most frequent one because: (a) it was an accusative ending, and the accusative was by far the most common case, as it became more and more the default oblique case, due to the loss of the dative and the restriction of the genitive to adnominal usages (Horrocks 1997: 122); (b) conjoined participles, if oblique, would almost never be co-referent with a non-primary term in the clause, such as an indirect object or a participial complement (cf. Whaley 1990); and (c) the -onta ending had spread to the neuter singular as well, as described in 2.3. On a more speculative level, considering also the fact that the consonant-stem 3rd declension was passing over to the a-stem 1st declension due to its formal identity in the accusative case (cf. Hatzidakis 1892: 379–380), even the participial genitive ending, could, at some point in the Medieval period, have passed from -ontos to -onta, which would then be the unique oblique ending of a 1st declension form. The example below, could, e.g., be interpreted as an absolute participle in the genitive case, (35):

ted

pro

ofs

(35) Sy autos panta hyperthemenos e:ke, ekeinu: to you.nom.sg self everything putting.aside come, he.gen.sg the son ergon poiu:nta. your work doing.acc.sg.masc.act ‘Put everything aside and come yourself, letting him do your work.’ (P.Oxy.120, 4th c. AD)

Un

cor

rec

Another potential piece of evidence for the existence of absolute genitive participles in [-onta] is the existence of gerunds with genitive subjects in MedG and contemporary MG dialects such as Cypriot and Cretan, as in (20) and (21). A further contributing factor in the preference for the -onta oblique ending must have been the final vowel -a, which is an ending characteristic of adverbs: superlative

JB[v.20020404] Prn:24/11/2004; 14:41

F: LA7608.tex / p.27 (1722-1766)

From participles to gerunds 

ins

Pub

lish

ing

Com

pan

y

(arista ‘best’, takhista ‘fastest’) and irregular adverbs (mala ‘well’) end in -a since AG times, and this form becomes generalised for all adverbs in the early MedG period.28 The external similarity of the oblique masculine (and neuter) participle to an adverb must have promoted the possibility of its being interpreted as an indeclinable adverbial modifier (for this view cf. Hatzidakis 1928 and Horrocks 1997: 123). The adverbial interpretation of the participle was strengthened by the fact that, as was discussed in 2.2, it is possible, since the time of the Koine, to observe a gradual limitation in the uses of the participle – though not in its frequency –, as some of its meanings are gradually replaced by finite subordinate clauses.29 In any case, the adverbial use is considered, in standard Grammars, to be the main function of the participle even from AG times (Schwyzer 1950: 387). Summing up, the change from participle to gerund could be interpreted, for all the languages involved, as originating in the domain of the adverbial participle: this was first analysed as an independent participial clause without reference to the matrix clause, fulfilling its agreement requirement in a local domain, and then as a non-finite, exclusively verbal non-agreeing element, fixed in the most frequently appearing form.

am

. Implications for syntactic theory

Joh

nB

enj

The above account of the consecutive stages in the evolution of the gerund and the motivation behind it have repercussions on our conception of gerundival and participial syntax. The discussion does not aim to innovate in current views of Verb Phrase structure. Rather, it aims to tabulate the implications of the historical research conducted above for the extant proposals concerning the representation of MG gerunds (Rivero 1994; Tsoulas 1996; Tsimpli 2000; Sitaridou & Haidou 2002), and to constitute a stepping stone for further theoretical elaboration on this topic.

-

. The status and external syntax of MG gerunds

Un

cor

rec

ted

pro

ofs

The structural position standardly assumed for gerund clauses is that of adjuncts to one of the verbal projections of the matrix clause, e.g. VP or IP (Tsoulas 1996: 445; Spyropoulos & Philippaki 2001: 157; Pires to appear) or even CP (Tsokoglou & Kleidi 2002: 279). On the other hand, an analysis of gerunds as adverbial modifiers in line with current analyses of adverbs such as Alexiadou (1997) allows one to capture the connection between gerunds and adverbs on the one hand, and participles and adjectives on the other in a more systematic way. According to Alexiadou (1997), adverbs and adjectives are in reality manifestations of one and the same category, with standard similarities and differences. The adverbial ending -ly (or MG -a, -os) is in reality an agreement marker with the verb, equivalent to the case endings of the adjective, which mark agreement with the noun. A corresponding situation obtains with gerunds and participles in MG, i.e. the de-

JB[v.20020404] Prn:24/11/2004; 14:41

F: LA7608.tex / p.28 (1766-1829)

 Io Manolessou

nB

enj

am

ins

Pub

lish

ing

Com

pan

y

scription proposed for the nominal domain can also apply to the verbal one: the participle can be informally termed as the “verbal adjective”, and the gerund as the “verbal adverb”, with the ending -ontas being interpreted as an agreement marker with the verb. In more detail: Adjectives and adverbs form a single category. The category A can be realised either as an adjective or an adverb, depending on its location. Thus, if it is located in a verbal Specifier (Tense, Aspect etc., according to meaning) it is an Adverb. If it is located in a nominal functional Specifier (which hosts the agreement features of the noun phrase), it is an adjective. Both realisations of A enter into Spec-Head relationships with the corresponding lexical head, V or N, resulting in the first case in the licensing of the adverbial ending and in the other in agreement with the noun. Correspondingly, gerunds and participles also form a single category, which will differ in realisation according to its location. Thus, location in a verbal specifier means that the element is a gerund, whereas location in a nominal specifier means that the element is an attributive participle. Both realisations of this category will enter into Spec-Head relationships with the corresponding lexical head, resulting, in the case of participles, in full nominal Agreement, and for gerunds in the checking of the agreement ending -ontas. On the contrary, the AG-type adverbial participles are (as discussed in Section 3) located in a verbal specifier, where a Spec-Head relationship with a lexical head is unavailable, require a more complex mechanism to ensure agreement. As far as language history is concerned, the developments can be seen as a re-analysis of the adverbial participle, interpreted as a verbal element not requiring agreement, an evolution which avoids a situation where a lexical category has mixed nominal and verbal characteristics.

Joh

. The internal structure of Passive Participles

Un

cor

rec

ted

pro

ofs

-

There is doubt in the literature concerning whether Passive Participles are derived syntactically or lexically (cf. mainly Laskaratou & Philippaki 1984). In the former case, there is uncertainty concerning the number and nature of functional projections constituting them. The historical overview, showing as it does a drive towards an unambiguously nominal status, argues in favour of the lexical derivation of Passive Participles in certain cases, i.e. without an articulated internal verbal structure, since there is clear evidence of the loss of all verbal categories (Tense, Aspect, Voice). Thus, in the case of the so-called “lexical” passives (for terminology and criteria of differentiation cf. Anagnostopoulou 2001), all functional verbal projections, including Voice should be absent. This is evidenced by the existence of unaccusative and unergative “perfect participles”, which, crucially, started to appear only in the Medieval period, when the split between the “verbal” gerund and the “nominal” participle took place.30 However, since not all perfect participles have the same stative interpretation, a more articulated verbal structure, which includes a Voice and an Aspect Phrase, should

JB[v.20020404] Prn:24/11/2004; 14:41

F: LA7608.tex / p.29 (1829-1886)

From participles to gerunds 

pan

y

perhaps still be accepted for other instantiations of this form, along the lines proposed by Anagnostopoulou (2001), mainly for cases where the participle is accompanied by a Prepositional Phrase expressing the agent or an adverb expressing manner/time, in i.e. environments where the verbal, eventive reading of the participle is still possible.

