Palmer L.R. The Greek Language

Подождите немного. Документ загружается.

< previous page page_209 next page >

Page 209

inscriptions v is found . According to the Roman grammarian Varro (on the authority of Ion) there

was a twenty-fifth letter called agrna, which was used for the velar nasal. The name implies a pronunciation

[ahma], and it follows that words like

must have exhibited the cluster [hm], and this would explain a

spelling like

. It is conceivable that was pronounced [hn], but the evidence for this is slight.

The articulation of l is roughly described by Dionysius of Halicarnassus, but he merely indicates the tongue posture

('rising to the palate') and the concomitant sounding of the windpipe, but omits the characteristic lateral release. For

all that, there is no doubt that this was a lateral consonant with dental rather than alveolar closure. A velar

colouring before a consonant is suggested by the occasional substitution of

in Cretan ( for ).

The descriptions of the ancients (e.g. Plato and Dionysius) indicate clearly that (r) was an apical trill. Plato refers

to the ' vibration of the tongue', while Dionysius of Halicarnassus notes that the tongue is raised against the palate

close to the teeth and 'beats back'

1 the breath. According to the grammarians, initial r was

aspirated, as was the second r in the medial cluster rr: in all probability this 'aspirated r' was a voiceless

.2 This

is supported by early spellings such as (Corcyra) and (Boeotian), and Latin transcriptions

such as rhetor and Tyrrheni. Where r followed an aspirated plosive the whole cluster was aspirated, as is evidenced

by Latin transcriptions such as Crhestos, Prhygia and Trhacem. Such an aspirated cluster appears in

+ ); here the r received its aspiration from the initial vowel of the second element and

communicated it to the preceding t.

The sound represented by

(digamma) was still preserved intact in Mycenaean (p. 44). Though lost at an early

date in Attic-Ionic, it was maintained in certain positions in other Greek dialects (for details see Chapter VIII). Its

articulatory closeness to the vowel

is shown not only by the origin of

1 There is a variant reading 'fanning back'.

2 The phonetic symbol subscript as in , etc. is used differently in Indo-European to designate the syllabic

sounds

, see pp. 215f.

< previous page page_209 next page >

< previous page page_210 next page >

Page 210

the two letters as variants of waw, but also by the substitution of

for to indicate the second element of

diphthongs, e.g. , etc. It is also used to represent the glide after . A voiceless counterpart

(etymologically traceable to original sw, see p. 238) is indicated by spellings such as (Boeotian) and

(Pamphylian).

S (s) stands for the sole fricative of Attic Greek. It corresponds to s in many of the related IE. languages and it is

still a voiceless sibilant in MnG. This value also emerges from its description by Dionysius of Halicarnassus, and it

is supported by transcriptions of Greek words on Indian coins and in Coptic, Armenian and Gothic. A voiced

allophone [z] appeared before voiced consonants, and this was rendered by 3 when the cluster originally denoted by

this letter had developed to [z] (see below). In certain dialects (Elean, late Laconian) [z] is rhotacized.

The rough breathing [h] is strictly speaking also a fricative consonant (a glottal fricative). This was represented by

, derived from the North Semitic hêt, which was a more vigorous and constrictive (pharyngeal) fricative than the

aspirate. This was because the letter he' had been used for the vowel e (see above). In the dialects of Asia Minor

the aspirate had been lost (for this 'psilosis' see p. 179) before the time of our earliest records, so that when the East

Ionic alphabet was adopted by the rest of Greece, the aspirate had no representative and it was ignored in writing.

In some scripts (e.g. at Tarentum and in the colony Heraclea)

the left half of , was used to denote the rough

breathing. Later (third to second century B.C.) this was used as a suprascript above the initial vowel by

Aristophanes of Byzantium with the counterpart

to mark the absence of the aspirate.

The inventors of the Greek alphabet appear to have regarded as units certain complex articulations, which we

should analyse as consonant clusters. , for instance, must have had the value [zd] to judge by <

< Further 'branch' is etymologically derivable from *osdos (cf. German Ast), and

is a reduplicated present tense (p. 262) < *si-sd-o, the root being *sed- 'sit'. That the cluster began with Is] is

shown further by the loss of a preceding n in compounds like

< , and the present stem ,

with aorist infinitive , passive .

