Palmer L.R. The Greek Language

Подождите немного. Документ загружается.

< previous page page_289 next page >

Page 289

form *s-bhi/*s-bhei (cf. Lat. sibi). A corresponding dual accusative

genitive-dative appears in Homer.

The phonetically-weak reflexive forms were strengthened by the addition of (p. 286), which also took on the

anaphoric functions of *swe:

, etc., and were contracted in Attic to , etc. In the plural early

Attic has the expected combinations

, etc. The pronominal stems were used to form possessive

adjectives by the addition of the accented thematic vowels:

< *tw-os (cf. Skt. tva-), but also (Hom.

etc.)<*tew-os (cf. Lat. tu-us). In the plural Aeol. , Doric show adjectival thematization, but elsewhere the

'contrastive' suffix -*tero- (p. 282) was used:

. In the third person < *sw-os (cf. Skt. sva-)

and < *sew-os (cf. Lat. suus). The corresponding plural forms are and , but they were

superseded in Attic by the genitive of

D. Numerals

Among the IE numerals only I-4 and 100 were declined.

1(a). *sem-s/*sm-jH2/*sem (cf. Lat., sem-el, etc.) >

, . The -v of the neuter ( < *-m) was generalized in

the declension (

, etc.), but Myc. still presents eme = hemei (dative singular).

1 (b). IE possessed another root *oi- with the meaning 'alone' and this appears in 'the one on the dice' (an n-

suffix, seen also in O.Lat. oino = unum, Ir. óin, Goth. ains) and

' alone' (cf. Avest. aeva).

2. *duwo (*dwo), masculine, *duwoi (*dwoi), feminine, neuter (cf. Skt. duva, dve). In Greek the distinction of

gender is lost: Myc. dwo, d(u)wouphi (p. 48); long o still in

(Hom. etc.), but later (Lat. duo, Goth. twai,

etc.). The Attic genitive-dative

is dissimilated from . The apparently plural inflexion of Elean

may in fact be an old dual parallel with Skt. genitive-locative dváyas (cf. O.Sl. dvoju), but other plural forms are

Ionic

, late Attic , etc.

3. *trejes, masculine/*t(r)isres, feminine/*tri (< *triH2) neuter; cf. Skt. tráyas, tisrás, Greek has lost the

feminine form:

(< < *trejes), (cf. Lat. tres, tria). The oblique cases had the zero grade *tri-:

accusative *trins >

(Cyrene, Cret. ), replaced in Attic by nominative ,

< previous page page_289 next page >

< previous page page_290 next page >

Page 290

genitive , dative , and this also appears in compounds, e.g. .

4. * qwwetwores * masculine/qwetesres* feminine/qwetwor* neuter (cf. Skt. catásras

>, with oblique cases *qwetur-ns*, * qwetur-om*, * qwetwr-su*.

Greek eliminated the feminine, and no Greek dialect preserves the original stem alternations. Doric < *qwet

(w)* -ores,, with w lost by dissimilation. Lesbian has the zero grade<*-twr- of the oblique cases (for qwe-*

in Aeolic, see p. 60; the vowel of Homeric is unexplained); Homeric (Attic )<*-twr-,

the stem form of the locative. Ionic

is not ancient but due to assimilation. In composition

< *qwetr* which appears in Myc. qet(o)ro-.

5. *penqwe* (Skt. páñca, Lat. quinque) > (Aeol. ).

6. *s(w)eks (Skt. sas-, Lat. sex) > .

7. *septm* (Skt. saptá, Lat. septem) > .

8. *okto (Skt. , Lat. octo) > .

9. *newn* (Skt. náva, Lat. novem) > , with the same unexplained prothesis as Armenian inn.

10. *dekm(t)* (Skt. dása, Lat. decem) > .

For 11-19 Greek uses a combination of 1-9 with : note < (but also ). In the numerals 20-90

the basic stem is a noun *dkomt-, an Ablaut variation of *dekmt*.