Com

. The internal structure of gerunds

Un

cor

rec

ted

pro

ofs

-

Joh

nB

enj

am

ins

Pub

lish

ing

Taking the verb phrase from the top, the historical evidence speaks clearly against the postulation of a CP projection for gerunds. This was already proposed by Tsimpli (2000: 143–144), reviewing previous literature on the topic (Tsoulas 1996 and Philippaki-Warburton 1995). Her arguments were purely synchronic, namely that gerunds (a) cannot be introduced by conjunctions (b) disallow operator movement (e.g. extraction of wh-phrases)31 (c) resist nominalisation and (d) do not provide a landing position for topics and dislocated elements. Apart from these crucial arguments, the historical data also supports the rejection of a CP projection. As described in Section 1.1 above, AG participles were frequently introduced by complementisers, i.e., causal, final etc. conjunctions. This possibility gradually disappeared, and thus we have not only theoretical but also tangible evidence of the loss of the CP projection.32 Furthermore, if one adopts the proposal of interpretation of gerunds as adverbs along the lines proposed by Alexiadou (1997), there is no need to posit a CP projection responsible for its semantic interpretation (temporal, causal etc.) thanks to a covert operator, as Tsoulas (1996) suggests. The interpretation of gerunds is identical to that of adverbs: it depends on the functional verbal specifier they are located in. As a final point, the inability of MG gerunds to occupy argument positions can also be connected with their lack of a CP projection: CP has often been argued to be responsible for turning the verb phrase into a sentence, capable of occurring as a complement; this is the equivalent of noun phrase syntax, where only full DPs can be arguments (cf. Longobardi 1994).33 The issue of the existence of a Tense projection in Greek gerunds is harder to discuss, mainly because tense has a quite complex structure and is not a simple propositional operator, and also because its relationship to Aspect is not entirely clear; it can thus not be represented as just one projection responsible for the temporal interpretation of the clause. Furthermore, since Tense phrases are standardly viewed as the locus where subjects check case, this projection is closely connected with the licencing of subjects in clausal gerunds. Tsimpli (2000: 142) takes the gerunds’ context-dependency for temporal interpretation, as well as their inability to express Future, as evidence of lack of a proper Tense projection; on the other hand, context-dependency for temporal interpretation is not sufficient for the rejection of a Tense Phrase. Here a distinction should be made between null subject, obligatory control gerunds on the one hand, and nominative subject gerunds on the other. For the first type, the relevant literature in other languages in general accepts the absence of a Tense projection,34 and the same claim is made for Greek (Sitaridou & Haidou 2002: 585). Nominative subject clausal gerunds, however,

JB[v.20020404] Prn:24/11/2004; 14:41

F: LA7608.tex / p.30 (1886-1951)

 Io Manolessou

pan

y

require some device for the licensing of their subject. For Sitaridou-Haidou (2002) this device is a Tense projection, bearing semantic and not morphological tense, where the gerundival subject ends up after movement out of the VP. Its presence as a separate projection is supported by the co-occurrence of temporal adverbs with gerunds, (36):

ing

Com

(36) Pernontas xθes o Kostas to ðromo, ton xtipise crossing yesterday the Kostas.nom the street, he.acc.sg hit.3sg.aor ena aftocinito. a car.nom ‘As Costas was crossing the street yesterday, a car hit him.’

ins

Pub

lish

It is true that a movement of the subject from VP to the postulated TP would be invisible in most gerund phrases, as e.g. in the example above, although the subject is almost always post-verbal in MG gerunds.35 This surface order can be seen as the result of movement of the gerund to a higher projection, past the subject. However, in perfective gerunds, a subject having moved to a Tense projection should precede perfective infinitival forms (located in Aspect), something which does not happen; the subject follows all the modifiers of the gerund, i.e. does not seem to have moved at all, cf. (37):

enj

am

(37) Exontas pjia vareθi i Maria, aftoktonise. having already had.enough the Mary.nom, kill-oneself.3sg.aor ‘Having had enough by now, Mary killed herself.’

rec

Na apelθomen eðo is tin Romanian kursevonta, zimiononta. sub come.1pl here to the Romania plundering, destroying ‘To come here, to Greece, in order to plunder and destroy.’ (Chr. M.3648)

Un

cor

(38) a.

ted

pro

ofs

-

Joh

nB

Word order data therefore speak against a TP as the host of the subject in gerunds. Furthermore, without conclusive independent motivation for the projection of TP in certain types of gerunds (even if not disjoint temporal interpretation), it is difficult to interpret why, in identical environments, TP is sometimes projected and sometimes not. A further problem with Tense in MG gerunds appears when the evolution set forth in Section 2, which shows clearly that there are morphological reflexes of the reduction of tense in MG gerunds, is considered. More specifically, in the Medieval period, it was possible to express anteriority/perfectivity through a special “aoristic/perfective” form, which later disappeared. The absence of a TP in MG gerunds would be one way of explaining why perfective gerunds such as vlepsontas, γrapsontas (‘having seen’, ‘having written’) are no longer possible, as in MedG or the S. Italian dialects. Similarly, futurity is another temporal interpretation that was available for participles/gerunds in earlier phases of the language, but is impossible in MG. In Hellenistic (Mayser 1926: 170) and MedG it was possible to express posteriority/futurity/finality through the imperfective form, modifying verbs denoting motion, (38):

JB[v.20020404] Prn:24/11/2004; 14:41

F: LA7608.tex / p.31 (1951-2014)

From participles to gerunds 

pan

y

b. O Anθenor eðievike tus θeus proskinonta. the Anthenor.nom go.3sg.aor the gods.acc.pl doing.homage ‘Antenor went to do homage before the gods.’ (War of Troy 1489)

Un

cor

rec

ted

pro

ofs

-

Joh

nB

enj

am

ins

Pub

lish

ing

Com

This usage has disappeared in MG, even though the form of the gerund has not changed at all. Positing, therefore, an identical functional structure for earlier and later forms of the gerund cannot capture this differentiation. On the other hand, nominative case licensing in gerunds does require an explanation, which is most readily furnished by the postulation of a TP, as Sitaridou and Haidou (2002), and Pires (2001, to appear) propose. Earlier proposals, such as default nominative case for gerund subjects (Philippaki & Katsimali 1995; Alexiadou 1995; also accepted in Tsokoglou & Kleidi 2002) face the difficulty that they constitute a Greek-specific solution, although exceptions to subject control in gerunds are not a Greek-specific phenomenon: they occur in several languages, often dismissed as nonnormative by standard grammars (Haspelmath 1995: 30) or belonging to an earlier historical phase (Babby & Franks 1998: 487; Egerland 1999). Furthermore, Torrego’s (1998) study on adverbial infinitive clauses in several Romance languages shows that other non-finite forms in adverbial function may also exhibit nominative subjects. Turning now to less complex issues, on the basis of the historical evidence, a case can be made for the projection of the other verbal categories, namely Voice, Mood and Aspect. Voice, first of all, as was discussed in 4.2, is the main distinguishing characteristic between the gerund, which is unambiguously active, and the participle, which is not exclusively passive. As for Aspect, the process of evolution has shown that aspectual distinctions were lost, along with temporal ones; however, the 20th c. saw their re-establishment, with the novel evolution of periphrastic active and passive Perfect gerunds (exo γrapsi, exo γrafθi). The developments examined in 2.7 above, with the creation of the new perfective gerund forms both for the active and the passive voice, show that a category “Aspect”, hosting a gerundival form of the auxiliary verb, does indeed exist in MG, contrasting with the previous periods. Tsimpli (2000: 146) terms this category “PerfP”, considering it a functional projection containing “aspectual features”, while Rivero (1994) terms it simply Aspect Phrase; so, terminological distinctions aside, the recent literature seems to agree on the existence of an aspectual projection.36 The projection of a Mood Phrase in MG gerunds and its exact location is more controversial. For Philippaki (1995), the [-ontas] ending is a suffix belonging to an Agreement projection and not to Mood, because its placement in MoodP, which occurs above Negation, should block the occurrence of negative gerunds (mi γrafontas), something which does not happen. For Rivero (1994) and Roussou (2000: 86–88), Mood exists, but located below negation, and for Tsimpli (2000), the [-ontas] suffix belongs in Mood, but the surface word order is due to the raising of the Negation µην to Mood. Remaining agnostic on the question of the relative order of NegP and MP, what must be pointed out is that the very fact that gerunds are negated by min is an indi-

