- See “Mathematical Linguistics” by W. Plath in C. Mohrmann (ed.) et al., Trends in European and American Linguistics, 1930–1960 (Utrecht: Spectrum, 1961), pp. 21–52. For a short critical bibliography, see also K. L. Pike and E. V. Pike, Live Issues in Descriptive Linguistics2 (Santa Ana, Calif.: Summer Institute of Linguistics, 1960), pp. 1–9. Significant is the series of Papers on Formal Linguistics edited by the University of Pennsylvania in co-operation with the National Science Foundation Project in Linguistic Transformation.
- Z. S. Harris, “Discourse Analysis,” Language XXXVIII (1952), 1–30; C. F. Hockett, “Two Models of Grammatical Analysis,” Word X (1954), 210–234; N. Chomsky, “Three Models for the Description of Language,” I.R.E. Transactions on Informal Theories, IT-2 (1956), 113–124; and Syntactic Structures ('s-Gravenhage: Mouton, 1957).
- See “The Logical Basis of Linguistic Theory” in Preprints of Papers for the Ninth International Congress of Linguists (Cambridge, Mass.: 1962), pp. 509–574. Cf. Linguistic Essays on the Occasion of the Ninth International Congress of Linguists (= Word XVIII [1962]) and specifically M. Halle's “Phonology in Generative Grammar,” ibid., pp. 54–72.
- For this assumption and for all italicized terms contained in it, see Syntactic Structures, page 13. For the various syntagmatic levels above and below the Sentence as a sequence, cf. V. B. Pickett, The Grammatical Hierarchy of Isthmus Zapotee (Baltimore: Waverly, 1960), page 11.
- F. W. Householder, Jr., “On Linguistic Primes,” Word XV (1959), 232. Cf. P. Lackowski, “Words as Grammatical Primes,” Language XXXIX (1963), 211–215. Of historical significance is the chart representing the “structural formula of the monosyllabic word in English (Standard mid-western American)” in B. L. Whorf, “Linguistics as an Exact Science,” Technology Review XLIII (Dec. 1940), 61–63, 80–83, and reprinted in J. B. Carroll (ed.), Language, Thought, and Reality, Selected Writings of Benjamin Lee Whorf (Cambridge, Mass.: The Technology Press of M.I.T., 1956), pp. 220–232.
- R. B. Lees, “Review of Chomsky, Syntactic Structures,” Language XXXIII (1957), 375–408.
- H. Contreras and S. Saporta, “The Validation of a Phonological Grammar,” Lingua IX (1960), 1–15.
- S. Saporta and H. Contreras, A Phonological Grammar of Spanish (Seattle: The University of Washington Press, 1962). Cf. E. C. García's Review in Word XIX (1963), 258–265. J. M. Anderson's M.A. Thesis, “The Morphophonemics of Gender in Spanish Nouns,” was published in Lingua X (1961), 285–296. Already in 1958, E. Haugen had dealt with a phrase-structure phonological grammar in “The Phonemics of Modern Icelandic,” Language XXXIV (1958), 55–58, pushing in this way a lead toward “formal' descriptions of so-called “critical” languages.
- Therefore, along with them, we assume that the ideal grammar generates: (a) all the grammatical sequences and only these; and (b) each linguistically significant sequence only once. Saporta and Contreras, page 13.
- D. C. Swanson, Review of A. Mirambel, La langue grecque moderne: Description et analyse (Paris: Klincksieck, 1959) in Language XXXVIII (1962), page 191. Swanson basically accepts Mirambel's phonemic system (or rather what seems to correspond to a phonemic system, because Mirambel's “phonèmes” at times are indistinguishable from phones), but makes no mention of B. E. Newton's “The Rephonemicization of Modern Greek,” Lingua X (1961), 275–284, since the latter article must have been published immediately after the former.
- For the “Language Question,” see the several entries in D. C. Swanson's Modern Greek Studies in the West (New York: The New York Public Library, 1960), pp. 34–35.
- Mr. ’Iωάνννες Γκνιãς, born in Cassandra, Chalcidice, on September 15, 1935, was educated there until he moved to Athens, where he stayed from 1954 to 1956. His further college-level education and army service forced him to travel in Volos, Crete, Larissa, and Athens again, where he settled in 1961. In May of 1963, with a very rudimentary knowledge of English, Mr. Ganyas came to Toronto, Canada, where he settled with the intention of pursuing a course of studies in engineering. My first contact with him began in July and ever since he has been very helpful in our effort to solve (or to recognize viva voce) some of the “hot” issues involved in the phonological analysis of dimotiki. I am extremely indebted to Mr. Ganyas for his collaboration.
