zemanekpharyngealization

92
THE ORIGINS OF PHARYNGEALIZATION IN SEMITIC PETR ZEMÁNEK en eni gma gma corporation, corporation, Ltd. Ltd. Praha Praha 1996 1996

Upload: fergananym

Post on 26-Aug-2014

37 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: ZemanekPharyngealization

THE ORIGINS OF

PHARYNGEALIZATION

IN SEMITIC

PETR ZEMÁNEK

eneniigmagma corporation,corporation, Ltd.Ltd.PrahaPraha 19961996

Page 2: ZemanekPharyngealization

2

The Origins of Pharyngealizationin Semitic/Petr ZEMÁNEK

- Prague :enigma corporation, Ltd. 1996(Prague : T.R.S. Polygrafické služby)

ISBN 80-9010704-3-9

Published by

enigma corporation, Ltd.P.O. Box 308

CZ 111 21 Praha 1

Page 3: ZemanekPharyngealization

3

© 1996 byenigma corporation, Ltd., Prague© 1996 by Petr ZEMÁNEK

All rights reserved, including those of translation into foreign languages. No part ofthis book may be reproduced or transmitted in any form or by any means, electronic ormechanical, including photocopy, recording, or any information storage and retrievalsystem, without permission in writing from the publisher.

Printed in the Czech Republic. ISBN 80-9010704-3-9

Page 4: ZemanekPharyngealization

CONTENTS

Contents . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . iAcknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . iiAbbreviations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. iii

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . 11 Characteristics of emphasis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3

1.1 Characteristics of pharyngealized articulation in Arabic . . . . . . . . . 41.2 Characteristics of glottalized articulation . . . . . . . . . . . . . . . . . . . . . 61.3 Characteristics of individual emphatic consonants . . . . . . . . . . . . . . 7

1.3.1 Emphatic˙t . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71.3.2 Uvularq . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91.3.3 Emphatic˙d . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111.3.4 Emphatic˙s . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 121.3.5 Emphatic˙z . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 131.3.6 Other emphatic phonemes . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13

2 The relation between pharyngealization and glottalization . . . . . . . . . . . . 162.1 Analysis of the alternations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20

2.1.1 Emphatic˙t and its alternation with voicedd . . . . . . . . . . . . . . 202.1.2 Uvularq and its alternation with voicedg . . . . . . . . . . . . . . . 242.1.3 Emphatic˙sand its alternation with voicedz . . . . . . . . . . . . . . 29

2.2 Analysis of the control set: Phonetically conditioned doublets . . . 322.2.1 Emphatic˙t and its alternation with non-voicedt . . . . . . . . . . 332.2.2 Uvularq and its alternation with non-voicedk . . . . . . . . . . . . 362.2.3 Emphatic˙sand its alternation with non-voiceds . . . . . . . . . . 39

2.3 Analysis of the control set: Phonologically and phoneticallynon-related doublets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 412.3.1 Doublets with the alternation ofb / r . . . . . . . . . . . . . . . . . . . 422.3.2 Doublets with the alternation of¯t / g . . . . . . . . . . . . . . . . . . . . 44

3 Summary and conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 463.1 Evaluation of the sets of alternations . . . . . . . . . . . . . . . . . . . . . . . . 463.2 Phonetic environment of the alternations . . . . . . . . . . . . . . . . . . . . . 493.3 Semantic distribution of the alternations . . . . . . . . . . . . . . . . . . . . . 503.4 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 503.5 Model of the development of pharyngealization in Arabic/Semitic 54

4 Supplement: Lists of doublets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . 81

Page 5: ZemanekPharyngealization

ACKNOWLEDGMENTS

The present study is an attempt at the explanation of the origins of the phary-ngealized articulation in Arabic and Semitic and is a culmination of several yearsof work, although many times interrupted. The interest in this type of articulationwas inspired already during my studies in the beginning of eighties by ProfessorPhDr. Karel PETRÁCEK, CSc., who also guided me through the first steps in thestudy of this phenomenon. I am glad to get the opportunity to express my gratitudefor his teachings and his inspiring ideas.

Through my work at this study, it has been at various stages consulted anddiscussed with several scholars, who have contributed in important ways to thegrowth of my knowledge and understanding at different stages of my journey.These include notably Arne A. AMBROS, Michael G. CARTER, Ladislav DROZDÍK,Jaroslav OLIVERIUS, Zdena PALKOVÁ , Stanislav SEGERT and Petr VAVROUŠEK.

Needless to say, none of these scholars are to blame for any errors that remainin the work.

Finally, I owe my thanks to James LANCASTER, who willingly corrected myEnglish. I know that I have surely missed mentioning others who truly deserveinclusion here, and to them I apologize most deeply.

Page 6: ZemanekPharyngealization

ABBREVIATIONS

AION Annali di Istituto Orientale Universitario di Napoli, Napoli.ArOr Archív orientální, Praha.BSOAS Bulletin of School of Oriental and African Studies, London.C consonant, consonants.GLECS Groupe linguistique des études chamito-sémitiques, Paris.JAOS Journal of the American Oriental Society, New Haven.JCS Journal of Cuneiform Studies, Baltimore.JNES Journal of Near Eastern Studies, Chicago.JQR Jewish Quarterly Review, London, Leiden.LAR laryngeal, laryngeals.LIQ liquid, liquids.MSA Modern Standard Arabic.OA Oriens Antiquus, Roma.R1(2,3) the first (second, third) radical of the root.SON sonant, sonants.SV semivowel, semivowels.V vowel, vowels.WZKM Wiener Zeitschrift für die Kunde des Morgenlandes, Wien.ZA Zeitschrift für Assyriologie und Vorderasiatische Archäologie,

Berlin, New York.ZDMG Zeitschrift der Deutschen Morgenländischen Gesellschaft, Wies-

baden.

Page 7: ZemanekPharyngealization

1Characteristics of emphasis

Introduction

One of the characteristic phonological features in Arabic, but also in majorityof other Semitic languages, is the so-called "emphatic" series. From the arti-culatory point of view, it is, especialy in Arabic with its dialects, a bunch ofvarious types of co-articulation. Under the heading "emphatic" we usually under-stand pharyngealized1 or glottalized consonants.2 Roughly it can be said that theformer are found in Arabic and Modern Aramaic (Neo-Assyrian, cf., e.g., HOBER-MAN 1988 or TSERETELI 1982), the latter in the Semitic languages spoken inEthiopia. In all of these languages, since they are still living ones, with the excep-tion of G ec ez, we have direct evidence on the phonetic realization of these con-sonants, while in case of other Semitic languages the character of the articulationcan differ. Traditionally, it is the pharyngealization that is ascribed to most of thedead Semitic languages. Modern Hebrew constitutes a special example in thisrespect because of cultural influence that results in de-emphatization in the pro-nunciation of originally emphatic sounds.

From the phonetic point of view, a lot of studies state that this correlationexhibits considerable growth, especially in the modern dialects of Arabic. Itshould be pointed out that this growth does not in many cases mean the growth ofnumber of pharyngealized consonants, but various types of co-articulation, likelabialization, (strong) nasalization, etc. This is, e.g., the case of emphaticb

˙(labia-

lization) or the case of emphatic˙n (nasalization). These issues will be treated indetail further on in the following part of this study.

However, there are still some aspects of this co-articulation feature in Arabicand Semitic languages that are not completely clear. The discussion on the originand type of development of this correlation is still going on. The aim of presentstudy is to investigate exactly these two aspects and attempt at a proposal of thedevelopment of the phonological feature of "emphasis" in Arabic and Semiticlanguages. The study tries to contribute to the discussion on which coarticulationof emphatics, i.e. glottalization and pharyngealization, predates the other. It isbased on data collected from the dictionary of Arabic (KAZIMIRSKI 1860), and thedata are evaluated from statistic, phonetic and semantic points of view.

The concept of emphatics in this study is mainly based on TRUBETZKOY’Sanalysis of co-articulation (Nebenarbeitskorrelation, 1939:122 ff.). The approach

1 Instead of the termpharyngealization, one can meet terms likevelarization(e.g., OBRECHT1968) or, rather rarely,uvularization(DOLGOPOLSKY 1977).2 In Semitic, especially post-glottalized ejectives.

Page 8: ZemanekPharyngealization

2 Petr ZEMÁNEK

to the phonological system used in this study proceeds from the non-statical,dynamic conception of the system, as given especially in the study of the phonolo-gical system of Czech by J. VACHEK,3 together with the reflexion of these con-cepts in the work of K. PETRÁCEK (1971, 1975, 1990:142-159 and 247-290). Theterminology used in this study corresponds mainly to the one used by ESSEN

(1979).

3 Dynamika fonologického systému soucasné spisovné ceštiny [The Dynamics of the Phone-mic System of Contemporary Literary Czech]. Praha 1968.