< previous page page_210 next page >

< previous page page_211 next page >

Page 211

The present stem must be reconstructed as *plangjo, with a further development to

However, this and

many other examples where

is traceable to a palatalized *d or *g suggest that there was an intermediate stage in

the shape of an affricated prepalatal or alveolar plosive [dz] which underwent metathesis to [zd]. Allen quotes as a

parallel the Slavonic development of IE *medhja-' middle', OCS mezda, but Russian meza' boundary'. He argues

that when the Semitic letter zayin was borrowed, the affricate pronunciation still existed, for it would have been

natural to represent a cluster zd as sd. That the plosives were still palatalized and not affricated in Mycenaean has

been argued above (pp. 31f.). The assimilation of [zd] to [zz] took place in the early Koine. The voiceless

counterpart of *gj is *kj and its Greek derivatives are represented by the digraphs tt in Attic and Boeotian and ss

elsewhere;

/ < . Here, too, phonetic considerations suggest an intermediate affricate

stage

or [ts]. In certain early Ionic inscriptions a special sign, the letter , is used, and this resembles the

Semitic tsade. The separate development in Attic and Ionic (p. 62) indicates that the affricate pronunciation (or the

palatalized plosive) was still maintained in the Attic-Ionic period. In the initial position a single letter 't- or s-is

used for the products of this combination: e.g.

/ , and it is likely that at least by classical times tt and

stood for true geminates [tt] and [ss], which had developed from the affricates.

That (x) and (y) had the values [ks] and [ps] is clear from their use in forms like

, and from grammarians' descriptions. The invention of letters

for the clusters suggests that in the eyes (or rather ears) of the devisors of the alphabet they had an intuitive unity

and a quasi-phonemic status. This was probably mainly due to the fact that these clusters could occur in the word-

initial position. Another factor was possibly the occurrence of

(originally I) with the suggested original affricate

value, and the new letters were invented on this analogy.

On the prosodies of Greek and their notation see pp. 242 ff.

< previous page page_211 next page >

< previous page page_212 next page >

Page 212

VIII

Phonology

1. Introductory

The sounds lying behind the letters of the Greek alphabet having been determined within fairly narrow limits, and

the same preliminary work being assumed for the other languages such as Latin, Sanskrit, Hittite, etc., the work of

comparison can begin. From the lexicons of these languages words of identical or similar meaning are picked out

and their 'vocables' or 'expressions' compared. A stock example is the word for 'father':

Gk. , Lat. pater, Skt. pitár-, Goth. fadar.

Each of these words consists of a string of phonemes in a given order and they are equated position by position:

If we find the same set of 'correspondences' in a number of words, the conclusion is drawn that such phonemic

parallelism of the expressions in words of identical or similar meaning cannot be accidental but is due to regular

but divergent development from an ancestral language. By various methods the most likely phonetic features of the

ancestral phonemes (to which the term 'diaphonemes' might be applied) are deduced. In the present instance the

word reconstructed for IE is symbolized as

. It should be stressed that the phonetic description of the

postulated diaphoneme is not an essential part of its definition. The notation IE *p is simply shorthand for the set

of correspondences p = p = p = f, etc. However, the determination of the phonetic features is essential for setting

up the phonemic structure of Indo-European. In each of the observed languages, too, the phonemes form structures,

and in the last resort our comparisons and historical deductions should deal with phonemes not as isolated units but

as components of structures.

In writing the historical and comparative grammar of the individual IE languages, the ancestral reconstructed forms

serve

< previous page page_212 next page >

< previous page page_213 next page >

Page 213

as triangulation points which facilitate the rather complex procedure of comparison and inter-linguistic cross-

reference. The comparative phonology of Greek states the representation of the ancestral diaphonemes (e.g. *p > a)

and illustrates this by adducing the correspondents in the related languages. The chapter falls into three main

sections. The first two deal with the vowels and consonants; the third discusses the syntagmatic relations and

especially the prosodies.