JB[v.20020404] Prn:24/11/2004; 14:41

F: LA7608.tex / p.32 (2014-2065)

 Io Manolessou

pan

y

cation for the existence of a Mood projection. Note crucially that this is a diachronic evolution, as in previous phases the Negation u: or ðen was also possible: AG participles allowed both kinds of negation, depending on the type of participle, and this was to a certain extent maintained in MedG as well, (39):37 U poseos to sinolon θelonta tu neg drink.gen.sg the total.acc.sg wanting the.gen.sg.neut γefθine. taste.inf ‘Not wanting to taste any drink at all.’ (Digenis G 4.381) b. ðen eγnorizontas o trisaθlios pjios ine. neg knowing the triply.miserable.nom who.nom be.3sg ‘Not knowing, the triply miserable, who it is.’ (Kechagioglou 2001: 316, 1632 AD)

Pub

lish

ing

Com

(39) a.

ins

It seems therefore that some syntactic mechanism is required in MG in order to account for the choice/stabilisation of the negation min.

am

. Conclusions

pro

ofs

-

Joh

nB

enj

This paper gives a detailed stage-by-stage exposition of the history of the Greek participle system. A new interpretation of gerund evolution is offered, taking into consideration (a) similar phenomena in other languages, showing that the transformation of participle to gerund is a cross-linguistically well-attested evolutionary path and (b) corresponding evolutions in the domain of the passive participle. Gerund evolution is viewed under the perspective of a gradual specialisation of an originally mixed category, possessing both verbal and nominal features, to a purely verbal category. Similarly, the evolution from passive participle to participial adjective is interpreted as the gradual specialisation of the same mixed category to a purely nominal one. The motivation behind the evolution towards the gerund is partly attributed to the complex agreement requirements of the adverbial participle. Finally, the implications of the diachronic data for our conception of participial syntax were discussed.

Notes

rec

ted

* I would like to dedicate this paper to Prof. Dimitra Theophanopoulou-Kontou, who was my first teacher in linguistics, my supervisor on behalf of the Greek State Scholarships Foundation, and, more importantly, my first model of scholarly integrity. I would also like to thank profs. Despina Cheila-Markopoulou and Amalia Moser for their comments on this paper.

Un

cor

. This development concerns only Standard MG; some dialects have followed diverging paths, which range from extension of the active participle to Present tense periphrases as well (Tsako-

JB[v.20020404] Prn:24/11/2004; 14:41

F: LA7608.tex / p.33 (2065-2139)

From participles to gerunds 

y

nian) to its total elimination (Asia Minor dialects) – for this cf. Mirambel (1961: 70–77) and Nakas (1991: 188–189).

ing

Com

pan

. The English gerund has a double status, both nominal (“the giving of gifts”) and verbal (“I hate giving gifts”) – for an overview of current treatments of this “mixed” status cf. Hudson (2000). On the contrary, the MG form has no nominal features, corresponding rather to the French gérondif. In more detail, the MG gerund does not possess any of the following properties (from Blevins 1994): (a) nominal inflection- genitive and plural forms (b) argumental function (subject, object, predicate) (c) determination through articles (d) adnominal modifiers such as genitive DPs. The MG gerund is uninflected and takes only verbal complements – nominative subjects, accusative objects and adverbial modifiers, cf. the references in Section 1.7. In this respect, Moser (2002: 111) is correct in doubting the applicability of the term “gerund” to the MG form; its is maintained here because of the wider cross-linguistic scope of the discussion.

lish

. Mainly: Jannaris (1897: 504–506), Hatzidakis (1924, 1927, 1928), Mirambel (1961), Mandilaras (1973: 352–373), Horrocks (1997: 120–124, 228–229).

ins

Pub

. Works treating AG participles (Mugler 1938; Oguse 1962; cf. also Jimenez 1987 and the references in Schwyzer 1950: 385) do not provide the kind of quantitative information that would be helpful to the present investigation. Also, large-scale computerised tagging of AG texts does not exist, and for later texts nothing is at all available, with the exception of the New Testament. Studies on the NT sometimes also provide comparative data for AG authors; cf. e.g. Fisher (1989).

enj

am

. For morphology and inflection tables cf. Smyth (1965: 81–85). The AG participle has separate forms for Present, Aorist, Future and Perfect. The difference between Present and Aorist is not temporal, but aspectual (imperfective vs. perfective). If one interprets the Perfect as also expressing aspect, and the Future as belonging to the field of modality, it could be argued that the AG participle does not have tense.

-

Joh

nB

. The term “spoken language” requires some qualification. It is generally accepted that AG literature is mostly written in literary dialects or in an high register, not representative of everyday spoken language. However, it is possible to find testimonies of colloquial language in certain types of texts – cf. Dover (1987) for examples. The medical treatises discussed here, written in a non-literary, factual prose style, are usually considered as containing a higher concentration of colloquial features than standard literature. . Gignac (1981: 46–47), Horrocks (1997: 66–67), cf. also Note 15 below for the reverse phenomenon.

pro

ofs

. Note that our evidence for this period comes almost exclusively from Egyptian papyri, mostly written by non-native speakers, whose command of the language may have been imperfect; so the picture emerging from participle use in papyri may not be representative of Greek in general, esp. since neither case nor adverbial participles exist in Egyptian (Schwyzer 1950: 386; Loprieno 1995: 55–56, 87).

rec

ted

. Cf. Mandilaras (1973: 356–358, 369), Mayser (1934: 67–70), and Soliman (1965: 103–105) for the papyri, de Foucault (1972: 173) for the literary Koine and esp. Blass and Debrunner (1961: 423) for the NT: “The NT authors tend to make the participial clause independent and to prefer the absolute construction. . . where a classical author would not have admitted it even as a special license.”

Un

cor

. Mendez-Dosuna (2000: 290–291) adduces another reason to consider the -onta ending as stemming from the masculine: gerunds are mainly subject oriented, and so are more likely to

JB[v.20020404] Prn:24/11/2004; 14:41

F: LA7608.tex / p.34 (2139-2198)

 Io Manolessou

Com

pan

y

have developed out of masculines, which have higher agentivity than neuters. An anonymous reviewer has suggested that the analogical presence in -onta forms of the final -n of the acc. sg. which appears in other masc. inflectional paradigms would provide additional proof that these are masculine forms. Such cases do exist (e.g. fenome kaθaran amfierosin. . . piontan = “I declare I am making a clear dedication”, Guillou Oppido, doc.24, 1053); however, the presence of the final -n is not significant, since in MedG there was a tendency for it to extend to any wordclass ending in a vowel, including neuter nouns (e.g. praγman, γramman), cf. Minas (1994: 62), Jannaris (1897: 544).

lish

ing

. List of the examples in Malalas in Wolf (1911: 54), additional instances in Hatzidakis (1892: 144) and Jannaris (1897: 207); cf. also Gelzer (1893: 198), Reinhold (1898: 57–58, Addenda), Mitsakis (1967: 158–159). Note that instances of the -onta ending often do not appear in the text of the edition of such early texts, but only in the apparatus criticus. This makes them difficult to locate, as they will not be mentioned in any index or introduction and, more importantly, will never be hit upon by a computer search, even when a machine-readable version the text exists.