Page 9: ZemanekPharyngealization

3Characteristics of emphasis

1.0 Characteristics of emphasis

Let us first try to resume the situation of the "emphatics" in Semitic:4

Akkadian5 ˙t ˙s qEblaite6 ˙t ˙s q ˙d ˙zUgaritic ˙t ˙s q ¯t7

Hebrew ˙t ˙s qSyriac ˙t ˙s qSayhadic8 ˙t ˙s q ˙d9

G ec ez ˙t ˙s q ˙d pArabic (Classical) ˙t ˙s q ˙d ˙z ( ˙l ˙r)10

Neo-Assyrian ˙t p ˇ˙c ˙kModern South Arabian11 ˙t ˙s q ˙d ˙z ˇ˙s

In all the Semitic languages, there are at least three of them:˙t, ˙s andq,12

which holds especially for the Ancient Semitic languages. This could point to the

4 If not quoted differently, our data are taken from MOSCATI ET AL. 1964:43-45.5 For Akkadian, cf. JUCQUOIS 1966:264-266; for the problem of the original articulation cf.KNUDSEN 1961.6 The situation in Eblaite is not completely clear. KREBERNIK 1982:179 and 1983 passimpostulates 4 emphatic phonemes, which are nevertheless not attested in the texts. His analysisoffers reconstruction of the system, graphical manifestation of the phonemes in the texts isnevertheless heavily dependent on the possibilities of the cuneiform writing. DIAKONOFF1984:4 characterizes these phonemes as non-voiced - cf. also the use of the symbol˙z fornon-voiced "emphatic" phoneme in Ugaritic - e.g. GARR 1986, SEGERT 1984, etc.7 In the ugaritological literature, this phoneme is represented as˙z - cf. e.g. SEGERT 1984,KTU 1976 passim etc.; for discussion see e.g. VOIGT 1990 etc.8 Old Epigraphic Southern Arabic. The name˙Sayhadicis used according BEESTON1981.9 For the characteristics of this phoneme, cf. RODINSON 1981, VOIGT 1990, etc.10 The last two mentioned sounds function rather as variants, cf. reading of Quran, arti-culation ofwa- ˙l ˙l ahi, bi-˙l ˙l ahietc. For more details, cf. e.g. PETRÁCEK 1952, FERGUSON1956,AMBROS 1981. For possible emphatic˙r already in Classical Arabic cf. SCHAADE 1911.11 i.e. mehri, š˙heri, soqotri etc., languages spoken in Yemen.12 Includingq, which is nevertheless a little bit problematic from the point of view of modernSemitic, e.g. Arabic.

Page 10: ZemanekPharyngealization

4 Petr ZEMÁNEK

fact that the emphatic sounds in older phases were non-voiced, which is also theway the proto-Hamito-Semitic phonological system is reconstructed.13

In the West Semitic (i.e. Arabic, Hebrew, Syriac, Ugaritic and Neo-Assyrian)the emphatic consonants are pharyngealized, in Ethiopian Semitic (here represen-ted by G ec ez), they are glottalized. For Akkadian, there is at least a possibility thatthe emphatic sounds were glottalized (cf. KNUDSEN 1961), and Modern SouthArabian probably stands in the borderline between glottalization and pharyn-gealization.14

In the following, we will characterize the pharyngealized and the glottalizedways of articulation:

1.1 Characteristics of pharyngealized articulation in ArabicThe characteristics of emphasis in Literary Arabic, from the synchronic point

of view, is not connected with special difficulties. Articulatorily, it is characterizedby the raising of the dorsum to the back part of the velum (GAIRDNER 1925), byslight backing of the typical dental and alveolar articulatory position to the rear.15

They are produced with a primary constriction involving the tongue tip and bladein the anterior region of the oral cavity, and a secondary posterior-inferior move-ment of the back of the tongue into the pharynx region. This secondary movementmay be accompanied by a lateral spreading of the tongue dorsum thus creating amore hollow configuration.

Accoustically, as shown in OBRECHT 1968, the main characteristics is thelowering of the second formant (F2), which can be in some cases considerable.16

OBRECHT also points to the fact that the velarization17 is the most effective in

13 Cf. e.g. DIAKONOFF ET AL. 1987:12, where the following emphatic phonemes are recon-structed:p, ˙t, ˙c, ˇ˙c, ˆ˙c, ˙k, ˙kw,

˙x,

˙xw.

14 Cf. NAUMKIN & PORKHOMOVSKY 1981:9 for Soqotri, where they describe a free variationof pharyngealization and glottalization.15 ALI & DANILOFF 1972 give the following values:˙s is shifted to the back for 2.89mm whencompared tos, for ˙t/t the difference is 2.76mm.16 OBRECHT 1968:24-26 gives the following types of differencies in F2: t/ ˙t - 1800/1200 Hz;d/ ˙d - 1750/1050 Hz;s/ ˙s - 1500/1050 Hz;z/ ˙z - 1500/750 Hz. From the point of view of thelowering of F2, the values for opposition ofk/q are very similar to the ones given above -2100/1200 Hz.17 The term "velarization" was later on replaced by "pharyngealization", which, at least inArabic, is more exact.

Page 11: ZemanekPharyngealization

5Characteristics of emphasis

influencing neighbouring sounds - both consonants and vowels. It induces atongue backing gesture on adjacent sounds, much more clearly than pharyngeals.

A little bit more problematical is the characteristics of emphasis in the dia-lects of Arabic. In most of the dialects, the emphasis covers much more con-sonants than in Literary Arabic. Especially the Magrib dialects exhibit a highdegree of emphasis in their phonological systems. E.g. ZAVADOVSKY 1981 in hisanalysis of Mauretanian dialect stated that each consonant can have its emphaticand non-emphatic variant.18 Such a situation lead some authors to distinguishbetweenemphatisantandemphatisables.19 Nevertheless the degree of emphasisis dependent on further criteria. It is not always the pharyngealization that is gathe-red to the new emphatics in the Arabic dialects, they are sometimes characterizeddifferently. E.g. emphatic

˙b is rather characterized as labialized than pharyngeali-

zed, emphatic˙mand ˙n as strongly nasalized (MALAIKA 1959), etc.It is surely also the phonetic nature of the feature in question - i.e. the pharyn-

gealization - that leads to the rise of further emphatic phonemes in Arabic. Thiscan be observed even in Literary Arabic, where we can mention e.g. the assi-milation ˙tt > ˙t ˙t, ˙dt > ˙d ˙t etc. at the VIII verbal stem. The pharyngealization canfurther effect on the consonants in the non-contact position - cf. GAIRDNER

1925:51, and sometimes it is the whole word that is considered "emphatised" - cf.D. COHEN1969:63-6, BANI YASIN & OWENS1987:303 etc. Also phonetic studiesshow that the pharyngealization can considerably affect its neighbourhood. Someremarks can be found by OBRECHT1968; the studies of GHAZELI 1981 and HAMDI

1990 must also be mentioned. The first states that the effect of pharyngealizationdoes not pass beyond the boundaries of a syllable, the second claims that accord-ing to his measurements the pharyngealization can spread over the syllable boun-daries, but only towards the end of the word (i.e. progressive, not regressivespread). It seems that this situation is caused mainly by phonetic-articulatory

18 A similar situation is also described by BANI YASIN & OWENS 1987:303-5 for Jordaniandialect, where the authors allow for the possibility that each phoneme has its emphaticcounterpart. Nevertheless, as independent emphatic phonemes they name only the following:

˙d, ˙t, ˙s, ˙m, ˙l, g,˙b,

˙w.

19 The terminology was adduced by D. COHEN 1969, based on HARRELL 1957:69ff where inhis analysis of the phonological system of Egyptian colloquial Arabic, where he speaks aboutprimary emphatics (i.e.˙t, ˙d, ˙s, ˙z), secondary emphatics (i.e.˙r, ˙l, ˙k,

˙b), that occur quite rarely

and mostly in words where primary emphatics occur, and marginal emphatics (g, f, ¯t,˘h, g,

h, n, w, y, ’) - i.e sounds, where there is phonetic possibility of emphatizing these phonemes,but there are no minimal distinctive pairs. HOBERMAN 1988 and 1989 gives also phonemesthat stop the spread of emphasis, i.e. phonemes that cannot be emphatised.

Page 12: ZemanekPharyngealization

6 Petr ZEMÁNEK

reasons, i.e. the articulatory organs that are in a specific position remain in such aposition even after the articulation of the pharyngealized sound.

Such a situation clearly leads to attempts to create another concept of em-phasis, i.e. a concept connected not only with one phoneme but with a higher unit -syllable or even word. From 1942 we have an attempt to characterize emphasis asa suprasegmental feature - the study by Z. HARRISon Morrocan Arabic. From thisstudy, there were several similar attempts suggesting a similar solution, e.g.ZAVADOVSKY 1981, etc.

Apart from purely linguistic criteria and arguments it is possible to adducealso other criteria that can shed further light on the role of emphasis in the Arabicdialects. This feature has been used also as a criterion for socio-linguistic study(ABD-EL-JAWAD 1986), but also in other studies we can find remarks on stylistic"deemphatisation" (KHALAFALLAH 1969:30) and the use of emphasis can also bedetermined socially - the younger and more educated the speaker is, the lowerdegree of emphasis used.20

1.2 Characteristics of glottalized articulationUnder glottalized sounds various types of sounds have been understood:

ejectives, implosives, creaky orglottalized vowels, pre-glottalized continuants etc.For the Hamito-Semitic languages, the ejectives and implosives are the mostimportant. We can find implosives in Chadic (b’, d’ andy’) and post-glottalizedejectives especially in Cushitic languages, but they occur in other groups ofHamito-Semitic as well. In the frame of Semitic, the area with the greatest spreadof glottalized sounds is Ethiopia, where in the Ethio-Semitic the glottalized soundsmay have emerged or preserved due to the areal influence of the Cushitic lan-guages. Other area, where glottalized articulation alternates with the pharyn-gealized one, is the Southern Arabia, in Modern South Arabian, i.e. Mehri, Š˙h eriand Soqotri. All of the glottalized sounds attested in Semitic are ejectives.

These facts mean that for the purpose of this work we can basically limit ourscope to ejectives, i.e. eggressive post-glottalized consonants, since these repre-sent the main articulatory manner in Semitic.