2. Vowels and Diphthongs

The vowel phonemes of late Indo-European are set forth in the following table:

The system comprises three degrees of opening (tongue height) along the two dimensions, front and back, plus the

'prosody' of length: all vowels may be short or long. The high vowels i and u have consonantal counterparts j and

w. In other words, these sounds may, or may not, function as syllabic peaks and so may be classed along with the

nasals m, n and the liquids l, r as 'sonants'. This grouping is in any case required for morphological or rather

morpho-phonological reasons.

The late Indo-European vowel system was preserved virtually intact in proto-Greek (for the dialect changes see pp.

59f.), as will emerge from the following representation:

*ag- 'drive'

, Lat. ago, Skt. ájati, O.Ice. aka.

*aks- 'axle',

, Lat. axis, Skt. áksas, NE ax-le.

*agros 'field'

, Lat. ager, Skt. ajrás, NE acre.

e *bher- 'carry',

, Lat. fero, Skt. bhárati, Goth. baíran.

*dekm * 'ten',

, Lat. decem, Skt. dasa, Goth. taíhun.

*s(w)eks 'six',

, Lat. sex, Skt. sás Goth. saíhs.

*esti 'he is',

, Lat. est, Skt. ásti.

o *domos 'house',

, Lat. domus, Skt. dámas.

*oktou 'eight',

, Lat. octo, Skt. ástau, Goth. ahtau.

*owis 'sheep',

, Lat. ovis, Skt. avis.

a is presented in all dialects except Attic-Ionic.

*mater- 'mother',

(Doric, etc.), Lat. mater, Skt. matá-.

< previous page page_213 next page >

< previous page page_214 next page >

Page 214

* bha- 'speak',

(Doric, etc.), Lat. fa-ri, fama.

*sta- 'stand', (Doric, etc.), Lat. sta-re, Skt. stha-.

In Ionic a > h , but in Attic this change was checked and reversed by an immediately

preceding ei, or r: e.g.

, but (Ion. ). On the mechanism and chronology of these

changes see pp. 62f.

*dhe- 'put',

, Lat. feci, Skt. dádhami.

*ple- 'fill',

, Lat. plenus, Skt. pratás.

*mens- 'moon', 'month',

, Lat. mensis, Skt. mas *, Goth. menops (< *men-ot-).

o *do- 'give',

, Lat. donum, Skt. dádati, danám.

*gno- 'know',

, Lat. (g)nosco, Skt. jñatá-, OE cnawan.

*oku 'quick',

, Lat. ocior, Skt. asú.

i *wid-'know',

(aor.), Lat. video, Skt. vidmá 'we know', Goth. witan.

*qwis* 'who',

, Lat. quis, Hitt. kwis.

u *jugom 'yoke',

, Lat. iugum, Skt. yugám, Goth. juk.

*rudhro- 'red',

, Lat. ruber, Skt. rudhirás, O.Sl. rudru.

*medhu- 'honey',

Skt. mádhu, O.Sl. medu, OE medu.

i *wis 'strength',

, Lat. vis.

*wisos 'poison',

, Lat. virus, Irish fi, Skt. visa- (with short i).

u *mus 'mouse',

, Lat. mus, Skt. mus*, OE mus.

*dhumos 'smoke',

, Lat. fumus, Skt. dhumá-.

*bhru- 'brow',

(on the prothetic vowel see pp. 221 ff.), Skt. bhru-, OE bru*.

The vowels, both short and long, occur in diphthongal combination with the sonants i and u (for the morpho-

phonological parallelism with m, n, r, l, see below on Ablaut).

*aidh- 'burn',

, Lat. aedes, Skt. edhas- 'firewood'.

*aiw-' vital force',

, Lat. aevum, Skt. ayu*, Goth. aiws.

*daiwer 'brother-in-law',

, Lat. levir (from devir), Skt. devár-, O.Sl. deveri.

ei *ei-ti 'he goes',

, Lat. it, Skt. éti, Lith. eit(i).