- Particularly useful also was Swanson's Vocabulary, since in it phonetic transcription is given whenever semantemes do not follow general rules. Our contacts with the informant force us to take into consideration some synchronic variants that denote a more demotic tendency in the results of our investigation. For example, during the search for minimal pairs, we have verified that, in the case of the informant, the nasal in the consonantal cluster [-mb-] is not present in words such as κάμπoς, καμπoṽρα, and κάμπoσoι where, contrary to Swanson's phonetic transcription, we always heard [kábos], [kabúra] and [kábosi]. Of course, one may assume several explanations for these phenomena, as analogy, influence of the written language on the spoken, loans, etc.; but on a strictly synchronic plane, one has to accept phonetic reality even in the case of borrowings such as καμπάνα that the informant pronounced [kabána].
- D. C. Swanson, in “English Loanwords in Modem Greek,” Word XIV (1958), page 30, posited twenty consonantal and five vocalic phonemes. Among the consonants he listed /j/ and/ç/. There were no affricate phonemes /ć/ and /ź/ = [ts] and [dz]. In 1959, he added /ŋ/ to the list included in the “Introduction” to his Vocabulary, page 60. In the same year, Mirambel was publishing La langue grecque moderne, in which he posited the phoneme /y/ as a semivowel, but no /ç/ or /ŋ /. In 1962, in the review to Mirambel's Langue grecque, Swanson declared (p. 192): “Curiously, Mirambel's list of ‘phonemes’ corresponds exactly to the list which should in fact be established for Modem Greek.' Mirambel had excluded the /ŋ/ and the /ç/, though it is not easy to be sure of the exact “phonemics” behind the whole list given by him. The latter phoneme was retracted by Swanson in footnote 8, page 193, of the same review (Language XXXVIII [1962]). Meanwhile, Newton's important article on re-monophonemicization of Modem Greek had appeared in December 1961. Mirambel's and Newton's phonemic descriptions are, among several, the ones that best describe dimotiki. They are, reorganized on the basis of correlation, as follows:
- Mirambel's quadrilateral system showing square correlations in the core system.
- Newton's triadic system showing no squaring in the core system.
- In essence, by postulating an hypothetical core consonantal system as the one following, one can only consider as affricates those “clusters” within the same order and never from two different orders:
- Phonetically possible affricates may be, for the voiceless series, [pf], [ts], and [kx], and never *[ps], * [px], * [tf], *[tx], * [kf], or * [ks]. Of course, independently from their phonetical possibilities, each affricate has to be tested functionally in order to see whether it is a phoneme or not. For the notion of “core system,” see A. Martinet, A Functional View of Language (Clarendon Press, 1962), p. 78, note 1, in which Martinet refers to a paper read by E. Dorfman at the Chicago Meeting of the Modern Languages Association on December 28, 1959. The same notion, however, had first been officially presented by Dorfman in the paper “The Function of the Square in Linguistic Evolution, “read at the Ninth Annual Northwestern Conference on Foreign Language Teaching in Eugene, Oregon, on April 11, 1958.—Dorfman's “core” should not be confused with C. F. Hockett's “core.” Cf. the latter's A Course in Modern Linguistics (New York: MacMillan, 1958), pp. 261–267.
- When the informant produces a group of minimal pairs (whether native or borrowed) such as /cámi/ τσάμι ‘pine’ and /źámi/, the researcher does not usually think of spelling or whether [ź] has the grapheme [dz] as in English. In the case of [zámi], if Greek spells it τζάμι ‘glass (in a window),’ and whether d or t is a component of the written cluster is irrelevant especially when seen historically in the framework of lexicographic developments. Cf. Mirambel, La langue grecque moderne, p. 341, and Newton, “The Rephonemicization,” p. 284, in which “there seems, therefore, no difficulty in regarding [ts] as a sequence of /ts/ and [dz] as a sequence of /tz/.”
- In this connection one might say that it is not difficult to view the graphemic cluster as an historical accident, and that if a voiceless affricate is represented by a t and an s, this representation may also be a graphic device invented (or transferred) to represent a newly created phone. In addition to the mere syntagmatic association of a /t/+an /s/, we know that /ć/ is also, if not basically, derived from ke.t or ke.t (much the same as from Classical Latin to Romance) as in ταíχλα ‘thrush’ κíχλα, or ρετσíνα ‘a type of a wine’ ρητíνα Cf. Catone, p. 34–35, and E. Boisacq, Dictionnaire étymologique de la langue grecque (Heidelberg: Winter, 1916), p. 462, and p. 840. For similar difficulties of phonological description, cf. H. and R. Kahane, “Problems in Modern Greek Lexicography” in F. W. Householder and S. Saporta (eds.), Problems in Lexicography (Bloomington: Indiana University Press, 1962), pp. 249–262. (Re. Catone, see note 19 below.)
