¹ Although less widely used than emphasis, other terms can also be found in the literature, e.g. velarization, uvularization, pharyng(e)alization, retraction, strong articulation, u-resonance, heaviness.

Because of typographical limitations, emphasis (except in transliterated terms cited in italics) in this paper is not marked by the conventional subscript dot but by a lowered dot preceding the letter: /.t/ represents emphatic /t/.

² These three terms perhaps denote the same feature (s). They are applied to overlapping classes of sounds. Sounds characterized by ?iṭbāq are (using the conventional symbols /.t/ and /.d/, although they are misleading for early Arabic in that the first represented a voiced stop and the second a voiced lateral spirant) /.s .d .t ./, by /.s .d .t . q x γ/, and by tafxīm the foregoing seven and in certain environments also /l r/ as well as some vowels and semivowels; later grammarians add /ħ 9/ to the last two classes. The term tafxīm is rarely used by the earliest grammarians; it is also more often found in the Orthoepic than in the grammatic literature.

³ In the preparation of this paper, I have profited from discussions with many colleagues, particularly Saad Gamal-Eldin, Hilmi Aboul-Fetouh, and Abdelghany Khalafallah, and from criticisms and suggestions made after oral presentation of earlier drafts at the meeting of the Canadian Linguistic Association in Montreal, June 1961, and at the 1961 Linguistic Institute in Austin, Texas. My greatest obligation is to Abd el-Rahman Ayoub, Professor of Linguistics at Cairo University, who read and criticized the first draft (written in Cairo in 1959). I am indebted to him for checking most of my data and for pointing out to me phonetic details about emphasis which I had earlier failed to observe.

⁴ A brief summary of the classificatory system devised by Arab grammarians for describing the sounds of Arabic is in Jean Cantineau, Cours de phonétique arabe 17–25 (Algiers, 1941; reprinted Paris, 1960), and in K. Vollers, The system of Arabic sounds as based upon Sibaweih and Ibn Ya'ish, Transactions of the Ninth International Congress of Orientalists 2.130–54 (London, 1893); on emphasis: Cantineau 23–4 and Vollers 147–51; also A. Schaade, Sībawaihi's Lautlehre 15–6, 18, 22–3, 38, 40–1 (Leiden, 1911); and I. Anis, Al-?aṣwīt alluγawiyyah ² 54–63 (Cairo, 1950). M. Bravmann, Materialien und Untersuchungen zu den phonetischen Lehren der Araber (Goettingen, 1934), is much less useful since it is based primarily on later orthoepic writers.

⁵ The pairing of these consonants is done by Sībawayh (late 8th century a.d.), who says: ‘if there were no emphasis (?iṭbāq), /.t/ would become a /d/, /.s/ a /s/, and // a //; /.d/ would disappear from the language, because apart from it, there is nothing else with its point of articulation.‘ (Regarding the use of these symbols, see fn. 2.) Translated from quotation in Anis, Al-?aṣwāt 59.

⁶ That in Classical Arabic there was in fact a contrast between emphatic and plain /l/ has been demonstrated by Charles A. Ferguson, ‘The emphatic l in Arabic’, Lg. 32.446–52 (1956).

⁷ The possibility of different analyses has been recognized before; see Ferguson, Lg. 32.451–2 (1956).

⁸ The uncertainty here reflects our lack of adequate synchronic descriptions for most modern dialects of Arabic.

⁹ This statement of vowel contrasts holds for most but not all of my informants. Some of them have eight contrasts stressed (as above), but only three unstressed /i a u/.

¹⁰ In all essential details, this paragraph follows Richard S. Harrell, The phonology of colloquial Egyptian Arabic (New York, 1957). Since phonetic details are readily available in Harrell, in W. H. T. Gairdner, The phonetics of Arabic (London, 1925), and elsewhere, they are not repeated here. To the above list of consonants /p q v ž/ should probably be added, because although of very limited occurrence and only in borrowings, they have been observed in the speech of some monolingual Cairenes. However, since this is not of central importance for our purposes, these consonants are here omitted.

¹¹ All of these features have been previously reported, but usually only one or two by any one investigator; see S. Spiro, A new practical grammar of the modern Arabic of Egypt 2–3 (London, 1912); T. F. Mitchell, An introduction to Egyptian colloquial Arabic 6–7 (London, 1956); Gairdner, Phonetics 20; Harrell, Phonology 69–71. Apparently the only other investigator of CrA who has observed the lateral spreading of the tongue is Mitchell, ‘Prominence and syllabication in Arabic’, BSOAS 23.369–70, fn. 3 (1960). Mitchell here describes the lip position for emphatic segments as ‘neutral’ and as ‘spread’ for the plain.

The emphatic segments could be described as having two points of articulation or biarticulated, as Sībawayh did; see Vollers, Transactions 2.150; Cantineau, BSL 46.1.xxv (1950); and Schaade, Sībawaih 15, 23.

¹² Most studies of ‘Egyptian Arabic’ are in fact based either exclusively or primarily on CrA, which should probably be recognized as a separate dialect. The dialect picture in Egypt is complex, probably more so than has generally been recognized, and has not been adequately investigated. Nevertheless it is doubtful that the label ‘Egyptian Arabic’ has any linguistic validity.

¹³ Harrell, Phonology 74.

¹⁴ Ibid. 73. With the pharyngeal spirants Harrell includes the uvular stop /q/. This, however, occurs only in classicisms or in borrowings from Classical Arabic and is always and only emphatic.

¹⁵ It was Ayoub (see fn. 3) who first drew my attention to the fact that the pharyngeal spirants do indeed show emphatic-plain contrasts. Emphatic pharyngeals have also been reported for other dialects; Haim Blanc, Studies in north Palestinian Arabic 71–2 (Jerusalem, 1953).