*deik- 'mark', 'show',

, Lat. dico, Goth. ga-teihan 'proclaim'.

*bheidh- 'trust', ' persuade',

, Lat. fido.

< previous page page_214 next page >

< previous page page_215 next page >

Page 215

*woida 'I know',

, Skt. veda *, O.Sl. vede, Goth. wait.

*oinos 'one',

'one on the dice', Lat. unus (O. Lat. oino), Goth. ains.

*woikos 'house, village',

, Lat. vicus, Skt. vésa-.

The diphthong ui, with two high components, is rare: an example is

< * su-ju-s.. Other instances are due to

inner Greek developments, e.g.

. < * Preconsonantally evolved to u, e.g. the optatives like (<

) and the diminutive for ). A similar treatment of antevocalically is exemplified in

(vi B.C.) for . On the Mycenaean forms i-ju, i-jo see p. 276.

*aug- 'increase',

, Lat. augeo, Skt. ójas, Goth. aukan, Lith. áugti.

*tauro- 'bull',

, Lat. taurus, Lith. tauras*, Welsh tarw.

*aus- 'draw (liquid, fire, etc.)',

, Lat. haurio, ON ausa.

*jeug- 'join',

, Lat. iugera (nom. plur.), Skt. yok-tár-.

*eus- 'burn',

, Lat. uro, Skt. ósati.

*leuk- 'shine',

, Lat. lux, Skt. rócate, Goth. liuhap, 'light'.

ouExamples are rare in Greek except in Ablaut forms alternating with

(see

below).

/ . Ablaut alternation occurs in the word for 'udder' (see p.

278) *e/oudhr-'

, Lat. uber 'fruitful', Skt. udhar*, gen. udhnas*, O E uder.

Long diphthongs *ai, ei, etc. also occurred in Indo-European. Greek shortened them before consonant (according

to Osthoff's Law a long vowel is shortened before sonant [i, u, r, l, m, n] + consonant. Thus and were

only preserved in final position. These forms will be discussed under Morphology.

3. Sonants

These are sounds (also called resonants, semivowels or semi-consonants) which, according to their phonetic

environment, can function as consonants, as second elements of diphthongs and as syllabic peaks. For Indo-

European, on morphophonemic grounds, we postulate (in addition to i and u)

and .1

1 On the sounds, see my Descriptive and Comparative Linguistics, 56-7. The small subscript circle

indicates the syllabic function of the sound. We may compare the syllabic character of l and n in little [litl]

and seven [sev.n].

< previous page page_215 next page >

< previous page page_216 next page >

Page 216

For

and the 'laryngeals', see below. The morphophonemic phenomenon in question is called Ablaut. By this is

meant the alternation of vowels observed in morphological systems such as noun and verb formation, noun

declension and verbal conjugation. English, sing/sang/sung/song and drive, drove, driven, drift are good examples

of this Indo-European phenomenon. The system is seen at its simplest in roots with a short vowel:

shows the alternation of e/o. 'fly' has the aorist and a derived noun 'feather'; , where the

vowel is lost altogether, exemplifies the 'zero grade'. We tabulate some examples of the three grades, e, o and zero:

*pet- 'fly'

*segh- hold'

bheidh- 'persuade', 'trust'

leiqw- * 'leave'

bheug- 'flee'

ghew- 'pour'

From these examples it emerges that in the zero grade the simple short vowel disappears but the second element of

the diphthong is left as a residue: in the above examples i and u. That the same rule holds good also for the

diphthongs em, en, el, and er is clear from the following table, which shows the reflections of

and in the

zero grade.1

'see clearly'

(aor. inf.)

< *drk-*

'sack', 'destroy'

< *prdh-*

'cut'

< *im-*

'kin' < *gn-*

'grief'

< *pndh-*

'make ready', 'equip'

< *stl-*

'be', 'become'

< *qwl-*

(*sem-s) 'one'

'once'

< *sm-*

It is important to note that these different Ablaut grades tend to appear in different parts of the morphological

system. Thus the zero grade is particularly associated with the thematic root aorist, the perfect middle, and the

verbal adjectives in -tó-(e.g. tn-to-s* < *ten-'stretch').