20-90. *wi-dkmti* (wi 'two' and the dual ending of the noun) >

(Doric, with shortened i). Attic etc. is

due to (a) a vowel prothesis (Hom. < ), (b) the assibilation of to , and (c) the introduction of -

o- from the numerals 30-90, which have the neuter plural

< *(d)-komt-H2. In the consonant cluster of

*dkomt-H2 the d- disappeared with consequent lengthening of the final vowel of penqwe*; hence

which provided the model for . (these two with the ordinal numeral as the first

element) and

with a reduplication (the forms and are also attested). For 40 an IE

neuter form has been constructed, being lengthened when d was lost. This is the source of Latin

quadraginta. The proto-Greek form would have been *qwetwra-konta*, which caused the change in the word for

< previous page page_290 next page >

< previous page page_291 next page >

Page 291

from *tri-konta (cf. Latin triginta) to triakonta before itself being altered to *qwetwarakonta * >

100. (d)kmtom* (Skt. satám, Lat. centum) >

, with a prothesis which is traced to *sem/*sm-* 'one' (but

elsewhere this appears as

, p. 260). The hundreds are based on an adjectival derivative / of *kmto*,

which Attic-Ionic changed to

under the influence of . Analogy of is seen in while

the

of has been taken over by , and .

1000. There is no general IE word, but Skt. sa-hàra-m is a compound of *sm* 'one' and a stem equatable with

(Lesb.), (Ion.)<*gheslo-. Attic is due to assimilation.

The ordinals were originally adjectives formed by adding the thematic vowel to the cardinal: e.g. Skt. sapiam-á-,

dasam-á-; Lat. septimus, nonus < *nov enos, decimus < *dekmos* . In and and have

been voiced. The other ordinals are formed with the suffix (p. 282): (hyper-characterized in Hom.

), / , (Dor. ) has the suffix attached to a stem *pr

(*prH2), which may link up with

'limit' (*perH2-/preH2-/prH:-): has the 'contrastive' - (p.

282) attached to a root meaning 'inferiority', ' deficiency' 'need', / 'lack', 'be ireflor'), which

links up with Skt.

'lack'<IE *deus-. The ordinals 20-90 undergo the regular phonetic transformations:

> > (p. 241): , etc. (Boeotian < *wi-knt-to* The suffix was carried

over to the hundreds and the thousand ( ).

1 Another possibility is that / replaced an original *pro-mo-s after this was specialized in the

seine 'headman', 'chief', 'prince'. The dialect forms < *pro-ato-s (for the suffix see p. 282). If this form was

due to the dissimilation of *pro-tato-s, the counterpart of pro-tero-s, then the 'common Greek' *pro-ato-s

must considerably post-date 'proto-Greek' *pro-mo-s.

< previous page page_291 next page >

< previous page page_292 next page >

Page 292

E. The Verb

I. Introductory

Whereas in noun inflexion the major concept is 'case' with 'number' as minor, verbal inflexion is dominated by

'person', which is modified not only by number, but also by 'voice' and 'tense'. In considering verbal stem formation

the predominant category is 'aspect', to which 'mood' must be added. These categories and concepts will now be

defined.

The 'persons' are those discussed under pronouns; in fact there is plausibility in the speculation that the verbal

inflexions originated in the attachment of pronominal elements to the verbal stem, although there is only slight

confirmatory evidence. To abbreviate reference, the nine theoretically possible endings (three persons x three

numbers) are numbered: singular 1, 2, 3; plural 4, 5, 6; dual 7, 8, 9. Tense in IE comprises a binary opposition

present:past, for there was no grammatical device to express 'future'. It is plausibly argued that at an earlier stage of

IE there was a single 'primitive' set of terminations for both present and preterite; e.g. *-m for first singular active.

At a later stage a deictic particle -i 'here and now' was added to emphasize the 'present' reference, and the opposed

unmarked *-m was then confined to the 'past'; *-mi/*-m then formed an opposition known as 'primary'/' secondary'

(hereafter P/S). Not all persons developed this opposition. The reference to the past, signalled by the secondary

endings, was emphasized in Greek (as in a few other IE languages, see p. 7) by the use of a prefix, the augment,

which plausibly originated in an accented adverb of time to which the verb was attached as an enclitic: e.g.