- Though in a synchronic analysis historical considerations may appear to be antithetical to some structuralists, the researcher cannot dismiss his extension into the safe grounds of the recorded immediate past, as well as the predictable grounds of the immediate future. Functional and structural researchers, in particular, usually do not see any dichotomy between synchronic and diachronic analyses. In essence, they view a synchronic description as a “horizontal system forming a link in a vertical or diachronic axis.” Cf. Dorfman, “The Function of the Square” (note 15, above) and for the concept of the square, first suggested by Martinet, see his “Rôle de la corrélation dans la phonologie diachronique,” Travaux du Cercle Linguistique de Prague VIII (1939), pp. 273–288.
- The core consonantal structure of Classical Latin shows a case in point when it shifted from a triadic system to a quadrilateral system (as Early Romance, Modern Italian, Portuguese), to a dual system (as in Modern French), while it remained triadic as in Castilian Spanish. A similar phenomenon of consonantal core reshuffling from a triadic to a quadrilateral system took place in the volutionary process from koine to dimotiki. Besides the disappearance of length and the pressure of monophthongization from Classical to Modern Spoken Greek, perhaps the most significant aspect of phonological change lies in the formation of the new series of voiced stops which characterizes, at the consonantal level, the relevant exponent of dimotiki.
- The same can be said of the dental and velar orders in which minimal (non-borrowed) pairs are the result of syntagmatic (such as syncopation and sonorization) diachronic accidents. For t/d, a typical case is /tíno/versus /dino/ (τεíνω ‘I tend to’ and ντúνω ‘I put on, I wear’). For k/g: /krína/ versus /grína/ (κρíνα ‘lilies’ and γκρíνα ‘sugar and coffee container’). As for the complex problem of voiced versus voiceless stops in special cases and areas, cf. Mirambel, “Le traitement du groupe nasale+occlusive dans les parlers néo-grecques et le problème de la classification,” Bulletin de la Société de Linguistique de Paris XXXIV (1933), 145–164.
- The whole range, from A. T. Kokkinakis and P. Andreou, Spoken Greek (New York: Cosmos, 1945) to N. Catone, Grammatica neoellenica (Roma: Edizioni dell' Ateneo, 1960), though not dealing with phonemic principles, gives an indication of the great difficulty encountered in the velar order when it comes to practical description. In this range, we have also included “classics” such as J. T. Pring, A Grammar of Modern Greek (London: University of London Press, 1950).
- The whole problem and its solution can be seen in Newton's new correlations between the velar and the palatal spirants (p. 282):
- [ça çe xa] = /xia xe xa/
- [ja je γa] = /γia γe γa/.
- Explanations are given in 4.0 where each decimal corresponds to its counterpart in 3.0. The key to our code appears below, on page 78.
- The number of unstressed syllables preceding a final stressed syllable could be more than six in polynominal compounds such as ἀερoθερμoδμναμική Our syllabic analysis, however, distinguishes between polynominal compounds and mononominal compounds in which components are either nouns or the like in conjunction with compounds such as prepositions, diminutives, verbal prefixes, and similar particles. Polynominal compounds, each of them theoretically unlimited in the number of components and more apt to be created in katharevusa, are not considered in the PW analysis of dimotiki. In some dialects, incidentally, it is possible to hear three or even four unstressed syllables after the stressed one. See G. Courmoulis, “Toνικά Tινά πρoβλήματα νέας ‘Eλληνικης” in the Scientific Yearbook published by the Faculty of Letters of the University of Athens, 1955-1956, pp. 439–468. Cf. Catone, p. 11.
- In other words, Ic may be present in a Sm only when preceded by Sm or Sm. A case in point is offered by the cluster /-tl-/ which is the Im of the Sm in /a-tlá-zi/, ἀτλάζι ‘satin.’
- It is a conditio sine qua non that if a T is present in Se, then it must be Tf.
- The obligatory rule is that the syllable preceding the Ic must have a Tn = Ø.
- When we checked our clusters against the partial list given by Mirambel (La langue grecque moderne, passim), we found several discrepancies, that (1) some clusters are just graphemic associations of individual phonemes, e.g. */sl/; (2) some clusters are given both as graphemic associations and as a phonemic unit, e.g. */ts/ and “/t/”; some clusters are shown as graphemic and phonemic clusters at the same time, e.g. sm and /zm/; some clusters as a combination of phonetic and phonemic, e.g. *[nx/; and some clusters were not given at all, e.g. /zb/, /zγ/, etc. Finally, it is difficult to see why Mirambel considers, among other clusters, /pt-/ as one of the “savants et rares” (pp. 51–52), while /tm-/ is not. In his Petit dictionnaire, /tm-/ occurs only once, while /pt-/ recurs six times and may be present in both dimotiki and katharevusa as, respectively, in /ptósi/ πτώση ‘fall’ and /ptósis/ πτωσις.