Pub

. CD-ROM PHI #7, Greek Documentary Texts, Packard Humanities Institute: -onta in CHR0013 (christian inscriptions of mainland Greece) and CHR0010 (christian inscriptions of Asia Minor).

Joh

nB

enj

am

ins

. One example of an [-onta] masculine participle occurs in Malalas, 29.43: jynekoðis exon frenas ce polypraγmonunta (= having.nom.sg. masc a female mind and being-meddlesome.acc. sg.masc.), cf. Wolf 1911.II.28. However, in opposition to the neuter [-onta] forms, which appear in equal proportion with the “correctly” inflected forms in this text, this masculine participle is an isolated case, contrasting with hundreds of “correct” forms, and could thus be considered as a scribal intervention. There is also the single example of a feminine form in -onta in the NT, ikusa fonin mian. . . leγonta Apoc.9.13–4 (= I heard one voice.acc.sg.fem. . . saying.acc.sg.masc, but (a) there are variant readings with the feminine form (BDF §136 n.3) and (b) this need not be a gerund form, but an agreeing participle, i.e. one more simple case of a masculine form instead of a feminine one, as occurs several times in the Revelation (4.1, 5.12, 11.15 etc.). So Hatzidakis (1928: 636) correctly rejects this example as the earliest instance of the gerund.

pro

ofs

-

. It is thus a misconception to consider early, Hellenistic, forms of a non-agreeing participle in [-ontas] as first attestations of the MG situation, since the [-s] suffix on the [-onta] ending appears only during the later MedG period. So, the examples quoted by Mandilaras (1973: 358) following Kapsomenakis (1938: 40) are not early precursors of the MG gerund: pantes ofilomen stefaniforuntas ce vuθituntas. . . iðene P.Oxy.1021.14–8, 54 AD, kata ton polemon symmaçisantes ce panijirizontas P.Oxy.705.33–5 (200 AD). These are nominative plural endings, a hypercorrect reaction against the -es ending which had begun to replace -as in the accusative, and are thus unrelated to the history of the participle.

ted

. Note crucially that Haspelmath (1994) considers that the formation of past passivemorphology participles from active-form intransitive and unaccusative verbs is a universal tendency, due to the resultative semantics associated with such verbs.

rec

. Cf. Minas (1994: 176–177) for examples of complement and attributive use in MedG S. Italian documents.

Un

cor

. The Aorist ending being reformed from [-anta] to [-onta] under the analogical influence of the Present.

JB[v.20020404] Prn:24/11/2004; 14:41

F: LA7608.tex / p.35 (2198-2263)

From participles to gerunds 

pan

y

. Thus, in the northern dialects, the vestiges of the present participle in -amenos and -umenos are exclusively attributive or substantivised: l’paminus ‘compassionate’ (< lipame ‘to be sorry’), ixuminus ‘rich’ (< exo ‘to have’), pirazaminus ‘passer-by’ (< perno ‘to pass’) etc. (Papadopoulos 1926: 96–97), and the same is valid for the Cypriot and the Rhodian dialects as well (Menardos 1925a: 66–67; Papachristodoulou 1958: 75–76).

Com

. Pace Tsoulas (1996) who accepts argumental readings. Examples like θimiθika ton Kosta oðiγontas to aftocinito (1996: 462) can only mean ‘I remembered Costas as I was driving’ and not ‘I remembered Costas driving’. For this cf. the discussion in Spyropoulos and Philippaki (2001: 158).

lish

ing

. The participial clauses in -omenos proposed in Tsokoglou and Kleidi (2002) are a different case: they concern not the resultative passive past participles discussed here, but relic present passive participles such as erxomenos, erγazomenos, which constitute a small non-productive subset within MG.

Pub

. For this cf. in detail: Svennung (1935: 425–432), Leumann, Hoffmann and Szantyr (1965: 383–384, 389, 392) and, more recently, Bauer (1993) and Egerland (1999). . Cf. Ambrazas (1990: 261): “Die Partizipien, die eine verallgemeinerte adverbiale Bedeutung genommen haben, lösen sich ganz vom Kasusparadigm und gewinnen die für Adverbien typische unflektierte Form”.

enj

am

ins

. All the languages examined here belong to the Indo-European family; in order to ascertain whether the phenomenon appears in other families as well, further research is required. Cf. for example Gordon (1982: 3), who, examining the history of the Hebrew participle claims that it “undergoes polarization; from an intermediate form with some nominal and some verbal qualities in Biblical Hebrew, it develops to become, in Modern Hebrew, either clearly nominal or clearly verbal in any given sense”.

Joh

nB

. Cf. Tsokoglou and Kleidi (2002) on the cases of adjective agreement within participial clauses in MG, where it is argued that such agreement is achieved not structurally, but through co-reference with the subject PRO which is in turn co-referent with the matrix subject: a multistep semantic path.

-

. A potential solution would be to consider, following Torrego (1998), that the subjects of non-finite forms displaying agreement (in her case, mostly inflected infinitives) are not PRO but pro. However, pro is inherently nominative in Torrego’s examples, something that is not valid for AG participial subjects.

pro

ofs

. From another viewpoint as well, the AG participle is a much more marked construction than the gerund: it possesses overt features belonging both to the nominal and to the verbal domain, whereas the gerund is a purely verbal category. The evolution of the passive participle towards adjectivisation can also be viewed under the perspective of a marked “mixed” category evolving into a purely nominal one.

ted

. There are some examples in the apocryphal Gospels, e.g. ðia to mi ekðoton emaftin piisasa Acta Thomae 215.21, Ljungvik (1926: 55), but these disappear in later texts, and indeed the articular infinitive is a very productive construction in MedG, cf. Mackridge (1997) for details of this construction.

rec

. For this cf. Hatzidakis (1892: 52 Anm.3), Jannaris (1897: §518b).

Un

cor

. See also Kurzova (1997) who sees the replacement of the participle through finite clauses within the general framework of a change from a system of paratactic adjuncts to one of clearly defined subordination.