Ejectives are produced with egressive pharynx air. The larynx moves upwardwith the vocal cords closed during the articulation of the consonant, which createsair pressure in the pharynx. After the release of the consonant articulation, the air

20 E.g. BANI YASIN & OWENS 1987:304 in note 8 mention that the lowest degree of empha-sis is in the category 30-45 years of age, with university education - 81% of non-emphaticvariants. The use of emphasis is also determined sexually - women use less emphasis thanmen.

Page 13: ZemanekPharyngealization

7Characteristics of emphasis

from the pharynx creates an egressive air stream originating from the pharynx. Wecan find glottalized plosives (p’, t’, k’ ) and glottalized affricates and fricatives (c’,c’ ~ s’).

It should be noted that it is physiologically impossible to produce other thanvoiceless ejectives, since the glottal closure does not allow participation of thevoice. On the other hand, implosives, which are attested in Chadic languages, tendto be voiced (cf. WEDEKIND 1990:128). Another point which is important forSemitic languages is that, according to WEDEKIND (1990:128) languages tend notto use the difference between a glottalized affricate (such asc’) and a glottalizedfricative (such ass’) for phonemic contrast. This might help in explanation of thearticulatory character of the Semitic˙sade.

1.3 Characteristics of the individual emphatic consonants in ArabicAs the basis of our analysis we include among the emphatic consonants the fol-

lowing five phonemes:˙t, ˙d, ˙s, ˙z andq.21 In addition, we will comment on thephonemes that can be emphatic or emphatized, but do not appear in the MSA.These phonemes can be traced especially in the dialects of Arabic or other Semiticlanguages. The phonemes that appear as emphatic in the phonological systems ofother Semitic languages are discussed briefly in the review of other Semiticphonological systems.

1.3.1 The emphatic˙tThe explosive unvoiced alveolar˙t is in Arabic situated in the center of the

Arabic emphatic sounds (mu˙tbaqa- cf. SIBAWAIH 1889, D. COHEN1969 etc.) andis attested in Literary Arabic as well as in almost all the dialects of Arabic. Theexception here is the situation in the peripheral Arabic dialects (Malta, Arabicdialects in Central Asia, Arabic dialects in Africa - Sudan, Chad, Kenya etc.). Incase of African dialects, this phoneme is substituted by its glottalized counterpart -cf. e.g. ZELTNER & TOURNEAUX 1986 as an injective for the Arabic dialects inChad. A similar situation can be found also in the Upper Egyptian dialects - cf.KHALAFALLAH 1969:30. Exceptionally, it is also possible to find it realized as theimplosived’ (in the dialects in Algerian Sahara - J. OWENS, personal communica-tion).

A close connection between the glottalizedt’ and pharyngealized˙t is shownalso by the perception similarity of the two sounds - cf. FRE WOLDU 1984-86 and

21 This phoneme is not always included among the emphatic phonemes.

Page 14: ZemanekPharyngealization

8 Petr ZEMÁNEK

1988.22 The possible alternation of these two types of articulation is also attestedby the data from Soqotri, where these two forms are perceived as allophones (cf.NAUMKIN & PORKHOMOVSKY 1981:9-10).

The data in FRE WOLDU 1984-86 and 1988 also show that this phoneme ismost probably the connecting bridge between the glottalized and pharyngealizedway of articulation. This phoneme in Sudan is the only one that allows the twofoldarticulation and both articulations are considered acceptable.

Although ˙t is attested in all the Semitic languages, its articulation in variouslanguages can be different. In all the Ethiopian Semitic languages it is realized asglottalized, the same is the situation in South Arabian and Neo-Assyrian. It isreconstructed for Akkadian, too, where it has been mostly represented by thegrapheme ford. The reason for this might be caused by the shortcomings of thecuneiform writing as indicated by MOSCATI ET AL. 1964:31. Nevertheless, thechoice ofdmight be caused also by other reasons. Our opinion is that analogicallywith the situation of the glottalizedk’ (q) in Akkadian (cf. KNUDSEN 1961) theemphatic˙t was in Akkadian realized as glottalized, or its graphical representationcan reflect the processes of the transformation of the glottalized articulation intothe pharyngealized one. In case of Hebrew and other Canaanite languages, thesituation is not completely clear as far as the manner of articulation is concerned,and we are still unable to decide which manner of articulation was the one used.

Outside the Semitic languages, in the Hamito-Semitic languages, we find thisphoneme only in the Cushitic and Omotic languages, represented as the glottalizedt’ (EHRET1980 and BOMHARD 1984 for the Cushitic languages, BENDER1975 forthe Omotic). For Chadic we find it only in the reconstruction of STOLBOVA 1986for Western Chadic, in the actual systems it is not attested (NEWMAN 1977 andJUNGRAITHMAYR & SHIMIZU 1981 do not reconstruct this phoneme even forProto-Chadic). In their reconstruction we find only voiced implosives. In Berberwe find it only in the reconstruction of Proto-Berber (PRASSE1972), APPLEGATE

1970:592-3 classifies emphasis in Berber as a suprasegmental phoneme,23 whichwould reflect a situation similar to the one in modern Arabic dialects or Modern

22 In these two studies, FRE WOLDU examines the perception similarity of pharyngealized andglottalized consonants in Tigrinya and Sudanese Arabic. In the perception test with theSudanese speakers, the glottalizedt’ is perceived as an acceptable articulation of pharyn-gealized ˙t.23 Emphatic consonants are treated there as a sequence of a consonant and an additionalfeature of pharyngealization. This additional feature is phonemic, but its domain is biggerthan the consonant (syllable, CV, etc.) and it has been found to affect nearly all of theconsonants of the phonemic system.

Page 15: ZemanekPharyngealization

9Characteristics of emphasis

Aramaic. The realization of the emphatic sounds in Berber is pharyngealized. ForEgyptian it is reconstructed in the protosystem only by RÖSSLER1971, whosereconstruction is nevertheless motivated by an obvious effort to postulate a clearlysymmetrical system. The reconstruction of the emphatic series is based only onanalogy with other series (voiced and non-voiced), i.e. analogically to the phono-logical systems in Semitic languages. No other analysis of Egyptian phonologyincludes emphatic˙t.

As we have seen, although this phoneme is not present in a number of actualsystems of the Hamito-Semitic languages, it is reconstructed for almost all theprotosystems (Semitic: MOSCATI ET AL. 1964, BOMHARD 1988, etc., Cushitic:EHRET 1980, Omotic: BENDER 1975, Chadic: STOLBOVA 1978, 1986, Berber:PRASSE1972, etc.) and similarly, it is reconstructed for the Hamito-Semitic (Afro-asiatic) system (cf. DIAKONOFF1988:34 ff., 1987; BENDER1975, BOMHARD 1984,HODGE 1987pr. etc.). The reconstructed manner of articulation is the glottalizedone.

1.3.2 Uvularq.Explosive non-voicedq is usually included by the old Arabic grammarians

among the emphatic consonants, though not in their center (i.e.mu˙tbaqa), but inthe mustacliya set. This phoneme is not always taken into consideration whenstudying the emphatic consonants. The reasons why we include it in our analysisare historical and systematic. We are aware of the fact that in case of articulativedefiniton of the emphasis (i.e. pharyngealization in case of Arabic) this phonemewould have to be excluded from the analysis. The features of this phoneme rankit rather with the back consonants (even pharyngeal - cf. DELATTRE 1971) ratherthan with the pharyngealized ones.

This phoneme is without doubt one of the most interesting phonemes of theArabic phonological system. This is i.a. proved by the rich bibliography connectedwith this phoneme and its problems. The realization of this phoneme in Arabic canbe very different (especially in Arabic dialects, but the differencies are tracableeven in MSA). In Arabic, it can be realized in the following way:24

- q - non-voiced uvular (CANTINEAU 1960:68; HARRELL 1962:5; DIEM

1972:9, etc.)

24 On the other hand, there is also some evidence for other phonemes that are realized asq -cf. the situation in Mauretania, where

˘h (non-voiced velar fricative) is realized asq (WILLMS

1972:1).

Page 16: ZemanekPharyngealization

10 Petr ZEMÁNEK

- g - voiced palatal (CANTINEAU 1960:68; ZAVADOVSKY 1962:7; DIEM

1972:9, etc.).- ’ - glottal stop (GAIRDNER 1925; FEGHALI 1919:25; CANTINEAU 1960:68,etc.).- g - voiced velar (postvelar) (CANTINEAU 1960:68, etc.).- k - non-voiced palato-velar, e.g. in Maltese (COHEN 1970) and other peri-pheral dialects of Arabic (e.g., HEINE 1982, OWENS1985, etc.- g - voiced affricate, e.g. in Iraqi Arabic -šarg ıfor šarq ı- eastern.

CANTINEAU 1960:68 differentiates the realization of this phoneme betweenthe sedentary population25 (q, ˙k, k, ’ - i.e. as a non-voiced consonant) and be-douins (g, g, g, g- i.e. as a voiced consonant). On the other hand, the differentarticulation can be based on other factors, too. This phoneme is used as a criterionfor sociolinguistic studies (cf. e.g. ABD-EL-JAWAD 1986, ROSENHOUSE1984,etc.).

According to GHALI 1983:436q in most Arabic dialects has a very lowfunctional load, nevertheless, such a situation may be caused by the above men-tioned variety of articulation of this phoneme.

In Classical Arabic, it is sometimes characterized as voiced (cf. medievalArabic grammarians, e.g. SIBAWAIH , ZAMA

˘HŠARI, cf. also SEMAAN 1968). Today,

in MSA it is realized as non-voiced postvelar or even more often, as uvular (cf.AL-ANI 1970, GIANNINI & PETTORINO1982:9 etc.).