- In reviewing Mirambel's La langue grecque moderne, Swanson (Language XXXVIII (1962), p. 195) says: “Mirambel denies (54–5) that word-final consonants occur in Modern Greek except -s and -n. Although these phonemes are statistically very frequent, this rule is not valid, since we must consider the presence of recent loans in Modern Greek, such as (from English) bridz (the card game), dzip, tánk (s), or from French likér, spór.” We disagree with Swanson on the ground that in most problems of borrowing and of languages in contact there are often structural boundaries which, though allowing the “import,” still forbid the assimilation of the new Tf into dimotiki. This is not a phenomenon of Modern Greek only, but one based on general principles. As a parallel case in point, who would assert that a Tf in Italian may be -k because we imported one or more words such as oc (as in la lingua d'oc)? The word oc in Italian will always be “foreign” to the structure of Italian until it becomes “naturalized,” i.e., until it follows the syntagmatic rules of the spoken chain that are typical of Modern Italian, and not of Provençal or Classical Latin. Therefore, when Migliorini speaks of oc (ók) on p. 170 of his Storia della lingua italiana (Firenze: Sansoni, 1961), he employs a morpheme foreign to the structure, but when he says oco [óko] on p. 181, then the morpheme is already integrated into Modern Italian.
- The admission of -p from American jeep or of -r from French liqueur and the like into Greek would imply that dimotiki (in its syntagmatic correlations) may theoretically accept any final consonant from borrowing. But the evidence rejects this supposition when we see that the tendency, if not the rule, of dimotiki is to “naturalize” in due time any borrowings by adding a vowel or any other morphological signal which has a parallel in the language. Otherwise, how would we explain that English bar becomes μπάρα [bára] in American Greek where it should have not, and μπαρ [bar] in Greece where it is still a foreign word? The so-called “problematic words” (pp. 37–38 in Swanson's list) such as blófa and pútinga (English bluff and pudding) tell us that they are borrowings old or popular enough to have acquired Demotic morphological garb, while dzip (<l jeep) has not yet had the time to do so. Cf. the case of to tirum and to tirúmi (< THE TEA-ROOM) as reported by Newton in “Grammatical Integration of Italian and Turkish Substantives in Modern Greek,” Word XIX (1963), p. 29.
- Thus, the phoneme /f/ is an initial element in a diphonemic cluster, as in 3–5(2) and 3.5.2, but never an initial element in a tri- or tetraphonemic “potential” cluster. In a case like this, /m/ becomes a Tn. The same behavior is displayed by /m/, /n/, and /k/. Thus, while the /f/ in /af-sti-rós/ is a Tn, in /a-ftó-Θ) ἀντóθι ‘ibidem,’ the /f/ is part of an Ia (3.5(2)).
- It is more “economical” to apply simultaneous rules whenever these constitute variants of the same basic nature. In this case they are three variants of the syllabic structure and the simultaneous application of these variants saves two steps. Whenever feasible, similar procedures are followed in the basic steps shown below.
- Cf. “La syllabe phonologique” in A. Rosetti, Sur la théorie de la syllabe ('s-Graven-hage: Mouton, 1959), pp. 23–25, and G. Straka, “La division des sons du langage en voyelles et consonnes peut-elle être justifiée?” in Travaux de linguistique et de littérature (Strasbourg: Centre de Philologie et de Littérature Romanes, 1963), pp. 23–25.
- In connection with the sequence of vowels /-ía/, we assume that at the phonological level there is no monophonemic unit called diphthong. Cf. 2.3 above, in which the semivowel (or semiconsonant) was considered an allophone of /i/. In dimotiki, incidentally, all combinations of vowels, with the restrictions of 3.4 and 3.4.1, above, are “grammatical.” Consequently, no sequence of two or more vowels is ever possible within a phonological syllable. This view is strengthened by the fact that, contrary to the development of Latin to the Romance languages, no paradigmatic diphthongs have ever emerged in the development from Classical to Modern Greek. Modern Greek “diphthongoids” are genetically syntagmatical, and as such are due to the association of vowels in the spoken chain. Cf. paradigmatic versus syntagmatic in L. Romeo, “Structural Pressures and Paradigmatic Diphthongization in East Romance,” Word XIX (1963), page 4, note 11. The term “diphthongoid,” as used here, is not to be confused with the same term as used by E. Petrovici in “La fonction phonologique des ‘diphtongues’ roumains,” Revue de linguistique I (1956), 27. Cf. L. V. Šˇcerba, Fonetika francuzkogo jazyka (Moskva: Izdatel'stvo literatury na inostrannyx jazykax, 1953), page 36.
WORD
Volume 20, 1964 - Issue sup2: Papers in Memory of George C. Pappageotes