The Greek representation of the IE vocalic allophones

1 Confusion maybe caused by the different use of the subscript circle in IE reconstructions and phonetic

notations, where

, etc. stand for the voiceless sounds; e.g. Welsh llan 'church' is represented as .

< previous page page_216 next page >

< previous page page_217 next page >

Page 217

(written

may be further exemplified from Attic (for the dialect representations, see p. 68).

*dekm * 'ten': , Lat. decem, Skt. dása, Goth. taíhun.

*(d)kmtom* 'hundred':

, Lat. centum, Skt. satám, Goth. hunda (plur.).

*septm* 'seven':

Lat. septem, Skt. saptá, Goth. sibun.

*newn* 'nine': , Lat. novem, nonus (< *noven-os), Skt. náva, Goth. niun.

*tntos* 'stretched':

, Lat. tentus, Skt. tatás.

* (negating prefix): , Lat. igngotus, Goth. unkunps.

*mldu* 'soft': , Lat. mollis, Skt. mrdú-.

*plt(H)u* 'broad':

, Skt. prthú-.

*rtko-* 'bear': , Lat. ursus (from *orcsos), Skt rksas.

*krd*. 'heart':

, Lat. cord-, Goth. haírto.

*dhr's-* 'dare':

, Skt. dhrsú-, dhrs-n-oti, Goth. ga-daúrsan.

A further sonant is postulated for IE, chiefly on the grounds of morphological analysis. This is

, the so-called

'shwa', a term taken over from Hebrew grammarians, who used it to denote an indeterminate vowel, like the first

sound of English about

. The shwa is postulated as a diaphoneme to account for equations like

Lat. pater, Skt. pitár.

The key to its place as a morphophoneme is given by the fact that shwa also occurs in the zero slot of the ablaut

alternations:

: Lat. status, Skt. sthitás.

Noteworthy is that the full grade is sta-, so that the alternation is

. On the basis of alternations like Latin donum,

datus and fe-ci/fa-cio, which can be paralleled in the related languages, we get alternations and . At first sight

it is difficult to fit such ablaut alternations into the simple and clear system of e/o/zero outlined above. There,

given the zero grade CC, it was possible to reconstruct the full grades simply by inserting the vowel e or o: CeC,

CoC: e.g. pt-, pet-, pot-. If the same morphophonemic notation is adopted for these skew examples, the notations

would be:

and . These postulated ancestral

< previous page page_217 next page >

< previous page page_218 next page >

Page 218

forms yield Or.

/ , / , and / . The transformation of the postulated diphthongs into the observed

long vowels standing in the e-grade positions must be formulated by rules. Since it would be an offence against the

sound-law principle to postulate three different treatments of one original diphthong, we have to posit three

different shwas:

and . It was this structural morphophonemic analysis which prompted the

postulation of three different sonants which appear as syllabic nuclei in the zero grade. They make possible a neat

and tidy representation of IE Ablaut relations.

E grade

Zero grade

pet- pt-

leiqw * liqw*

bheudh- bhudh-

hem- nm-

gen- gn-

derk- drk-

qwel* qwl-*

These IE diaphonemes, quasi-algebraic notations representing morphophonemic relations, were gradually endowed

with real phonetic features. It is a reasonable guess that they represent laryngeal consonants, and they have been

given the notations H1 H2, and H3. In 1927 the Polish scholar Kurylowicz showed that with certain IE words

beginning with *a- and *o-their Hittite correspondents began with an h-phoneme.*

Lat. ante Gk. Hitt. hanti*

Lat. os 'bone' Gk. Hitt. hastai*

The evidence indicated that while the postulated H1 has left no trace (ezi < H1esti), H2 and H3 had real reflections

for the first time in an IE language.

The morphophonemic ramifications of the laryngeals will be discussed in detail under Morphology. For the present

it will suffice to state some basic rules.

In word-initial position (a) before consonants (the examples are most frequent before sonants) the laryngeals

develop to

< previous page page_218 next page >