< *é-bhero-m. The augment appears only in the indicative.

For 'voice', too, there was in IE a binary opposition active/ middle. The active verb was used to present an activity

proceeding from a subject outwards: when the event took place within the subject or was reflected on the subject,

then the middle voice was used. The stock example is

'he washes' (an external subject), v. 'he washes

himself'. The Sanskrit grammarians expressed the opposition between active and middle as parasmai-padam 'word

for another' v. atmane-padam' word for oneself '. Inherent in the middle is the notion of

< previous page page_292 next page >

< previous page page_293 next page >

Page 293

the 'passive', formal grammatical distinction of which developed gradually within Greek.

Mood conveys the attitude of the speaker towards the event (in the broadest sense) referred to by the verb. 1. The

indicative is the unmarked member, and its function may be defined as neutral attitude: it states the verbal event as

a fact. 2. The subjunctive projects the event into the future and may be semantically divided into (a) expectation

('prospective' or 'eventual') and (b) will. (the 'voluntative'). 3. The optative similarly divides into (a) wish (the

'optative' proper) and (b) contingency (the 'potential'). 4. The imperative is used for giving commands.

Aspect

In IE the verb had three main stem forms which are called 'tense' stems although their functions are not concerned

with distinctions of 'time', but with 'aspect'. 1. The verbal event may be presented as a global unit, an item of

history, without further embellishment or 'definition'. The corresponding stem is known as the 'aorist' (

'undefined'). 2. The event may be presented in all its immediacy as a process, a kind of 'eye-witness' aspect; the

corresponding stem is known most inappropriately as the 'present' stem, although a preterite is formed from it (see

below on the imperfect). On the manifold ways of forming 'present' stems, see pp. 262 ff. 3. The speaker may refer

to a state of affairs which has come about as a result of an event. The 'perfect' stem is then used.

These three oppositions are known as the 'aspects' of the verb. This term is a mistranslation of the word used by

Slavonic philologists, who for each verb distinguish two vidy, stems denoting 1, action in progress and 2,

completed action. These were regarded as the two 'species' of the verb. It has been suggested that the three 'tense'

stems of IE were not concerned with 'aspect', but the analogies of Greek and Sanskrit usage leave no alternative to

the above exposition. The temporal references of the verbal forms are made by augment and secondary inflexion

and are in any case confined to the indicative:

'I am carrying'/ < *e-bher-o-m 'I was carrying',

whereas the modal forms (including the infinitive) based on the same stem have no temporal reference.

< previous page page_293 next page >

< previous page page_294 next page >

Page 294

Thematic and Athematic

The terminations in the example just quoted require a fundamental morphological distinction which divides the

Greek verbs into two opposed systems: the stem immediately preceding the inflexion shows a vowel -o- (added to

the root *bher-), which alternates with -e- in the conjugation: e.g.

. Other verbs show no such

vowel: / . The two conjugational types differ not only in the personal endings

(in 1, 2, 3, and 6), cf. , > , but also in the formation of the moods and the

infinitive (see below).

Augment

The augment appears as an optional feature (on Mycenaean see p. 49) in early Greek (e.g. Homer) as it is in the

earliest Sanskrit.

Before consonants the normal form of the augment is *e-. In verbs beginning with a vowel, this was lengthened

already in late IE by contraction with the augment'

, Skt. , (Dor.) 'he was', Skt. 'was' (*e-es-t >

*est). This was the so-called 'temporal' augment contrasting with the 'syllabic' augment. From a synchronic point

of view, in Greek vowel lengthening constituted augmentization in the case of verbs with initial vowel, and this

was applied to words which originally began with an s:

/ (*si-sd-o/*e-si-sd-o-m), /

(*saljo-/ * e-salj-o-man) .

In some Greek verbs beginning with w- the augment appears in a lengthened form: . The root wra-, which

appears in the Homeric aorist participle , has the indicative . In Vedic the long augment may be

used not only before v-, but also before j, n and r, so that the phenomenon may be IE (in the limited area discussed

p. 7) In Attic the long augment is disguised by quantitative metathesis:

< < (cf.