JB[v.20020404] Prn:24/11/2004; 14:41

F: LA7608.tex / p.36 (2263-2324)

 Io Manolessou

Com

pan

y

. The split evolution of active gerund vs. non-active (not just passive) participle lends credibility to recent proposals (Kratzer 1996; Arad 1999) concerning the status of external arguments, which are argued to belong to a Voice projection and not to the VP proper. The Greek gerund, being a verbal category, can be interpreted as possessing the category Voice, which thus licenses an Agent; the gerund is therefore active. The participle, on the other hand, having lost all verbal categories, has no Voice and no external argument. Therefore, any verb which is not active (be it passive, unaccusative or unergative) might form “passive participles”. . Cf. similar tests in Stowell (1982) for English.

lish

ing

. The arguments in Katsimali (2003) against the existence of a CP projection in AG participial clauses are dubious: the fact that AG adverbial participles are not accompanied by complementisers introducing object clauses, such as hoti or hina, is due to their meaning and their function in the clause, and not to the absence of a CP projection. Furthermore, the optional appearance of a complementiser does not entail the absence of a CP projection, as e.g. in the English complement clauses optionally introduced by that.

ins

Pub

. In the latest papers asserting the existence of CP in Greek gerunds (Roussou 2000 and, following her, Sitaridou & Haidou 2002), the projection argued for is in fact a kind of Mood projection and has no bearing on the discussion in the present paragraph. Cf. also Pires (to appear) for additional arguments (involving indirect questions and complementisers) against a CP projection in English clausal gerunds.

nB

enj

am

. Cf. Alexandrova (1995) for Bulgarian, Babby and Franks (1998) for Russian, Pires (2001) for English. The seminal discussion of Stowell (1982) also argues against a Tense projection in gerunds, based on the internally determined future interpretation of infinitives as compared with the “completely malleable” temporal interpretation of the gerund, which is determined externally by the semantics of the control verb. The opposite view is defended in Tsokoglou and Tsimpli (2002), who argue in favour of a TP projection in MG gerunds based on examples with independent temporal adverbs, e.g. Fevγontas noris simera to proi, θa ise sto Parisi avrio ‘Leaving early this morning, you’ll be in Paris tomorrow’.

Joh

. The exceptions are topicalised or focused subjects, cf. Rivero (1994). . Cf. Moser (2002) for the most recent account of the semantics of aspect in MG gerunds.

ofs

-

. For participle negation in Classical and Hellenistic Greek cf. Moorhouse (1948), and for later developments cf. Jannaris (1897: 430–432), Landsman (1988–1989: 26) and references cited therein.

pro

References

Un

cor

rec

ted

Aerts, W. J. (1965). Periphrastica. Amsterdam: A. Hakkert. Alexandrova, G. (1995). Participial clauses in Bulgarian, Italian and Spanish: Argument structure, agreement and case. In K. Zagona (Ed.), Grammatical Theory and the Romance Languages. Selected papers from the 25th Linguistic Symposium on Romance Languages (LSRL XXV) Seattle, 2–4 March 1995 [Current Issues in Linguistic Theory 133] (pp. 1–12). Amsterdam: John Benjamins. Alexiadou, A. (1995). Subject positions in Modern Greek. Studies in Greek Linguistics, 16, 242– 253.

JB[v.20020404] Prn:24/11/2004; 14:41

F: LA7608.tex / p.37 (2324-2451)

From participles to gerunds 

Un

cor

rec

ted

pro

ofs

-

Joh

nB

enj

am

ins

Pub

lish

ing

Com

pan

y

Alexiadou, A. (1997). Adverb Placement: A case study in antisymmetric syntax [Linguistik Aktuell/Linguistics Today 18]. Amsterdam: John Benjamins. Alexiadou, A. & E. Anagnostopoulou (1999). Tests for unaccusativity in a language without tests for unaccusativity. In A. Moser (Ed.), Greek Linguistics ’97. Proceedings of the 3rd International Conference on the Greek language (pp. 23–31). Athens: Ellinika Grammata. Ambrazas, V. (1990). Vergleichende Syntax der baltischen Partizipien. Vilnius: Academy of Sciences, Inst. of Language and Literature. Anagnostopoulou, E. (2001). On the distinction between verbal and adjectival passives: Evidence from Greek. Paper presented at the Tübingen Workshop on Participles. Arad, M. (1999). On little v. MIT Working Papers in Linguistics, 33, 1–25. Babby, L. & S. Franks (1998). The syntax of Adverbial Participles in Russian Revisited. Slavic and East European Journal, 42(3), 483–515. Bauer, B. (1993). The coalescence of the participle and the gerund/gerundive: An integrated change. In H. Aertsen & R. J. Jeffers (Eds.), Historical Linguistics 1989 (pp. 59–71). Amsterdam: John Benjamins. Blass, F. & A. Debrunner (1961). A Greek Grammar of the New Testament and Other Early Christian Literature. Transl. by R. W. Funk. Cambridge: Cambridge University Press. Beshevliev, V. (1963). Die protobulgarischen Inschriften. Berlin: Akademie Verlag. Bees, N. (1910). Über die Konstruktion von diaferein (gehören) mit dem Genetiv. Glotta, 2, 118–124 & 300. Blevins, J. (1994). A lexicalist analysis of gerundive nominals in English. Australian Journal of Linguistics, 14, 1–38. Browning, R. (1983). Medieval and Modern Greek. Cambridge: Cambridge University Press. Cheila-Markopoulou, D. (2003). Protasiakotita ce metoçices ðomes stin aleksanðrini ce meseoniki ellinici (Sentence and participial structure in Alexandrian and Medieval Greek). In D. Theophanopoulou-Kontou et al. (Eds.), Sinxrones tasis stin ellinici γlosoloγia. Meletes afieromenes stin Irini Philippaki-Warburton (Curent trends in Modern Greek Linguistics. Studies dedicated to Irini Philippaki-Warburton) (pp. 128–143). Athens: Patakis. Dieterich, K. (1898). Untersuchyngen zur Geschichte den griechishen Sprache von den hellenistischen Zeit bis zum 10. Jhrh. N. Chr [Byzantinisches Archiv., Heft 1]. Leipzig: Teubner. Dover, K. J. (1987). The colloquial stratum in classical Attic prose. In K. J. Dover, Greek and the Greeks, Collected papers (Vol. 1) (pp. 16–30). Oxford: Blackwell. Egerland, V. (1999). Sulla sintassi delle costruzioni assolute participiali e gerundive nell’ italiano antico ed il concetto di anacoluto. Revue Romane, 34(2), 181–204. de Foucault, J. A. (1972). Recherches sur la langue et le style de Polybe. Paris: Belles Lettres. Fisher, W. F. (1989). The Participle in the Greek Pentateuch: A descriptive analysis and comparison to New Testament usage. Ph.D. Dissertation, SW Baptist Theological Seminary. Frisk, H. (1928). Participium und verbum finitum im Spätgriechischen. Glotta, 17, 56–66. Gelzer, H. (1893). Leontios von Neapolis, Leben des heiligen Iohannes des Barmherzigen, Erzbischofs von Alexandrien. Freiburg: Akademische Verlagsbuchhandlung. Giakoumaki, E. (1993). Iðiazuses sintaksis rimaton tis kriticis ðialektu (Special syntax of verbs in the Cretan dialect). Lexicografikon Deltion, 18, 39–53. Gignac, F. T. (1981). A Grammar of the Papyri of the Roman and Byzantine Periods (Vol. II), Morphology. Milano: Cisalpino-Goliardica. Gordon, A. (1982). The development of the participle in Biblical, Mishnaic and Modern Hebrew. Afroasiatic Linguistics, 8(3), 122–179. Harris, M. (1978). The Evolution of French Syntax: A comparative approach. London: Longman.