Non-voiced explosiveq is attested in all the Semitic languages (MOSCATI ET

AL . 1964:44 and 45 table). The differences are in the realization of this phoneme -in a number of languages (Ethiopian, South Arabian, Modern Neo-Assyrian) it isrealized as glottalized. Its articulation in the dead languages is not completelyclear, there are indications, however, that point to the glottalized articulation (cf.KNUDSEN 1961, BOMHARD 1988:115-116 - both authors study the influence ofglottalization and pharyngalization on the neighbouring vowels, non-voicednessof the emphatics in Akkadian and the application of the Geer’s law (GEERS1945).The pharyngealized articulation would be then limited to Arabic and most pro-bably also to Canaanite languages. This region, together with Arabic, seems to bein case of emphasis the innovative area.

25 i.e. especially in the villages, the situation in the towns can be different.

Page 17: ZemanekPharyngealization

11Characteristics of emphasis

1.3.3 Emphatic˙dThe voiced explosive˙d is in Arabic traditionally included among the empha-

tic sounds. Its articulation is given in KÄSTNER1981 and AL-ANI 1970. In someregions of the Arab world it merges with emphatic˙z (Iraq, Tunisia, etc.).

In many studies, it is possible to find this phoneme listed as a member of theSemitic protosystem (cf. e.g. BROCKELMANN 1908:43, also MOSCATI ET AL. 1964as ¯d), but as early as by VILENCIK (1930:89ff) we find another position. MARRAS-SINI (1976:337) speaks on the delicacy of the problem of the classification of˙d inProto-Semitic, cf. also RODINSON (1981). In other Semitic languages it is notpossible to find the phoneme in all the systems, and furthermore the corresponden-ces with other Semitic languages show the possibility that this phoneme mighthave another type of co-articulation than pharyngealization (cf. M. COHEN

1931:10; MARTINET 1953:72 etc.). Also the correspondences with other Hamito-Semitic languages (cf. DOLGOPOLSKY1983:130-31 for Cushitic, BALDI 1987pr.for Hausa, CONTI 1976 for Egyptian) offer a different picture. The phoneme inloanwords and correspondences is often realized asl or d (in Hausa and otherChadic languages also as implosived’).

The situation in Gec ez shows that in graphical representation the phonemewas differentiated from˙s. RODINSON1981, based on further data, speaks about alateral character of this phoneme, i.e. not about emphasis understood as pharyn-gealization or glottalization. A similar situation is found in Arabic dialects (GAIR-DNER 1925:20, RABIN 1952:33, CANTINEAU 1960:54ff., 284ff.; D. COHEN

1963:11, COLIN 1930:92, 101-4 etc.), in the articulation of some Coranic recitersand in Modern South Arabian languages (Mehri, Š˙h eri, Soqotri - BROCKELMANN

1908:132), where the phoneme is realized asl. Besides these data, there is also acorrespondence withld - loanwords from Arabic in Spanish -a ˙d- ˙dayca> aldea, al-q a˙d ı> alcalde(RODINSON1981:103), in Maltese withl - ri ˙d a> rela, rama˙d an>ramelanetc. (COLIN 1930:101). Today the lateral character of˙d is rarely doubted,though sometimes the phoneme is characterized slightly differently - as cacuminal(MARRASSINI 1976:337), and generally the˙d as an emphatic (i.e. pharyngealizedor glottalized) is not included into the Semitic protosystem. Cf. also HODGE

1983:149 in the Hamito-Semitic (Afroasiatic) context.Nevertheless it is not possible to exclude definitely the connection of˙d and

the sibilant series. This idea has appeared already in VILENCIK 1930 and BLAKE

1946, and the combinatorics (incompatibility) also show that this phoneme preser-ves almost completely the incompatibility with sibilants and not withl, as mightbe supposed on the basis of its possible lateral character. There are also a numberof doublets of this phoneme with laterals that are mentioned in KURYŁOWICZ

1973:28, STEINER 1977 (passim) and FISCHER1968:59-60. One of the best ana-

Page 18: ZemanekPharyngealization

12 Petr ZEMÁNEK

lyses known to us is VOIGT 1991, where the phoneme is included to the protosy-stem as lateralized sibilants. From such a position even the most peculiar corre-spondences can be explained.26 That is why our position is that˙d should beclassified in the Semitic protosystem as a lateralized member of the sibilant series.

1.3.4 Emphatic˙sIn Arabic, from the first attestations by the Arabic grammarians (SIBAWAIH ,

˘HAL IL - cf. ROMAN 1977, etc.) this sibilant is put among the emphatic sounds, i.e.pharyngealized (the so-calledmu˙tbaqa) series. It is one of the phonemes thatappear as pharyngealized in all the Arabic dialects in the role of "emphatisant".27

It is attested in all the Semitic languages with the exception of Neo-Assyrianand some Arabic dialects that have lost the emphatic series (Malta - D. COHEN

1970, Chadic Arabic - ZELTNER & TOURNEUX 1986, ROTH 1979, Arabic pidginsin Sudan, Kenya and Nigeria - e.g. OWENS1985:233-4, HEINE 1982 etc.). In olderworks the supposed articulation is pharyngealized, in recent studies other solutionsare offered (cf. VOIGT 1986, STEINER 1982, CLAASSEN 1971, CARDONA 1968etc.). In other languages, where we have the possibility to see the phonetic realiza-tion, ˙s is realized in a different way. In Modern Hebrew it is realized as affricate(c) and this usage prevails today even in liturgical praxis with Biblical Hebrew. InModern Ethiopian languages it is realized asc, in G ec ez it is usually characterizedas emphatic. Also in Modern South Arabian we can see a tendency towards anaffricated realization of this sound (both articulations, i.e. pharyngealized andaffricated, are considered equal - MILITAREV , personal communication). Outsidethe Semitic languages we find one indirect indication for an affricated realizationof this sound - in Old Egyptian it corresponds with the phoneme¯d (i.e. g in Semi-tistic transcription) - cf. CONTI 1976, which might reflect a correspondence of anonvoiced (c) and voiced (g) affricate. Cf. also GARBINI 1972 and 1984.

Beside this manner of articulation, in some works it is possible to find at-tempts to characterize the original articulation of this phoneme as glottalized (s’ -cf. ROMAN 1990 etc.). ROMAN 1981-2 characterizes the sound as "pseudo-glottalized", and the position of VOIGT 1986 is similar.

26 Cf. the correspondences withq in Aramaic (Arabic’ar ˙d-un - Aramaic ’arq a - the earth)and correspondences with˙s in a number of languages, e.g. Hebrew’ere ˙s - Arabic ’ar ˙d-un -the earth.27 For terminology, cf. COHEN 1969.

Page 19: ZemanekPharyngealization

13Characteristics of emphasis

Interesting data is found in Yemen, where this phoneme in some dialects canbe realized asst (BEHNSTEDT1986:5-9). A similar situation can be found also inloanwords in Arabic - cf. lat.castra> qa ˙sr - D. COHEN 1962:120.

As far as its position in the protosystem is concerned, it seems that it shouldbe classified as a glottalized affricate or fricative.28

1.3.5 Emphatic˙zThe articulation of this phoneme in Classical Arabic places the phoneme

clearly among the pharyngealized emphatic consonants (cf. SIBAWAIH ). It isarticulated as voiced pharyngealized alveolar spirant or interdental (¯d - AL-ANI

1970:48 etc.). The Arabic pharyngealized explosive˙d often merges with thisphoneme in the dialects of Arabic, e.g. in Iraq, Tunisia etc. In Classical Arabic, itsfunctional load is minimal (cf. HERDAN 1962:54 - only 0,6%) to such a degree thatfor the majority of the roots with˙z it is hard to find minimal correlation pairs.

In other Semitic languages, this phoneme is attested only scarcely, MOSCATI

ET AL. 1964:43-5 mentions it only for Arabic and Epigraphic South Arabian.29

The grapheme transcribed as˙z in Ugaritic represents an unvoiced phoneme (cf.MOSCATI ET AL. 1964:43, VOIGT 1991 etc.).

Outside the Semitic languages this phoneme can be traced only in Berber (cf.WILLMS 1972 and WOLFF 1981:177), where it is possible to expect the influenceof Arabic. In every other family of the Hamito-Semitic languages this phoneme isnot attested.

Sometimes this phoneme has been postulated for the Proto-Semitic phonolo-gical system - cf., e.g., MOSCATI ET AL. 1964:24. On the other hand some authorsspeak about this phoneme as an innovation in the Arabic system. Cf. e.g. GARBINI

1972:42 (footnote 1), PETRÁCEK 1981:165 etc., who speak of an innovation in theAmoreic and especially the Arabic area (cf. also LOPRIENO1977:135).

1.3.6 Other emphatic phonemesThe five above mentioned phonemes occur in Classical Arabic (and also

MSA). Nevertheless the number of emphatic phonemes in the dialects of Arabicis much higher. Especially in the "central" Arabic dialects30 the correlation of

28 Note that according to WEDEKIND 1990:128 it is not necessary to distinguish betweenfricative or affricate character of this type of ejectives, since languages tend not to use thisopposition for phonemic contrast and the two types can easily be variants of one phoneme.29 Sayhadic in BEESTON’s terminology - BEESTON1981.30 In the so-called peripheral Arabic dialects, the emphasis is mostly lost.

