^{page 578 note 1}For levels of analysis see Firth, J.R., ‘The technique of semantics’, TPS, 1935, 36–72,Google Scholar and ‘Modes of meaning’, Essays and Studies (English Association), 1951, 118–49, especially 119–21. Also Robins, R.H., ‘The phonology of nasalized verbal forms in Sundanese’, BSOAS, XV, 1, 1953, 138–15.CrossRefGoogle Scholar

^{page 578 note 2}This is an analysis of Istanbul Turkish in slow, careful style. I was very fortunate in having the help of Professor Fahir Iz of Istanbul University, who kindly consented to act as my informant, when he held a lectureship at SOAS. It is a great pleasure to acknowledge the sacrifice of his time in this work.

^{page 578 note 3}Firth, J.R., ‘Sounds and prosodies’, TPS, 1948, 127–52.Google Scholar

^{page 578 note 4}Forms of the nominal including suffixes of the type for deriving nouns from verbs and verbs from nouns are not included in the analysis.

^{page 578 note 5}As two examples of such treatment see Deny, J., Grammaire de la langue turque (dialecteosmanli), Paris, 1921,Google Scholar and N. McQuown and Sadi Koylan, Spoken Turkish (Holt Spoken Language Series), Ling. Soc. of America, 1946.

^{page 578 note 6}This structure is not to be confused with the structure (C)VCəC which is a dissyllabic structure. See p. 586.

^{page 579 note 1}See Appendix pp. 589–91 and Plates i and u for palatograms showing the difference in articulation of a few consonants with front and back vowels.

^{page 579 note 2}‘Word’ here includes both what is called the unsuffixed form and the suffixed form. For division into base and suffix see p. 581.

See also note 1 above.

^{page 579 note 3}Seen on palatograms as having a generally greater degree of contact of tongue on palate.

^{page 579 note 4}In phonemic analyses, the consonants and vowels are considered separately, and apart from k and l, which have long been singled out for special notice, the difference in articulation of consonants with back and with front vowels does not appear to have been noted in Western publications. Soviet linguists have recognized this difference in articulation but treat it as allophonic, after the way of Russian Phonetics. See Baskakov, N.A., Karakalpakskij jazyk II. Fonetika i morfologija. Čast' pervaja, Moscow, 1951;Google ScholarRamazanov, S. and Khismatullin, Kh, Tatar tele grammatikasy, Kazan, 1954,Google Scholar and Grammatical Supplement to Dmitriev, N.K. (ed.), Russko-ČuvaŠskij slovar, Moscow, 1951.Google Scholar

^{page 579 note 5}The prosodies for polysyllabic words will be the same, the formulae are given for monosyllabic words for simplicity of statement.

^{page 579 note 6}Examples are given in the Turkish orthography which is in general an adequate guide to the phonetic realization if one bears in mind note 4 above. In the few instances that it is inadequate, examples are given in I.P.A. in brackets after the orthographic form.

^{page 580 note 1}In the orthography open vowels are represented by o a ö e, and close vowels by u i ü i.

^{page 580 note 2}In loanwords and onomatopoeic words O can operate with α in all syllables of the word, e.g. loan: komposto CαC-CαC-Cα and onomatopoeic word: ṡorolop Cα-Cα-CαC.

^{page 580 note 3}Some alternative forms are possible in structures of this type, e.g. kapi/kapu

^{page 581 note 1}It is not intended to indicate in the formulae exactly where w prosody ends and y prosody begins but only to show that there is a w beginning and y ending.

^{page 581 note 2}See Firth, J.E., ‘Sounds and prosodies’, TPS, 1948, 143.Google Scholar ‘Even if's true and strew should happen to be homophonous, the two structures are different: and ‘cvw.’

^{page 581 note 3}See p. 580 for the treatment of some loanwords.

^{page 581 note 4}See pp. 586–7 on benim, bizim.

^{page 582 note 1}Values are given in I.P.A.

^{page 582 note 2}In this analysis words with daş/taş would be treated as compound words rather than suffixed forms as daş/taş is always w prosody, e.g. arkadaş ^wVCCV-CVC, meslektaş ^yCVCCVC. ^wCVC. However, H prosody operates here as for C_p suffixes and the analysis therefore suggests that this is a transitional stage of a word being used as a suffix.

^{page 582 note 3}This applies to all base final V structures. See p. 580.

^{page 582 note 4}Where there is a long vowel in Base final, represented in the orthography by vowel + ğ, e.g. dağ [da:] tuğ tu:iğ [ i:], Base final structure is treated as (C)VC because the integrating prosodies at the junction of Base final and Suffix initial are the same as for Base final C, and not as for base final V. This structure will fit into the k system of Base final C. dağ [ da:] is a monosyllable with the structure CVC. Here ğ represents what is known to have been a voiced velar plosive in an earlier stage of the Turkish language. The vowel is long only in syllable final position, i.e. in absolute final and pre-consonantal position, e.g. dağ [da] dağda [ da:da]but short in prevocalic position, with a voiced transition from the first vowel position to the next, e.g. The phonetic exponents of Base final C in this CVC structure are therefore vowel length in syllable final and voiced glide in syllable initial position. In rapid colloquial (which is not being dealt with here) the forms daği and dağa may be pronounced as one syllable [da:] making no difference between the three forms dağ, dağa, and daği.