JB[v.20020404] Prn:24/11/2004; 14:41

F: LA7608.tex / p.38 (2451-2567)

 Io Manolessou

Un

cor

rec

ted

pro

ofs

-

Joh

nB

enj

am

ins

Pub

lish

ing

Com

pan

y

Haspelmath, M. (1994). Passive participles across languages. In B. Fox & P. J. Hopper (Eds.), Voice. Form and function (pp. 151–177). Amsterdam: John Benjamins. Haspelmath, M. (1995). The converb as a cross-linguistically valid category. In M. Haspelmath & E. König (Eds.), Converbs in Cross-Linguistic Perspective: Structure and meaning of adverbial verb forms – adverbial participles, gerunds (pp. 1–55). Berlin: Mouton de Gruyter. Hatzidakis, G. N. (1892). Einleitung in die Neugriechische Grammatik. Leipzig: Breitkopf and Härtel. Hatzidakis, G. N. (1924). Paθitice metoçe meta enerγiticis simasias (Passive participles with active meaning). Athena, 36, 197–199. Hatzidakis, G. N. (1927). Peri tis metoçis tu paθitiku parakimenu en ti meseonici ce nea Ellinici (On the perfect passive participle in Medieval and Modern Greek). EEFSPTh, 1, 3–34. Hatzidakis, G. N. (1928). Simvoli is tin istorian tis ellinicis γlosis. Peri ton metoxon (Contribution to the history of the Greek language. On participles). Proceedings of the Academy of Athens, 3, 634–645. Hatzidakis, G. N. (1934). Parektetamena moria tis neas ellinicis (Suffixed particles of Modern Greek). In G. N. Hatzidakis, Γlosoloγice erevne A’ (pp. 223–226). Athens: Academy of Athens. Horrocks, G. (1997). Greek: A history of the language and its speakers. London: Longman. Hudson, R. (2000). Gerunds and multiple default inheritance. UCL Working Papers in Linguistics, 12, 417–449. Hult, A. (1991). On the Development of the Present Active Participle in Bulgarian [Commentationes Slavicae Gothoburgenses 6]. Göteborg: University of Göteborg. Iordanidou, A. (1985). La diglossie en Grèce: Étude d’un cas précis, le participe. Ph.D. Dissertation, Université de Paris VII. Jannaris, G. (1897). An Historical Greek Grammar, Chiefly of the Attic Dialect. London: MacMillan. Jimenez, M. A. D. (1987). La distribucion de participio concertado y subordinada adverbial en la lengua literaria griega del s. V a.C. Emerita, 55(1), 51–62. Kapsomenakis, S. (1938). Voruntersuchungen zu einer Grammatik der griechischen Papyri. Beiträge zur Herstellung und Deutung einzelner Texten. München. Karanastasis, A. (1997). Γramatici ton elinikon iðiomaton tis Kato Italias (Grammar of the Greek dialects of Southern Italy). Athens: Academy of Athens. Karla, G. (2002). Metafrazontas metoçes se ðimoði loγo to 1600 (Translating participles in colloquial language in 1600). In P. Agapitos & M. Pieris (Eds.), T aiðonin cinon pu γlika θlivate. . . Acts of the 4th International Conference Neograeca Medii Aevi (November 1997, Nicosia) (pp. 185–205). Herakleion: Panepistimiakes Ekdoseis Kritis. Katsimali, G. (2003). Xalari ipotaksi. I periptosi tis metoçis se γenici apoliti (Loose subordination. The case of the genitive absolute participle). In D. TheophanopoulouKontou et al. (Eds.), Sinxrones tasis stin ellinici γlosoloγia. Meletes afieromenes stin Irini Philippaki-Warburton (pp. 144–157). Athens: Patakis. Katsoyannou, M. (1995). To rima sto eliniko iðioma tis Kalavrias (The verb in the Greek dialect of Calabria). Studies in Greek Linguistics, 15, 542–553. Kavˇciˇc, J. (2001). On the use of the participle functioning as a finite verbal form in the chronicle of John Malalas. Linguistica, 41, 77–92. Kechagioglou, G. (2001). Pezoγrafici anθoloγia. Afiγimatikos γraptos neoelinikos loγãïs [Prose Anthology. Narrative Modern Greek prose Vol. 1]. Thessaloniki: Manolis Triantafyllidis Foundation.

JB[v.20020404] Prn:24/11/2004; 14:41

F: LA7608.tex / p.39 (2567-2685)

From participles to gerunds 

Un

cor

rec

ted

pro

ofs

-

Joh

nB

enj

am

ins

Pub

lish

ing

Com

pan

y

Keil, J. & A. Wilhelm (1931). Monumenta Asiae Minoris Antiqua (Vol. III), Denkmäler aus dem Rauhen Kilikien. Manchester: Manchester University Press. Kester, H. P. (1994). EUROTRA Grammar: Adnominal participles and infinitives in Germanic and Romance languages. Torino: Edizioni dell’ Orso. Klaffenbach, G. (1933). Book review of Keil and Wilhelm (1931). Deutsche Literaturzeitung (12/3/1933), Heft, 11, 492–499. Kratzer, A. (1996). Severing the external argument from its head. In J. Rooryck & L. Zaring (Eds.), Phrase Structure and the Lexicon (pp. 109–137). Dordrecht: Kluwer Academic. Kroch, A. (2001). Syntactic change. In M. Baltin & C. Collins, The Handbook of Contemporary Syntactic Theory (pp. 699–729). Oxford: Blackwell. Kühner, R. & C. Stegmann (1966). Ausführliche Grammatik der lateinischer Sprache (Teil II, Satzlehre). Hannover: Hahnsche Buchhandlung. Kurzova, H. (1997). Notes on Greek participial constructions. In E. Banfi (Ed.), Studi di linguistica greca II (pp. 107–127). Milano: Franco Angeli. Langholf, V. (1977). Unmovierte Partizipien im Griechischen. Hermes, 105, 290–307. Landsman, D. M. (1988–1989). The history of some Greek negatives: Phonology, grammar and meaning. Glossologia, 7–8, 13–31. Laskaratou, G. & I. Philippaki (1984). Lexical versus transformational passives in Modern Greek. Glossologia, 2–3, 99–109. Leumann, M., J. B. Hofmann, & A. Szantyr (1965). Lateinische Grammatik (Bd. 2), Lateinische Syntax und Stilistik. München: Beck. Ljungvik, H. (1926). Studien zur Sprache der apokryphen Apostelgeschichten. Uppsala: Lundequistska Bokhandeln. Lindstedt, J. (2002). Is there a Balkan verb system? Balkanistika, 15, 323–336. Longobardi, G. (1994). Reference and proper names: A theory of N-movement in syntax and logical form. Linguistic Inquiry, 25, 609–665. Loprieno, A. (1995). Ancient Egyptian. A linguistic introduction. Cambridge: Cambridge University Press. Mackridge, P. (1997). The medieval Greek infinitive in the light of dialectal evidence. In K. Konstantinides et al. (Eds.), ΦIΛEΛΛHN. Studies in honour of R. Browning (pp. 191–204). Venice: Istituto ellenico di studi bizantini e postbizantini. Mandilaras, B. (1973). The Verb in the Greek Non Literary Papyri. Athens: Ministry of Culture. Mayser, E. (1926). Grammatik der griechischen Papyri aus der Ptolemäerzeit (Bd.I, 1926), Formenlehre (Bd.II.2, 1934), Satzlehre (Bd.II.3, 1935). Berlin: De Gruyter. Menardos, S. (1925a). Peri tis ðiaθeseos ton cipriakon rimaton (On the the aspecti of the Cypriot verb). EEFSPA, 1, 75–86. Menardos, S. (1925b). Cipriaci γramatici. Rimata (Cypriot Grammar. Verbs) Athena, 37, 35–79. Mendez-Dosuna, J. (2000). Book review of Horrocks (1997). Journal of Greek Linguistics, 1, 274– 295. Minas, K. (1994). I γlosa ton ðimosievmenon meseonikon elinikon engrafon tis Kato Italias ce Sicelias (The language of the published Medieval Greek documents of Southern Italy and Sicily). Athens: Academy of Athens. Mitsakis, K. (1967). The language of Romanos the Melodist. München: Beck. Mirambel, A. (1961). Participe et gérondif en grec medieval et moderne. Bulletin de la Societé Linguistique de Paris, 56, 46–79. Moorhouse, A. C. (1948). On negativing Greek participles, where the leading verbs are of a type to require µÞ. Classical Quarterly, 42, 35–40.