Page 20: ZemanekPharyngealization

14 Petr ZEMÁNEK

emphasis developed considerably. There are dialects where we can find an empha-tic variant of every phoneme (e.g. in Mauritania - cf. ZAVADOVSKY 1981, Mor-rocco - cf. HARRELL 1965, Jordan - cf. BANI YASIN & OWENS 1987, etc.). Instudies on such dialects we meet the idea of conceiving the emphasis as a supra-segmental phoneme.

The occurrence of additional emphatic phonemes in other Semitic languagesis rather rare. MOSCATI ET AL. 1964:43-5 does not mention any other emphaticphonemes, SWIGGERS1981 adduces emphaticˇ˙s for Mehri and Š˙h eri. Within theframework of glottalized phonemes, it is possible to find glottalizedpandc (Neo-Assyrian - TSERETELI1978:34-8, Gec ez - e.g. VOIGT 1989:634 etc.).

As far as the Hamito-Semitic languages are concerned, several other empha-tic phonemes are postulated. The largest number can be found in the reconstruc-tions of DIAKONOFF (1988, 1984, 1987 etc.) -p, ˙t, ˙c, ˇ˙c, ˙k, qh, ˙kw. In actual systemsof the Hamito-Semitic languages we find pharyngealized phonemes in Berber,non-voiced ejectives in Cushitic and Omotic and implosive sounds in Chadic (b,d, y). Generally the character of the original emphatic phonemes is postulated asnon-voiced, only in Chadic do we find voiced implosives and in Berber voicedpharyngealized consonants.

The situation in Arabic is rather different. Already in Literary Arabic (MSA)we find certain tendencies towards the expansion of the pharyngealized type ofarticulation (emphatic˙l - cf. PETRÁCEK 1952, FERGUSON1956, etc.), also thepossibility of velarization/pharyngealization ofr (> ˙r) in Classical Arabic - cf.SCHAADE 1911:14-16. In the Arabic dialects, the growth of this correlation beco-mes considerably dynamic. From the point of view of emphasis we can divide theArabic dialects into two groups - to the peripheral dialects, where the emphasisdoes not grow or decreases - dialects in Malta (D. COHEN 1970), in Central Asia(G. TSERETELI 1956), in Africa (ROTH-LALY 1972, HEINE 1982, OWENS 1985etc.), the second group is formed by the central dialects, where the pharyn-gealization increases very dynamically. Here, in fact, are all the Arabic-speakingcountries. It seems that in the Arabic Magrib (especially in Morocco and Maureta-nia) the dynamics of emphasis is so high that it covers the whole phonologicalsystem - cf. especially ZAVADOVSKY 1981.

Even more "emphatic" sounds can be found in Neo-Assyrian, where theexpansion of "emphasis" is the highest of all the Semitic languages (cf. TSERETELI

1982, HOBERMAN 1988 and 1989 etc.). Emphasis is nevertheless treated diffe-rently in those studies: it is not connected with the individual emphatic phonemes,but is considered as a so-called suprasegmental feature, or, as in HOBERMAN 1988and 1989 in the framework of so-called autosegmental phonology.

Page 21: ZemanekPharyngealization

15Characteristics of emphasis

Nevertheless, from the articulatory point of view, under so-called "emphasis"we can find very different types of articulation. Apart from emphatics from Classi-cal Arabic, which are pharyngealized, we encounter labialized (

˙b, etc.), and stro-

ngly nasalized (˙m, ˙n - MALAIKA 1959) consonants.It is clear that the number of "emphatic" consonants in the modern dialects of

Arabic is growing, on the other hand, it is also clear that not all of them can becovered by pharyngealization. It seems that in this case use of a more general termlike co-articulation or secondary articulation would be more suitable.

Page 22: ZemanekPharyngealization

16 Petr ZEMÁNEK

2. The relation between pharyngealization and glottalization

2.0 In all reconstructions of the Hamito-Semitic phonological system the conne-ction between glottalization and pharyngealization is widely accepted. Both typesof articulation are taken as two realizations of the co-articulation (Nebenarbeits-korrelation, cf. TRUBETZKOY 1939, PETRÁCEK 1988:32-4, 1990:284-290).

As has been observed several times, the emphatic phonemes in Semitic aregenerally realized by two basic ways of articulation, i.e. as pharyngealized andglottalized.31 The problem of which of the two articulations was the original oneappeared relatively early, and during the discussion the question has been solvedin different ways,32 however, recently the opinion that the glottalization is theoriginal way of articulation in proto-Semitic is gaining ground. This can be seene.g. in recent reconstructions of the proto-Semitic and proto-Hamito-Semitic(Afroasiatic) phonological system,33 where the emphatics are reconstructed asglottalized and voiceless.

The relation between glottalization and pharyngealization is confirmed alsoin studies by FREWOLDU (1984-6, 1988), which show a certain auditory similaritybetween glottalization and pharyngealization. In this study the author i.a. showsthat Tigrinya emmigrants in Sudan, when speaking the Sudanese dialect of Arabic,often realize the Sudanese pharyngealized sounds as glottalized (ejective). Thenative speakers of the dialect then usually accepted the ejective articulation of theemphatic ˙t as a normal variant, while at other emphatics they pointed out theauditory differences and described it as foreigners’ speech. The Tigrinya speakersusually realized the difference in the articulation of˙d, ˙sand ˙z, but usually did notnotice the difference between ejectivet’ and pharyngealized˙t.

The similarity between glottalization and pharyngealization is reflected alsoin the situation in contemporary South Arabian languages (Mehri, Š˙h eri, Soqotri),where the glottalized articulation of pharyngealized sounds is perceived as a validvariant (cf. SWIGGERS1981, NAUMKIN & PORKHOMOVSKY 1981:9).

31 Of course, other types of articulation can be effective, too, especially in the broaderconcept of emphasis, as it is generally applied in the case of Arabic dialects, where one canobserve types of coarticulation like labialization (emphatic%b, ˙m), (strong) nasalization(emphatic ˙n), etc.32 Cf., e.g., LESLAU 1957:325 in favour of the pharyngealization as proto-Semitic, KNUDSEN1969 for glottalization as original. For an attempt at an alternative solution (emphasis =uvularization) cf. DOLGOPOLSKY 1977. Naturally, a number of other studies have appeared.33 Cf., e.g., ROMAN 1981 and BOMHARD 1988 for Proto-Semitic, DIAKONOFF 1984, 1987,BOMHARD 1984 for Proto-Hamito-Semitic, etc.

Page 23: ZemanekPharyngealization

17Pharyngealization and Glottalization

These data clearly show the relation between pharyngealization and glotta-lization, but do not solve the problem which predated the other. We, however,accept the arguments that glottalization was the original one, that have beenarticulated already in MOSCATI ET AL. 1964:23-4,34 and that pharyngealizationappears as secondary from some of originally glottalized consonants, especiallyglottalized t’ and k’. Then, the pharyngealized˙t forms the core of the newlyformed series, whileq, due to articulatory reasons,35 is not pharyngealized, butshifted to the back and is not accompanied by coarticulation. Even in SIBAWAIH

(1889:302ff)q is not ranked together with other emphatics (mu˙tbaqa- ˙t, ˙d, ˙s, ˙z),but with a broader set ofmustacliya.36

The secondary character of pharyngealization is, according to our view, sup-ported also by the further development of pharyngealization in Arabic, especiallyin its dialects. Arabic is then probably the originator and without any doubt thepropagator of pharyngealization.37

A similar approach, i.e. the concept of secondary character of pharyngealiza-tion, can be observed also in studies on other Semitic languages. In recent studieswe find clear statements that the correlation of emphatic phonemes that we knowfrom Arabic is of heterogeneous origin and moreover, that the pharyngealizationis secondary (cf. GARBINI 1972:141, note 1, similarly also 1984; DOLGOPOLSKY1977, ARO 1977, ROMAN 1981, etc.). The concept of this correlation (as well asthe concept of the whole proto-Semitic phonological system) is dynamic andpharyngealization is regarded as an innovation.

In most of the branches of the Hamito-Semitic languages, this correlation isalso represented, and only Egyptian forms an exception.38 In Chadic we find

34 "a) the Ethiopic "emphatics" are voiceless and, apart from Arabic, so the Semitic "empha-tics"−almost without exception; b) the Ethiopic "emphatics" do not appear to influence thetimbre of neighbouring vowels and, again apart from Arabic, this seems to be the norm inthe Semitic languages (cf. however for certain facts in Akkadian SODEN, 1952, p. 12); c) thephenomenonq > ’ in some Arabic dialects can only be explained by way of glottalization."(Ethiopic "emphatics" stand for ejective consonants).35 The backing articulation, when applied to back consonants, has a different manifestationthan when applied to dentals and alveolars. Cf. DELATTRE 1971.36 Mustacliya = mu˙tbaqa+ q,

˘h.

37 Cf. the spread of pharyngealization in modern dialects of Arabic. The pharyngealizedconsonants in Berber are usually ascribed to Arabic influence, too (APPLEGATE1970).38 Emphatic phonemes are nevertheless reconstructed on the proto-Egyptian level, cf., e.g.,RÖSSLER1971 in his systematic reconstruction; CONTI 1976 reconstructs only emphatic˙t.Sometimes, Egyptian grapheme for voiced affricateg (= ¯d in egyptological transliteration)is interpreted as corresponding to Semitic˙s.

Page 24: ZemanekPharyngealization

18 Petr ZEMÁNEK

implosives (b’, d’, y’ - NEWMAN 1977).39 Cushitic offers a number of post-glotta-lized ejectives and implosives, with maximum of 11 emphatics in DOLGO-POLSKY’S (1973) reconstruction. Berber has pharyngealized consonants, cf., e.g.,the reconstruction of K. PRASSE(1972), where we find˙t, ˙d, and ˙z. Generally, thepharyngealized articulation of Berber emphatics is ascribed to the influence ofArabic.