¹⁶ The one example of emphatic VC reported by Harrell, Phonology 78, does not hold for CrA, in which ‘voice’ and ‘whip’ are homophonous: /.s o.o.t/. Since ‘voice’ is written with initial ṣād (usually /.s/) and ‘whip’ with initial sīn (usually /s/), one can of course find speakers who, knowing the difference in spelling, will if asked about it differentiate between them. Telling in this regard is the fact that this word for ‘whip’ is literary and neither generally known nor used by people who use whips. The further fact that ‘examples where emphasis … does not extend through the entire syllable are … common … in traditional spelling’ (ibid.) is irrelevant to the question.

¹⁷ Ibid. 71–8.

¹⁸ The following are examples of the traditional analysis. J. S. Willmore, The spoken Arabic of Egypt (London, 1901), writes only the four emphatics but refers to others; he also cites numerous examples which show that some items now plain were emphatic in Classical Arabic. Gairdner, Phonetics, writes also emphatic and plain /l/, but in his Egyptian colloquial Arabic (London, 1926) writes only the four emphatics. Mitchell, Introduction, writes only the four emphatics but cites minimal pairs for emphatic and plain /l r/, and in the glossary uses notes to indicate emphatic /l/ and an asterisk for emphatic /r/. Gairdner, Phonetics, and Mitchell also recognize /q/ as emphatic.

¹⁹ As far as I know, this analysis has never appeared in print; examples of it are in the unpublished MSS by Saad Gamal-Eldin, Morphophonemics of colloquial Egyptian Arabic (University of Texas thesis, 1959), and Hilmi M. Aboul-Fetouh, A morphological study of Egyptian colloquial Arabic (University of Texas dissertation, 1961). Both authors state that all consonants show emphatic-plain contrasts.

²⁰ Again, as far as I know, this analysis has never appeared in print; an example of it, for a different dialect, is in the unpublished MS by Abdelghany A. Khalafallah, A descriptive grammar of sại:di colloquial Egyptian Arabic (University of Texas dissertation, 1961).

²¹ A transcription in terms of analysis B might be useful in Arabic texts for speakers of English, who tend to hear only the vocalic and not the consonantal differences. To them /?ab/ and /?.ab/ differ only in the vowels; the vowel of the first is identified with that of English bet and that of the second with but.

²² As far as I have been able to determine, this alternative was first suggested by Ferguson, although he did not mention it in publication until 1956 (Lg. 32.451–2). However, Zellig S. Harris, who published a somewhat similar analysis of emphasis for Moroccan Arabic in JAOS 62.309–18 (1942), gives Ferguson credit for suggesting it.

²³ Phonology 78–82, from which unless otherwise indicated all quotations in this section are taken. In his monograph, Harrell uses the term prosodic, not suprasegmental. Since, however, his use of this term differs from its use by some other linguists, e.g. Firth, and since he has informed me that by prosodic he means nothing more nor less than what American linguists traditionally call suprasegmental, to avoid confusion I will substitute suprasegmental for his prosodic and use only the former term in discussing his analysis. A revised and clearer statement of Harrell's analysis is in his article ‘A linguistic analysis of Egyptian radio Arabic’, Contributions to Arabic linguistics 26–30 (Charles A. Ferguson, ed.; Cambridge, Mass., 1960).

²⁴ Blanc in a review of Harrell, Word 15.541 (1959).

²⁵ See also the pertinent remarks by Blanc in Word 15.539–43 (1959).

²⁶ Bernard Bloch, Lg. 24.46 fn. 43 (1948).

²⁷ Kenneth L. Pike, Phonemics 14, 63 (Ann Arbor, 1947).

²⁸ Charles F. Hockett, A course in modem linguistics 168 (New York, 1958). For a brief but exceptionally useful discussion, see Rulon S. Wells, ‘The pitch phonemes of English’, Lg. 21.27–39 (1945).

²⁹ The segmental nature of emphasis in CrA is discussed in more detail in Gamal-Eldin, Morphophonemics 8, and Aboul-Fetouh, Morphological study 4, 15.

³⁰ There is complete agreement among all investigators of CrA on syllable structures. Those given above are either identical with those given elsewhere or mechanically convertible into them; e.g. Harrell, Phonology 30–1 et passim; Mitchell, BSOAS 23.370–1 (1960).

³¹ A. A. Hill, in a lecture at the 1961 Linguistic Institute, suggested the terms phoneme, prosodeme, and suprasegmental; phoneme to be restricted to segmentals, suprasegmental to those features which extend over more than one segment and which pattern as non-segmental morphemes (e.g. pitch and stress), and prosodeme to those features which extend over more than one segment but pattern with (segmental) phonemes as segmental morphemes. See now the published version of that lecture, Lg. 37.457–68 (1961). In these terms, emphasis is a prosodeme. In Hockett's terms, emphasis is a component, not a phoneme, since (however complex phonetically) structurally it is unitary, one of 'the not-further-decomposable elements out of which all larger phonological elements, up to whole utterances …, are built.' See 'Linguistic elements and their relations', Lg. 37.44 (1961).

³² These terms in phonologic analysis are used as defined by Hockett, Manual of phonology 150–4 et passim (Baltimore, 1955).

³³ After oral presentation (fn. 3), W. S. Allen and G. L. Bursill-Hall drew to my attention a reference to emphasis in CrA by Firth, ‘Sounds and prosodies’, TPS 1948 [reprinted in Papers in linguistics 1984–51 121–38 (London, 1957)]. Although different terms have been used, analysis D as here outlined seems to be the same analysis as suggested by Firth. Unfortunately he never published such an analysis, referring only by way of illustration to CrA. In his terms, emphasis would be a syllable prosody.