). Sometimes is used ( , see below) and the alternation h/e may be due to lost laryngeal

consonants.

< previous page page_294 next page >

< previous page page_295 next page >

Page 295

II The Personal Endings

The personal endings may now be tabulated, those of the perfect being reserved for treatment below. First the

active:

Primary

Secondary

1. * -mi/-o >

, -w *-m >-n/-a

2. *-si >

*-s > -s

3. *-ti > *-t > -Ø

4. *-mes >

*-me- > ( / )

5. *-te(s) >

*-te >

6. *nti > / /

*-nt > -n

7. *-wes/-wos *-we/-we

8.? ?

9.? ?

1. The thematic inflexion -w (

, Lat. fero, etc.) may possibly owe its length to a lost laryngeal (see below); *-m

regularly > -v, while Cm > Ca: *es-m * 'I was' >

, *e-deiks-m* > etc. On the optative form < *-

oi-m*, see p. 310.

2 and 3. The Greek representation of the IE endings has been much discussed, and no general agreement has been

reached. A synchronic structural analysis will bring out the facts which have to be explained, and this shows a

clear difference between the athematic and the thematic conjugations. It seems a priori unlikely that phonetic and

analogical processes operating on a single set of IE endings can have produced this distinction between the two

classes.

The athematic verbs show regularly 2, -V:

, etc. If this were a recharacterization of < *-si one

would expect

, etc. Another difficulty is offered by the forms for 'thou art'. Given the general system

represented by , etc., we should expect Greek , and this form may account for Homeric ,

which would be purely graphic for the long syllable (cf. Plautine ess in Latin). Within a syn-chronic system *esmi,

*ess, *esti, *ess would be levelled to essi, and this form is actually found in Aeolic and Doric. The Attic

falls out

of the system. Yet it is taken as the representative Greek form, equated with Skt. asi and an IE *esi is constructed.

This strange form, however, requires the postulation of an ad

< previous page page_295 next page >

< previous page page_296 next page >

Page 296

hoc IE 'sound law', simplifying the -ss- of the *es-si, required by the system. Thus *essi cannot be assigned to the

late IE constructed on the basis of Greek and Indo-Iranian; it is a postulated early IE form arrived at by internal

reconstruction and so should be written **essi. Given the succession early IE **essi > late IE *esi, it would be a

contradiction to say that Greek preserves the ancient **essi. On the contrary, within the synchronic system it

appears as a levelling of *ess, rather like 2,

(Epidaurus) for . If we start with *-mi, *-si, *-ti, what

requires explanation is the disturbance of the system by the introduction of the anomalous 2, -s. It is no explanation

to say that the athematic class 'borrowed' the secondary ending

. Starting with the Greek facts, we should

say that the anomaly is probably what has been inherited. Thus Attic

is not to be equated with Skt. asi. The

equation is made more difficult by the accentuation of the Greek word, for the other persons are enclitic (as is

and the difficulty is merely underlined by the necessity to postulate additionally an analogical influence of

'you

will go'.

Nor is there any direct evidence within Greek for 2, -si in the thematic conjugation. Here 2,

contrasts with 3,

. Purely synchronic analysis suggests that in both persons an -i-has been added to the stem . In 3, there is

zero inflexion (the bare stem suffices) with addition of the primary (deictic) -i-. In 2, the normal inflexion -s has

been added. That the second person was formed from 3 has the greater statistical probability.

Methodologically it is preferable to treat the problem structurally; we have two contrasting sub-systems in the

singular active of the athematic and thematic verbs. The desirability of making the structural contrast the basis of

comparative treatment is given further point by the contrast of these two verbal classes in the formation of

subjunctive, optative, imperative and infinitive. We start with an agreed point: in the contrast -mi/-o, the

lengthened thematic vowel of the latter suggests, as we saw, an original *-oH. This guides the search of the

comparatist. He will look for evidence of structurally opposed -mi and -H verbal systems. This is a feature of

Hittite, and it is in the

conjugation that a third singular -i is found. It has been plausibly suggested that

paradigms like that of sak- 'know' 1,

< previous page page_296 next page >

< previous page page_297 next page >

Page 297

(sdkhi), 3, sakki may provide a parallel for the Greek thematic endings -w/-i.