Glottalization is thus the quantitatively prevailing type of articulation, andaccording to its behaviour in loanwords, it seems that the emphatic consonantswere non-voiced (cf. DIAKONOFF 1991-92:59). The Hamito-Semitic evidenceoffers also an indication of the possible development from glottalization to pha-ryngealization. MERLINGEN (1978) speaks of Chadic implosivesb’ and d’ assounds characterized by "laryngealized voicing" or "creaky voice". This mightpoint to a certain degree of articulatorily similarity and the direction of successionof the two ways of articulation.

The transition from glottalization to pharyngealization must have been aprocess which had to leave some traces in the system. It is probable that after therelease of the glottal occlusion applied in the articulation of the glottalized con-sonants, the way for the participation of voice opens (postglottalized ejectives areattested only as non-voiced), and such a possibility could result in the existence ofdoublets alternating in non-voiced emphatic and its voiced counterpart. Thesetypes of alternation were noticed already in BROCKELMANN 1908 (e.g. 160, 162for ˙s/z), MAIZEL ’ 1983, cf. also PETRÁCEK 1987, 1990:288; for "emphatic"q cf.CANTINEAU 1960:182, CONTI 1976, ROMAN 1981 etc. According to us, in themoment of substitution of glottalized articulation by the pharyngealized one, partof the roots with the glottalizedt’ andk’,40 and, with some reservation, possiblyalso the roots with emphatic˙s41 change after the release of the glottal occlusionto roots with the voiced counterpart of the originally glottalized consonants.

39 JUNGRAITHMAYR & SHIMIZU 1981 reconstruct the following "emphatics" (i.e. implosivesand ejectives):b’, d’, y’, k’, s’ ; STOLBOVA (1986), obviously in a maximalist reconstruction,reconstructs the emphatic series in the following way:p, ˙t, c, c (c), k, kw, q. It should be saidthat "emphatic" in the Russian concept corresponds rather to the "gepreßt" from TRUBETZ-KOY’S (1939)gepreßt x nicht gepreßtopposition.40 The situation at the shift from glottalizedk’ to uvularq is in Arabic obscured by the fact,that the voiced counterpart which is reconstructed in the protosystem is changed to itspalatalized variant (g > g).41 The question of the original articulation is not clear, the discussion has not been concludedand the possibility remains of original affricated articulation, which would set this consonantsomewhat apart from our speculations.

Page 25: ZemanekPharyngealization

19Pharyngealization and Glottalization

Such a type of alternation is attested also from other Semitic languages,where e.g. in Old Babylonian graphemic representation, the signs withd are usedfor words with etymological˙t (cf. MOSCATI ET AL. 1964:31; SODEN 1952:§29,§26b.42

For the test of our hypothesis we used the method, which is in Semitisticsrelatively common, i.e. the collecting of the so-called doublets. This method isbased on the character of the root in Semitic languages (or in Arabic, to be moreprecise), which is formed in the vast majority by 3 consonants that serve as basisfor the morphological derivations and that bear some "basic meaning". Doubletsthen are such types of pairs of roots that differ in only one of the three radicals inthe specified position, i.e. for alternation˙t/d an example for a doublet would bee.g.bl ˙t/bld.

This method has been used e.g. in the discussion about the original characterof the root in Semitic, namely as the supporting argument for the biradical chara-cter of the Semitic root, as shown e.g. by the so-called PR3 series.43 For a radicalexplanation of this phenomenon cf. EHRET 1989.44 Furthermore, the method isrelatively often used for showing the original character of a phoneme, cf., e.g.,STEINER 1977 passim for doublets likes/l, ˙d/s, ˙d/l, etc.

On the other hand, it has to be said that this method cannot be considered areliable one. It is, beside the formal differences of one element in the root, basedon semantic similarity that can be influenced very often by subjective criteria ofthe researcher, and often these criteria can be hardly defined by some rules, sincethe existence of doublets can be caused not only by historic relations between thetwo roots, but also by the subsequent phonetic and phonological processes, phone-tic similarity, etc.

42 VON SODEN 1952: §29b says: "In der Schrift werdend, t und ˙t aA (= Altassyrisch)nirgends, aAK (= Altakkadisch) fast nie unterschieden ...". Cf. also GELB 1961:29 and thefollowing discussion. These problems were obviously caused by the character of the cunei-form script which did not represent the emphatic phonemes. These were in Akkadiansubstituted by signs for auditory similar sounds. Nevertheless, in Greek loanwords we findnon-voicedt for emphatic ˙t (DIAKONOFF 1991-92:59).43 Cf., e.g., MOSCATI ET AL. 1964:72-73, where there are given examples like Hebrewprd"to separate",prm "to tear",prs "to split", pr ˙s "to break down" etc., which are interpreted asa biconsonantal root basepr with the basic notion "to divide". For critique, cf. VOIGT 1988,for an alternative explanation of this phenomenon, cf. PETRÁCEK 1987.44 This study is an attempt to assign certain types of meanings to the root extensions. Suchan explanation goes probably too far, when one realizes that a number of such alternationscan be caused by phonetic reasons. Another problem is that EHRET concentrates on R3 only,while Semitic offers data on the alternation of both R1 and R2 as well.

Page 26: ZemanekPharyngealization

20 Petr ZEMÁNEK

It is exactly these types of processes, i.e. phonetic and phonological, that arevery frequent in Semitic languages and especially in Arabic. A good overview ofthe richness of these processes in Semitic languages is given in MAIZEL ’ 1983which deals with the processes of the expansion of the roots in Semitic, with themain processes allothesis, metathesis and chaining (of the type PR3 series, cf.above).

Relatively frequent are also alternations based on clearly phonetic reasons,e.g. the alternation of semivowels (w, y), laryngeals (c/ ˙h, h/ ˙h, c/’ etc.), alternationslike d/ ¯d/s, etc.

These facts lead us to the conclusion that the use of this method without someaccompanying correction procedures is not sufficiently valid. With the assumptionthat it is difficult to avoid the subjectivity when collecting the data, we havedecided to use the method together with a control set of data that were collected inthe same way as the sets that are at the centre of our attention. The control sets thatwe use are of a twofold character: the first consists of doublets, where there is aclear phonetic relation in the alternation, i.e. doublets with alternations˙t/t, q/kand

˙s/s, and the second that contains doublets of phonetically clearly non-relatedsounds. For the choice of the second set clearly quantitative criteria were used. Wehave chosen the consonants according to their relative frequency as given inHERDAN 1962:54, and the resulting pairs werer/b (occurrence frequencies 7,7%/ 5,7%) and ¯t/g (occurrence frequencies 1,8% / 1,9%). Such a choice offers,despite the subjectivity of the input procedure the possibility to compare theresults of the data collection, since it offers information on relative frequencies ofthe doublets. It should also allow for setting up a more contoured picture of thesituation and processes that took place in Arabic.

2.1 Analysis of the alternations of non-voiced emphatics with their voiced coun-terparts2.1.1 Emphatic˙t and its alternation with voicedd

The emphatic˙t has the central position in our hypothesis, according to whichthis phoneme represents the connecting bridge between the older type of arti-culation of emphasis - glottalization - and the newer type - pharyngealization.45

The doublets of˙t/d we consider as the residue of the period when the glottalizedt’ was changing into pharyngealized˙t. After the loss of the glottal stop a way

45 The other phoneme could beq, but this consonant does not fit exactly to the model ofdevelopment from glottalization to pharyngealization. Its articulation in Arabic cannot becharacterized as pharyngealized, but uvularized.

Page 27: ZemanekPharyngealization

21Pharyngealization and Glottalization

opened for the participation of voice in the articulation and part of the roots witht’ changed into roots with voicedd, and other part changed to roots with pharyn-gealized ˙t, with overlapping of the two sets. It is exactly this overlapping thatdocuments this change. Because of reasons given above the type of articulationthat emerged at˙t, the correlation of emphatic pharyngealized consonants is cre-ated.

Then, in the part of the doublets where thet’ changed tod, originally as anallophone, thed received gradually phonematic character. The voiced explosived thus has two origins - one is the old phoneme (*d as opposed to*t etc.), and theother is the original variant of glottalized/pharyngealized˙t.

In the dictionary of A.B. KAZIMIRSKI (1860) we find a relatively large num-ber of these doublets (cf. Table I):

Table I: Alternation of ˙t/d in Arabic

Alternation ˙t/don the position

R1 R2 R3 Total

Number of dou-blets

40 37 32 109

total number ofroots in the dic-tionary

d ˙t

274 212

d ˙t

249 200

d ˙t

259 214

d ˙t

782 626

% 14,59 18,8614,85 18,5 12,35 14,9513,93 17,41

Such a frequency, especially the frequency of doublets with the emphatic˙t,is clearly very high. It is hard to imagine that such a frequency could be a result ofpurely random similarity, although the phonetic similarity in this case plays aconsiderable role.

The amount of the onomatopoetic (or, somewhat more widely, descriptive)phonemes in the corpus is relatively low and is almost exclusively restricted to thegroup of ˙t/d = R1 (R1: T’T’; Tqq; Tnn; TnTn;R2: hThT; R3: 0). Normally, thesetypes of roots might influence the results of the analysis, since the semanticsimilarity can be caused by the character of these words, i.e. the phonetic imitationof some sound or process. Here, in this case, considering to the very limitednumber of these phonemes (less than 2% in the highest possibility), we can suppo-se that their occurrence will not influence the analysis.