JB[v.20020404] Prn:24/11/2004; 14:41

F: LA7608.tex / p.40 (2685-2796)

 Io Manolessou

Un

cor

rec

ted

pro

ofs

-

Joh

nB

enj

am

ins

Pub

lish

ing

Com

pan

y

Moser, A. (1988). A History of the Greek Perfect Periphrases. Ph.D. Dissertation, University of Cambridge. Moser, A. (2002). Metoçes, γerunðia, -ontas. Xronos ce apopsi (Participles, gerunds, -ontas. Tense in perspective). In Ch. Clairis (Ed.), Recherches en linguistique grecque. Actes du 5e colloque international de linguistique grecque. Sorbonne, 13–15 Septembre 2001 (Vol. II) (pp. 111–114). Paris: L’Harmattan. Mugler, C. (1938). L’evolution des constructions participiales complexes en grec et en latin. Strasbourg. Muñio Valverde, J.-L. (1995). El gerundio en el espanol medieval (s.XII– XIV). Malaga: Libreria Agora. Nakas, A. (1985). Metoçika, i pos xrisimopiun ti metoçi o Seferis, o Makrijianis, o Elitis (Participials, or how Seferis, Makrijianis and Elitis use the participle). In A. Nakas (Ed.), Γlosofiloloγika (pp. 136–174). Athens: Kalvos. Nakas, A. (1991). Γia ti metoçi ce to rimatiko epiθeto se -tos opos emfanizete stin Cini Nea Ellinici ce tis ðialektus (On the participle and the verbal adjective in -tos as it appears in Standard Modern Greek and its dialects). In B. A. Nakas (Ed.), Γlosofiloloγika (pp. 177–204). Athens: Smili. Oguse, A. (1962). Recherches sur le participe circonstanciel en grec ancien. Wetteren. Papachristodoulou, Ch. (1958). Morfoloγia ton roðitikon iðiomaton (Morphlogy of the dialects of Rhodes). Dodekanisiakon Arxion, 3, 9–106. Papadopoulos, A. (1926). Γramatici ton vorion iðiomaton tis neas ellinicis γlosis (Grammar of the Northern dialects of the Modern Greek language). Athens: Sakellariou. Papadopoulos, Th. (1977). Nikolau Sofianu, γramatici tis cinis ton Ellinon γlosis (Nikolau Sofianu, Grammar of Standard Modern Greek). Athens: Kedros. Petersmann, H. (1979). Zur Entwicklungsgeschichte der motionslosen Partizipia im Griechischen. Die Sprache, 25, 144–166. Philippaki-Warburton, I. (1995). Diaxronici θeorisi tis θesis ton englitikon mesa stin protasi (A diachronic approach to the position of clitis in the sentence). Studies in Greek Linguistics, 15, 123–134. Philippaki-Warburton, I. & G. Katsimali (1995). Control in Ancient Greek. In G. Drachman et al. (Eds.), Greek Linguistics ’95 (pp. 577–598). Salzburg. Pires, A. (2001). Clausal and TP-defective gerunds: Control without tense. Preceedings of NELS, 31, 389–406. Pires, A. (to appear). Clausal gerunds in minimalism. Univ. of Maryland Working Papers in Linguistics 12 (Special volume on Control and Binding). Reinhold, H. (1898). De graexitate patrum apostolicorum librorumque apocryphorum Novi Testamenti quaestiones grammaticae. Ph.D. Dissertation, Halle. Rivero, M.-L. (1994). Noun movement in the languages of the Balkans. Natural Language & Linguistic Theory, 12, 63–120. Roberts, I. & A. Roussou (2002). The history of the future. In D. Lightfoot (Ed.), Syntactic Effects of Morphological Change. Oxford: Oxford University Press. Roussou, A. (2000). On the left periphery: Modal particles and complementisers. Journal of Greek Linguistics, 1, 65–94. Rupprecht, K. (1926). Zwei Probleme der griechischen Syntax. 1: Der Gebrauch maskuliner Formen bei Femininen. Philologus, 81(NF 35), 101–106. Rydå, S. (1988). Present and Aorist Participles in Contemporary Greek Newspapers (Dissertationes Neohellenicae Universitatis Stockholmiensis 2). Stockholm.

JB[v.20020404] Prn:24/11/2004; 14:41

F: LA7608.tex / p.41 (2796-2911)

From participles to gerunds 

ted

pro

ofs

-

Joh

nB

enj

am

ins

Pub

lish

ing

Com

pan

y

Schwyzer, E. (1942). Zum sog. genitivus absolutus statt participium coniunctum im Griechischen. Emerita, 10, 98–104. Schwyzer, E. (1950). Griechische Grammatik (Bd. II). München: Beck. Seiler, H.-J. (1958). Zur Systematik und Entwicklungsgeschichte der griechischen Nominaldeklination. Glotta, 37, 4–67. Setatos, M. (1985). Paratirisis sta rimatika epiθeta se -menos ce -tos tis Cinis Neas Ellinicis (Observations on the verbal adjectives in -menos and -tos in Standard Modern Greek). Studies in Greek Linguistics, 5, 73–87. Sitaridou, I. & K. Haidou (2002). I arçi tis nomimopiisis ton ipocimenon sta ellinika Γerunðia (The licensing of subjects in Greek gerunds). Studies in Greek linguistics, 22, 583–592. Sklavounou, E. (2000). Ta epiθeta tis neoellinicis se -menos (The Modern Greek adjectives in -menos). Studies in Greek Linguistics, 20, 499–510. Smyth, H. W. (1956). Greek Grammar. Cambridge, MA: Harvard University Press. Soliman, A. H. (1965). Xrisis tis γenicis en tis metaptolemaicis papiris (The use of genitive in the postptolemaic papyri). Ph.D. Dissertation, University of Athens. Spieker, E. H. (1885). On the so-called genitive Absolute and its use especially in the Attic orators. American Journal of Philology, 6, 310–343. Spyropoulos, V. & I. Philippaki-Warburton (2001). “Subject” and EPP in Greek. Journal of Greek Linguistics, 2, 149–186. Stowell, T. (1982). The tense of infinitives. Linguistic Inquiry, 13(3), 561–570. Svennung, J. (1935). Untersuchungen zu Palladius und zur Lateinischen Fach- und Volkssprache. Uppsala: Almquist & Wiksells. Taube, M. (1981). Participe et gérondif en vieux russe. Folia Linguistica Historica, 2, 125–132. Torrego, E. (1998). Nominative subjects and pro-drop Infl. Syntax, 1–2, 206–219. Triantafyllidis, M. et al. (1941). Neolinici Gramatici tis Dimotikis (Modern Greek Grammar of the Demotic). Athens: OESV. Tsoulas, G. (1996). Notes on temporal interpretation and control in Modern Greek gerunds. York Papers in Linguistics, 17, 441–470. Tsimpli, I.-M. (2000). Gerunds in Greek. Journal of Greek Linguistics, 1, 131–169. Tsokoglou, A. & S. Kleidi (2002). I ðomes ton metoxon -ontas ce -menos (The structure of participles in -ontas and -menos). In Ch. Clairis (Ed.), Recherches en linguistique grecque. Actes du 5e colloque international de linguistique grecque. Sorbonne, 13–15 Septembre 2001 (Vol. II) (pp. 279–282). Paris: L’Harmattan. Tzartzanos, A. (1989). Neoelinici Sintaksis tis Cinis ðimotikis (Modern Greek Syntax of Standard Demotic). Thessaloniki: Kyriakidis. Vaillant, A. (1964). Manuel du vieux slave. Paris: Institut d’ Etudes Slaves. Weierholt, K. (1963). Studien im Sprachgebrauch des Malalas. Symb.Osl. Suppl.18, Oslo. Whaley, L. (1990). The effect of non-surface grammatical relations on the genitive absolute in Koine Greek. In K. Dziwirek et al. (Eds.), Grammatical Relations: A cross-theoretical perspective (pp. 457–472). Stanford: CSLI. Wolf, K. (1911–1912). Studien zur Sprache des Malalas (Vol. I & II). München: Straub.