Recent contributions have veered round to the derivation of all the Greek 2 and 3 endings from *-si and *-ti. They

all require, however, the postulation of phonological and analogical processes, the complexity of which reduces

plausibility. The latter are essentially levelling processes resulting in uniformity and regularity; yet in the end we

are left with the anomalous

, etc. and the necessity to suspect 'borrowing' of the secondary ending. Moreover,

the Mycenaean evidence has been overlooked. If *bhere-si > *phere-hi > phere-hi-s, the last form would have

remained trisyllabic until the loss of -h-, and the same would have been true of 3, *phere-hi modelled on the 2

form. According to the Linear B spelling rules a form like ekhehi would be written e-ke-i, just as e-ke-e stands for

the infinitive ekhehen. What we have, however, is e-ke, that is a disyllabic form with a final diphthong -ei. This

evidence is thus decisive against the derivation of

< *bhere-si-s.

In 3, the primary *-ti > *-si in Mycenaean, Arcado-Cypriot, Attic-Ionic and Lesbian. There are a few forms with

zero ending which have been extracted from 2:

, on the analogy of / , etc. The secondary

-t being lost in Greek, the sigmatic aorist repaired the loss by taking over -e from the perfect.

4. *-mes survives in West Greek, but the other dialects have an unexplained . A similar form occurs in Hittite -

weni/-wen, and this resembles the Sanskrit dual ending -vas/-va. On the possible P/S opposition *-mes/-me, see

below.

5. All Greek dialects present

and, as with 4, there is no distinction between primary and secondary.

6. survives in West Greek; in the other dialects > with compensatory lengthening for loss of -v. The

athematic ending was

, e.g. *s-enti (as in Osco-Umbrian sent), whence Doric for . Mycenaean offers

kitijesi = ktiensi, with the zero grade of the root *ktei-. With roots ending in a long vowel (e.g.

'blow') there

would have been 1, contraction with the vowel of

, 2, shortening of the long vowel (Ost-hoff's Law) >

> > . Reduplicated verbs show the inflexion *-nti: > > (also

< previous page page_297 next page >

< previous page page_298 next page >

Page 298

, and these forms were influenced by the contracted verbs and in Ionic received pro-

perispomenon accentuation

After consonant *-nti * > , which appears in West Greek inscriptions , while Ionian has a few

corresponding forms in

, e.g. (Homer). The ending was mostly recharacterized to > >

, the last being the regular Attic-Ionic ending in the perfect: . In Attic-Ionic this inflexion has spread

elsewhere: (Homer), 'they go' (Homer and Attic), the reduplicated presents and the

verbs ( , etc.).

The secondary ending -(e/o)nt lost the final plosive: hence , etc. In the athematic type *-ent appears in

'they were' and in the optative (p. 309) , etc.; *-ont is rare (Horn. 'they went' and in the verbs); *-nt in

, , etc. (on the Attic-Ionic substitution of from the s- aorists, see p. 62).

7, 8, 9. Traces of the dual survive: 8. (both primary and secondary) is in origin secondary ( = Skt.

ábharatam), and 9, with primary

and secondary / , cf. Skt. -tam ( = Skt. ábharatam).

Middle

Primary Secondary

*-(m)ai >

*-(m)a(?) >

*-soi > / *-so > /

*-toi > *-to >

* >

*-dhwe >

*-ntoi > *-nto >

The majority of the Greek dialects have 1, , 2, , 3, 6, as the primary endings. However, the

secondary endings are 2,

, 3, and 6, , and it is plausible to assume that the primary endings, as in the

active, were made by the addition of -i. In fact Arcado-Cypriot (and Mycenaean) show in 3 and 6

and . In

1, if it is plausible to suppose that

is a recharacterization of , then the opposition / could be

accounted for by assuming *-(m)aHi/ *-(m)aH, (to posit *-(m)oH2i/*-(m)oH2 would raise phonological problems

concerning the laryngeal). At all events it is arguable

< previous page page_298 next page >