Table II shows the phonetic distribution where the alternation˙t/d is preserved(only the consonants in the contact position are given; this means in the case of R2

Page 28: ZemanekPharyngealization

22 Petr ZEMÁNEK

both contact positions before and after the alternation is given - the third numberis the total of the preceding two):

Table II: Phonetic environment of the alternation˙t/d in Arabic:a) voiced consonants:

Alternations of ˙t/don the position:

R1 R2 R3 Total

consonants:brzc

glmnh

w/y

(R2)3202223454

(R1) (R3) (∑)5 2 74 2 60 0 02 3 51 2 33 1 44 5 91 2 32 1 31 7 8

(R2)2512152225

1212196111391017

Total: 27 48 27 100

% 67,5 67,6 75,75 69,45

b) non-voiced consonants:

Alternation of ˙t/don the position:

R1 R2 R3 Total

consonants:’

¯t

˙h

˘hsšfq

(R2)10523011

(R1) (R3) (∑)0 2 20 0 02 3 53 1 44 1 52 0 21 1 23 0 3

(R2)12210101

421278355

Total 13 23 8 44

% 32,5 32,39 25,8 30,55

Page 29: ZemanekPharyngealization

23Pharyngealization and Glottalization

The table clearly shows a great dominance of the voiced consonants in thecontact positions of the alternation. It is remarkable that the occurrence of thevoiced consonants is proportional, in most of the cases the number of occurrencesis about 10 (9-11). Onlymand especiallyw/yoccur more frequently. The voicedconsonants are mostly from the group of the so-called "glides" or the transitoryzone [(C) - LIQ - SON - LAR - SV - (V) - cf. K. PETRÁCEK 1971]. Onlyb, gandzdo not enter this zone.46

The group of non-voiced consonants is clearly much smaller than the groupof voiced ones. Also the distribution of these phonemes is much less proportionalthan is the case at the first group. Half of them does not occur at all the subgroups(according to the position of alternation, i.e. R1, R2 and R3). The phoneme with thehighest occurrence (˙h) comes also from the group of the "glides".

It is evident that the character of the neighbouring phonemes played a sub-stantial role in preservation of (and, possibly, to a certain degree in the causing)the alternation of˙t/d, and served also as the catalyst of the transition. It is probablythe voicedness that helped in the preservation of this alternation, and moreover, inthe contact position, there are consonants from the zone of the transition betweenconsonants and vowels (cf. PETRÁCEK 1971), that exhibit a high degree of voiced-ness (liquids, sonants, semivowels).47

A certain number of doublets in this set are interrelated phonetically. Namely,they are the following: at R1: 17.Tcs- 18.Tcz, 29.Tnn- 30.TnTn; at R2: 31.mTT-33.mTw/y; at R3: 2.blT - 24. lbT, 10.zrT - 11.zlT. The alternations in these roots(beside the alternation of˙t/d) can be explained by phonetical changes.

Morphologically most of the doublets preserve the alternation in its verbalmanifestation; nominal patterns are rather rare.

Semantic analysis could also show some interesting facts. That is why wetried to divide the doublets into groups that are evident in the data. The presentanalysis is based only on our own division, and is to be considered tentative.

Although the semantic division here is rather speculative, we consider ituseful to mention it here, since it can bring interesting facts to the environmentwhere the alternation is preserved. This holds also for the semantic division of thedoublets with the alternation ofq/gand ˙s/z.

The Table III shows the semantic division of the doublets with the alternationof ˙t/d:

46 According to RUŽICKA in a number of studies (e.g., 1954) and PETRÁCEK 1955, 1975, it ispossible to countg among the glides due to a supposed transition ofc > g.47 For the question of the degree of sonority, cf. GVOZDANOVIC 1985:89.

Page 30: ZemanekPharyngealization

24 Petr ZEMÁNEK

Table III: Semantic division of the doublets with˙t/d in Arabic.

Alternation of ˙t/d on the position: R1 R2 R3 Total

violence, aggressivity 4 13 9 26

actions connected with movement 10 6 7 23

actions and activities connected withspace

2 3 4 9

spiritual and corporal qualities 10 4 3 17

taboo actions and activities 2 4 3 9

production of sounds 5 1 0 6

rest 7 6 6 19

Total 40 37 32 109

It is evident that most of the features that we have managed to distinguish inthe data are somehow connected with some emotional charge (aggressivity,violence, etc.), spiritual and corporal qualities (mostly pejorative) or with dyna-mics (movement, etc.). These three groups form almost 60% of the whole. Emo-tional charge can without doubt be seen also in the taboo actions and activities (8doublets).

Thus, with 53 doublets we can speak of emotional charge, with 23 doubletsof dynamics (especially movement; a certain dynamics could be also ascribed tosome of the doublets that are here gathered under the "rest" label). This dynamicscan be to a high degree strengthened by the fact that most of the doublets areverbs.

We assume that it is also this dynamic and emotional charge in the semanticsof the doublets that plays its role in preserving the remnants of the transition oft’> ˙t/d > ˙t, d.

Nevertheless, in our tentative semantic analysis we did not find any regularsemantic change due to the variant used. That is why we consider this alternationas caused mainly by phonetic features, mostly by the voiced environment of thealternation.

2.1.2 Uvularq and its alternation with voicedg:As has been said above when describing the uvularq, this phoneme can

hardly be associated with the other "emphatic" phonemes of Arabic, both from the

Page 31: ZemanekPharyngealization

25Pharyngealization and Glottalization

phonetic and phonematic point of view. This phoneme enters our analysis only onhistorical grounds.

There are many reasons why we should consider this phoneme as originallyglottalized, one of them is the alternation with the glottal stop’ in a number ofArabic dialects (e.g. in Egypt and elsewhere). Here, we would like to add anotherargument.

Based on the hypothesis formulated above, we suppose also with this phone-me the transitionk’ > q , i.e. a change in the manner of articulation. After therelease of the glottal closure, there is a way opened for the participation of voice,and thus for the transitionk’ > g . In Arabic, this transition is somewhat obscuredby the fact that the originalg was palatalized tog (we assume this transitionaccording to thecorrespondences among theSemitic languageswithout examiningthe reasons for this process). In spite of this, as our data show, the traces of thetransitionk’ > q in Arabic are preserved in a number of doublets.

In the dictionary of A.B. KAZIMIRSKI (1860) we find the following number ofdoublets ofq/g:

Table IV: Alternation ofq/g in Arabic:

Alternationq/gon the position

R1 R2 R3 Total

Number of doublets 50 21 36 107

total number of rootsin the dictionary

g q

357 413

g q

205 248

g q

322 363

g q

884 1024

% 14,00 12,1 9,46 8,43 11,18 9,91 12,10 10,45

As opposed to the alternation of˙t/d, the relative (i.e. percentual) number ofthe doublets with the alternation ofq/g is smaller. We suppose that this relativelysmaller number of doublets is caused by the related palatalizationg > g, whichwithout doubt influenced the possibility of the preservation of the traces of thechange.

As onomatopoeic (or descriptive), the following roots can be determined:Qšš, ˙sQQ, šQQ,and hQhQ. Their number is so small that even here we canexclude any considerable influence of these words on our analysis.

The following table gives the phonetic environment of the alternation ofq/g,summarizing the consonants in all the contact positions of the alternation:

Page 32: ZemanekPharyngealization

26 Petr ZEMÁNEK

Table V: Phonetic environment of the alternation ofq/g in Arabic:a) voiced consonants:

Alternations ofq/g in the posi-tion:

R1 R2 R3 Total

consonants:bd

¯drzc

lmnh

w/y

(R2)24390254322

(R1) (R3) (∑)0 1 11 0 10 0 01 2 30 0 01 3 40 2 21 2 31 1 24 1 53 1 4

(R2)52021262134

874141813961010

Total: 36 36 28 90

% 72,00 69,23 84,85 72,00

b) non-voiced consonants:

Alternation ofq/g in the posi-tion:

R1 R2 R3 Total

consonants:’

¯t˙h

˙sf0

(R2)1214105-

(R1) (R3) (∑)0 0 01 0 12 0 21 0 13 0 31 0 12 3 5- 3 3

(R2)0020102-

135551123

Total 14 16 5 35

% 28,00 30,77 15,15 28,00

Page 33: ZemanekPharyngealization

27Pharyngealization and Glottalization

Just as with the preceding group of doublets (˙t/d) the phonetic environmentof the alternation is formed mostly by the voiced consonants. Most frequent areliquids and semivowels, the role of laryngeals, esp.h, is also substantial. Into thegroup with higher occurrence come the soundsr, l, h, w/y(with occurrence over10), while at the other extreme standszwith one occurrence only.

Non-voiced consonants in the environment of the alternation are much lessfrequent than voiced ones. It is interesting that here the phonemef is very promi-nent. It occurs more than twice as often as the other most frequent phonemes (˙h,s, š). On the other hand, in the case of this alternation, the occurrence of one"emphatic" phoneme (˙s) is also attested here.48

In this group also occur roots of the type R1R2R2 which are here under thenon-voiced consonants where the third consonant is labelled with 0 in the table.It is interesting that with the preceding group this type of root was of very lowoccurrence.

Again it is possible to say that it is the voiced environment, especiallysounds with high degree of voicedness that are close to vowels (liquids, semi-vowels, laryngeals) which served as the catalyst for the transition and also wheremost traces of the transition are attested.