rec

Primary sources

Un

cor

Acta Thomae: Bonnet, M. (Ed.). (1883). Acta Thomae. Leipzig.

JB[v.20020404] Prn:24/11/2004; 14:41

F: LA7608.tex / p.42 (2911-3042)

 Io Manolessou

Un

cor

rec

ted

pro

ofs

-

Joh

nB

enj

am

ins

Pub

lish

ing

Com

pan

y

Apocalypses Apocryphae: Tischendorf, C. (Ed.). (1866). Apocalypses Apocryphae. Leipzig: Mendelssohn. Assizai: Sathas, K. N. (Ed.). (1877). Asize tu vasiliou ton Ierosolimon ce tis Cipru, Cipriaci nomi, vizantina simvolea, kritice diathice. Meseonici Vivliothici 6. Paris: Maisonneuve et Cie . BGU: W. Schubart et al. (Eds.). (1895–). Ägyptische Urkunden aus den Königlichen (Staatlichen) Museen zu Berlin. Berlin. BGV: Legrand, E. (1880–1913). Bibliothèque Grecque Vulgaire. Paris: Maisonneuve. Chr. M.: Schmitt, J. (Ed.). (1904). The Chronicle of Morea, London: Methuen. Chr. P. : Chronicon Paschale in J. P. Migne (Ed.), Patrologiae Graecae cursus completus, 92, 70– 1023. Digenis E: Jeffreys, E. (Ed.). (1998). Digenis Akritis. The Grottaferatta and Escorial versions. Cambridge Medieval Classics 7. Cambridge: Cambridge University Press. Digenis G: Jeffreys, E. (Ed.). (1998). Digenis Akritis. The Grottaferatta and Escorial versions. Cambridge Medieval Classics 7. Cambridge: Cambridge University Press. Ev. Matt: Evangelium secundum Matthaeum. In K. Aland et al. (Eds.). (19912 ): Novum Testamentum Graece et Latine. Stuttgart: Deutsche Bibelgesellschaft. Evang. Thomae.: Tischendorf, C. (1876). Evangelia Apocrypha, Leipzig, 140–163. Guillou, Messina: Guillou, A. (Ed.). (1963). Les actes grecs de S. Maria di Messina. Enquête sur les populations grecques de l’Italie du Sud (XI–XIVe sc.). Palermo: Istituto Siciliano di studi bizantini e neoellenici. Hp.Mul.: Hippocrates, Gynaikeia. Hp.Steril.: Hippocrates, Peri aphoro:n. Kallimachos: Pichard, M. (Ed.). (1956). Le Roman de Callimaque et de Chrysorrhoé. Paris. Leontios: Gelzer, H. (Ed.). (1893). Leontios von Neapolis, Leben des heiligen Iohannes des Barmherzigen, Erzbischofs von Alexandrien. Freiburg i.B. / Leipzig: Akademische Verlagsbuchhandlung von J. C. B. Mohr. Livistros E: Lambert, J. A. (Ed.). (1935). Le roman de Libistros et Rhodamné publié d’après les manuscrits de Leyde et de Madrid avec une introduction, des observations grammaticales et un glossaire. Amsterdam. Machairas: Dawkins, R. M. (Ed.). (1932). Leontios Makhairas. Recital concerning the Sweet Land of Cyprus entitled “Chronicle”. Vol. I. Oxford: Clarendon Press. Malalas: I. v. Thurn. (Ed.). (2000): Ioannis Malalae Chronographia. Berlin: De Gruyter. P. Fay.: Grenfell, B. P. & A. S. Hunt (Eds.). (1900). Fayum Towers and their Papyri. London. P. Grenf.ii: Grenfell, B. P. & A. S. Hunt (Eds.). (1897). New Classical Fragments and oother Greek and Latin Papyri. Oxford. P. Lips.: Mitteis, L. (Ed.). (1906). Griechische Urkunden der Papyrussammlung zu Leipzig. Leipzig. P.Cair.Zen.: Edgar, C. C. (Ed.). (1925–1951). Zenon papyri. Cairo: Imprimerie de l’Institut français d’archéologie. P.Giss: Eger, O., Kornemann, E., & Meyer, P. (Eds.). (1910–1912). Griechische papyri im Museum des oberhessischen Geschichtsvereins zu Giessen. Leipzig & Berlin. P.Hib.: Grenfell, B. P. et al. (1906–1955): The Hibeh papyri. London: Egypt Exploration Fund. P.Tebt.: Grenfell, B. P. et al. (1902–1976): The Tebtunis papyri. London/NY: H. Frowde, Oxford University Press. Pap.Cair.Masp.: Maspero, J. (Ed.). (1911–1916). Catalogue général des antiquités égyptiennes du Musée du Caire: Papyrus grecs d’époque byzantine. Cairo. PGM: Papyri Magicae Graecae, ed. by K. Preisendanz et al. Leipzig & Berlin, 1928, 1931, 1941. Pind.Olymp.: Pindarus, Olympian Odes.

JB[v.20020404] Prn:24/11/2004; 14:41

F: LA7608.tex / p.43 (3042-3080)

From participles to gerunds 

Un

cor

rec

ted

pro

ofs

-

Joh

nB

enj

am

ins

Pub

lish

ing

Com

pan

y

POxy: Grenfell, B. P. et al. (Eds.). (1898–). The Oxyrhynchus Papyri. London: Egyptian Exploration Fund. Rev.Eg.: Revue d’Égyptologie. Published by La Société Française de l’Égypte, 1933– Romanus Melodus: Maas, P. & Trypanis, C. (Eds.). (1963). Sancti Romani Melodi Cantica. Oxford: Clarendon Press. Tacitus: P. Cornelius Tacitus, Annales. Theocr.Idyll.: Theocritus, Eidyllia. UPZ: Wilcken, U. (1922–1937): Urkunde der Ptolemäerzeit. Berlin: De Gruyter. Velthadros: “Velthandros ce Xrisandza”, in E. Kriaras (Ed.). (1955), Vizantina ipotika mithistorimata. Athens: Aetos. Vita Epiph.: Vita Epiphanii, in J. P. Migne (Ed.). (1858), Patrologiae Graecae cursus completus, Vol. 41. War of Troy: Papathomopoulos, M. & Jeffreys, E. M. (Eds.). (1996). O polemos tis Troados – The War of Troy. Athens: MIET. X.Anab.: Xenophon, Anabasis.