Some loanwords of Arabic origin having a long vowel in Turkish where the Arabic had or are similarly treated as having CVC structure. Here the phonetic exponents of base final C are long vowel in syllable final and glottal stop in syllable initial, e.g.:– cami (, Redhouse'sTurkish dictionary, London, 1880,Google Scholar gives CVCVC and camii menşe [mēn∫ε:] and menşei [m–n∫ε?i]

^{page 582 note 5}Base final C is underlined to show its position in syllable initial and syllable final.

^{page 583 note 1}When the terms for there is voicelessness and plosion in syllable final position and voice with absence of plosion in syllable initial, unless a nasal precedes. In a y prosody word, there is a palatal glide and in a w prosody word, there is a velar glide (with no friction), both glides being represented in the orthography by ğ.

These glides are the phonetic exponents of one of the four terms of the Base final subsystem and are therefore C in the structure.

^{page 583 note 2}Structures with base final C_p with the value k may have no phonetic exponent for base final C in the form with suffix cik, e.g. eşekçik or eşecik, structure

^{page 584 note 1}Systems for C are not given unless relevant.

^{page 584 note 2}See pp. 588–9 for examples of such forms.

^{page 584 note 3}See S prosody above.

^{page 584 note 4}3rd per. plur. possessive suffix lari is analysed into plur. suffix lar which comes under Z prosody on p. 587, and 3rd per. sg. poss. suffix VCn.

^{page 584 note 5}There is an exception to this, yarimşar, i.e. the exponent is sibilance with base final C.

^{page 585 note 1}The pronominal forms bu, şu, o, kendi, and the suffix ki are similarly treated as C fina structures, with a 1 term system for C, dental nasal in all positions except absolute final, when there is no phonetic exponent, e.g. bunu bunlar but bu Burada, şurada, etc., are treated as compound words, as bu, şu have their absolute final forms. This analysis agrees with the history of the Turkish language. The treatment of these pronominals as having base final C structure instead of the customary base final V structure has the same advantage as for the 3rd per. sg. poss. suffix in that no n variants have to be set up. su ‘water’ is the only noun which has up to the present been considered irregular; that is because of its form suyu instead of the expected susu for 3rd per. sg. poss., and the form suyun instead of the expected sunun for gen. suffix (see p. 587). Here there is no reason for it to be considered irregular, as it is treated as a base final C structure. The phonetic exponent of base final C is a palatal glide in syllable initial but there is no phonetic exponent in syllable final position, e.g. suyun sular . Cf. Uzbek cyb [ suw] ‘water’, also CVC structure, where the phonetic exponent of base final C is bilabial Motionless continuant in syllable initial and syllable final.

^{page 585 note 2}As final C of VC_n is a nasal, it is classed as a continuant and comes under the k sub-system. See p. 582.

^{page 585 note 3}Not to be confused with C_n in the structure VC_n, where C_n is suffix final and has no phonetic exponent in absolute final position.

^{page 586 note 1}The phonetic exponents of I and a here are the same, i.e. close, back, rounded vowel.

^{page 586 note 2}The phonetic exponents of the first C are voiced frictionless glide in syllable initial position, Oğul [ o'ul], and vowel length in syllable final position, Oğlu [ O:'lu]. Cf. dağ, daği, etc., on p. 582, n. 4.

^{page 586 note 3}Note also words like rükünden CVCəC-CVC, rüknü CVCəCV, where the second C of the base is a plosive, and vakit CVCəC, vakti CVCə-CV, where both are plosives, which have the same syllabic structure as the examples given above. Loanwords like sarp, fevk, zamk, where the 2nd C of the base is a continuant and the 3rd C is a plosive have similar structure to native Turkish words (see pp. 578–9), i.e. CVCC. In words of Arabic origin having final CC structure, where the Cs are homorganic, the second C has no phonetic exponent in syllable final position, e.g. hissi histen sirn sirdan muhikki muhikten muhilli muhilden gammi gamdan CVCC-CVC, hakki haktan .

^{page 586 note 4}In this structure C_n is always in absolute final position and is different from C_n in the structure VC_n. See p. 585, n. 3.

^{page 587 note 1}The post-position ile does not come under this group. Its spoken forms, i.e. -la/-le, -yla/-yle, permit it to be treated as a suffix. In that event, another prosody, ‘i’ has to be set up to cover its junction with base final. Suffix structure is C_iV. The phonetic exponent of i prosody with base final V will be yotization, e.g. usturayla and with base final C there will be no .exponent, e.g. vapurla

^{page 588 note 2}See pp. 582–3.

^{page 589 note 1}These palatograms were made by Yildiz Serpen, a native speaker of the Istanbul dialect. It is a pleasure to acknowledge her participation in this rather arduous work. Her willingness and enthusiasm greatly contributed to its success.

^{page 589 note 2}For the technique of palatography see Firth, J.R., ‘Word-palatograms and articulation’, BSOAS, XII, 3 and 4, 1948, 857–64,CrossRefGoogle Scholar and Firth, J.R. and Adam, H.J.F., ‘Improved techniques in palatography and kymography’, BSOAS, III, 3, 1950, 771–4.CrossRefGoogle Scholar

^{page 589 note 3}I am pleased to acknowledge the assistance with the technical side of the palatographic work given by Mr. H. J. F. Adam, Chief Technician at SOAS, who also drew the Palatogram Figure.