As it has been the case at the˙t/dalternation, here also we can find some dou-blets whose interrelations can be explained by phonetic reasons. Namely they arethe following: at the alternation on the position of R1: 7.Qdd- 11.Q ¯d ¯d; 19.Qrf -37.Qlf; 30.Qff - 32.QfQf; at the position of R3: 6. ˙hbQ- 7.

˘hbQ; 10.drQ - 12.

rdQ; 28. frQ - 29. flQ.

Morphologically also the verbs are dominant and nominal (mostly adjecti-val) meanings are not so frequent.

The tentative semantic analysis is given in the following table.

48 Root ˙sQQ; it is especially the combination of these two consonants that is very commonin Arabic.

Page 34: ZemanekPharyngealization

28 Petr ZEMÁNEK

Table VI: Semantic division of the doublets ofq/g in Arabic:

Alternation ofq/g in the position: R1 R2 R3 Total

violence, aggressivity 16 9 8 33

actions connected with movement 2 3 6 11

actions and activities connectedwith space

0 2 2 4

spiritual and corporal qualities 7 2 5 14

taboo actions and activities 0 1 4 5

production of sounds 0 1 2 3

dryness 4 0 0 4

"draw in, remove" 9 0 0 9

rest 12 3 9 24

total 50 21 36 107

The table shows, as was the case of the semantic analysis of the alternation of

˙t/d, the dominance of meanings of emotional charge and dynamics (violence,taboo actions and spiritual and corporal qualities - 52 doublets, movement - 20doublets).

Unlike the alternation˙t/d, however, the meanings in the individual groups ofdoublets (i.e. R1, R2 and R3) are not distributed so proportionally. Here we havetwo semantic features that occur only with one group (dryness, "draw in, remove"only at R1). The dominant meaning is violence, in the second position standspiritual and corporal qualities.

Again, it seems evident that the emotional charge and dynamics helped in thepreservation of the traces of the transition ofk’ > q and it is possible to assumethat to a certain degree it served as the catalyst for the process. Nevertheless themost important role is without doubt played by the phonetic environment of thealternation.

Page 35: ZemanekPharyngealization

29Pharyngealization and Glottalization

2.1.3 Emphatic˙sand its alternation with voicedz.As we have already pointed out, the original manner of articulation of this pho-

neme is still a question to be solved. It is not clear whether it was originally an affri-cate which might be indicated by the situation in Modern Hebrew and also ModernEthiopic Semitic (cf. also an indirect evidence - correspondence with the Egyptianphonemeg,49 or, as others assume it was an originally glottalized phoneme, as inthe case of˙t. Alternatively, it could be a glottalized affricate, which is the solutionoffered (implicitly) by DIAKONOFF 1991-92:56ff.

Nevertheless, whatever side we choose, the fact remains that in the lexicon ofArabic we can find a number of doublets with the alternation of˙s/z, as the data givenhere show.

The following table gives the numbers of the doublets with the alternation of

˙s/zthat were found in A.B. KAZIMIRSKI 1860:

Table VII: Alternation of ˙s/zin Arabic.

Alternation ˙s/zin the position

R1 R2 R3 Total

Number of dou-blets

26 26 27 79

total number ofroots in the dictio-nary

z ˙s

280 229

z ˙s

195 151

z ˙s

238 184

z ˙s

713 564

% 9,28 11,35 13,33 17,2111,34 14,6711,08 14,01

Here, the relative frequency of the doublets, especially when compared to thedoublets of˙t/d, is considerably lower. Possibly the sibilant character or the differentoriginal manner of articulation might have played a role.

The number of doublets of onomatopoeic or descriptive character is very lowhere, in fact it is only one -SrSr. The influence of this type of roots on our analysisis then negligible.

The following tables show the phonetic environment in the contact positions ofthe alternation of˙s/z:

49 In Egyptological transliteration¯d.

Page 36: ZemanekPharyngealization

30 Petr ZEMÁNEK

Table VIII: Phonetic environment of the alternation˙s/zin Arabica) voiced consonants:

Alternations of

˙s/zin the posi-tion:

R1 R2 R3 Total

consonants:bdrc

glmnh

w/y

(R2)3343011016

(R1) (R3) (∑)3 4 70 2 22 2 43 0 30 0 03 2 52 3 52 1 33 0 31 1 2

(R2)1060141222

11514611075610

Total: 22 34 19 77

% 84,61 72,34 70,37 75,49

b) non-voiced consonants:

Alternation of

˙s/zin the posi-tion:

R1 R2 R3 Total

consonants:’

˙h

˘hšfqk

(R2)2000020

(R1) (R3) (∑)1 2 31 0 11 0 13 0 31 2 30 2 20 0 0

(R2)2120111

7233451

Total 4 13 8 25

% 15,39 27,66 29,63 24,51

Page 37: ZemanekPharyngealization

31Pharyngealization and Glottalization

Also in the case of alternation of "emphatic"˙swith its voiced counterpart -zwe observe a clear majority of the voiced consonants in the neighbourhood ofthe alternation, especially in case of R1. Also in the overall figures this alternationexhibits one of the highest rates of voiced consonants in its neighbourhood. Theoccurrence of voiced consonants here is not so "proportional" as is the case of thealternation of ˙t/d. We can divide the consonants into two groups according totheir frequency - a group with higher frequency (14-10 -b, r, l, w/y) and lowerfrequency (5-7 -d, c, m, n, h). Beside these two groups we findg with only oneoccurrence. Again, as it was the case with the previous alternations, the con-sonants belonging to the transitory zone (PETRÁCEK1971) can be found here withthe highest frequencies (i.e. liquids and semivowels). All of the consonants foundhere belong to the group with high sonority.

With the non-voiced consonants it is remarkable that the occurrence ofcompletely non-voiced contact is infrequent in the case of groups R1

50 and R3;but also the occurrence of non-voiced environment at the group of R2 is relativelysmall. It is also remarkable that the higher frequency can be found at’ andq,sounds that are not very stable and tend to alternate with other sounds.51 Besidethese two sounds with the highest frequency and with exception ofk with onlyone occurrence, the frequency here is proportional.

It seems that in case of this alternation the influence of the voiced environ-ment is even stronger than is the case of the alternation˙t/d andq/g. Such a typeof environment would point especially to the phonetic character of this alter-nation.

As far as the phonetically related pairs of doublets are concerned, in this setwe find the following ones: at R1 it is 1.S’b - 2.S’mand 12.Srm- 18.Slm; at R3

we find 11.rqS- 18.qlS.52

The following table offers the tentative semantic distribution of the doublets,divided according to the most frequent semantic groups:

50 In case of R1 it is only ’ andq that occur in the contact position!51 Cf. the situation in the dialects of Arabic -’ often changes to simply a vocalic reflex, andq can be realized in a number of ways - voiced or unvoiced.52 I.e. both metathesis and alternation of liquids -r/l .

Page 38: ZemanekPharyngealization

32 Petr ZEMÁNEK

Table IX: Semantic division of the doublets of˙s/zin Arabic:

Alternation of ˙s/zin the position: R1 R2 R3 Total

violence, aggressivity 11 6 10 27

actions connected with move-ment

0 3 7 10

actions and activities connectedwith space

2 4 0 6

spiritual and corporal qualities 1 3 2 6

taboo actions and activities 1 2 0 3

production of sounds 3 2 0 5

rest 8 6 8 22

Total 26 26 27 79

Similarly to the preceding two cases, we also observe here the dominance ofsemantic notions connected with emotionality and dynamics. The most promi-nent notions are quite clearly activities connected with violence and aggressivity(27 doublets), second is movement (10 doublets). These two notions togetherrepresent almost half of the doublets. Then come taboo activities and corporaland spiritual qualities that without doubt also bear an emotional charge.

Here too it is possible to say that it is the dynamics and the emotional chargethat represent the environment where the alternation of˙s/zis preserved.

2.2. Analysis of the control set: Phonetically conditioned doublets

This set is included here in order to achieve more plastical information aboutthe behaviour of the doublets with emphatic phonemes in Arabic. Here, five setsare analysed, three for the alternation of emphatic vs. non-voiced non-emphatic,i.e. the alternations of˙t/t, q/kand ˙s/s, and two sets of doublets with variation ofnon-emphatic and phonologically non-related consonants chosen according totheir frequency. The alternation ofb/r stands here for the group of consonants

Page 39: ZemanekPharyngealization

33Pharyngealization and Glottalization

with the highest frequencies, and the alternation of¯t/g for the consonants with thelowest frequencies. The frequency data are taken from HERDAN 1962:54, al-though these data may differ from the results obtained from KAZIMIRSKI 1860.The reason for the difference is mainly in the fact that HERDAN (based on GREEN-BERG1950) takes into account only verbal roots, while our data cover both verbaland nominal roots, while another reason for difference could be caused byGREENBERG’S use of other dictionaries for his analysis.

2.2.1 Emphatic˙t and its alternation with non-voicedt:

The number of doublets with this type of alternation, which is based clearlyon phonetic features, is relatively low, especially when compared to the numberof doublets with the alternation of˙t/d. The data are summarized in the followingtable:

Table X: Alternation of ˙t/t in Arabic.

Alternation ˙t/t onthe position

R1 R2 R3 Total

Number of doub-lets

10 22 14 46

Total number ofroots in the di-ctionary

t ˙t

151 212

t ˙t

238 200

t ˙t

189 214

t ˙t

578 626

% 6,62 4,71 9,24 11,00 7,4 6,54 7,96 7,34

In case of this alternation, only the alternation on the position of R2 givesnumbers little bit higher, but still in no way comparable to the numbers at the˙t/dalternation.

The following tables show the phonetic environment at the contact positionsof the alternation of˙t/d:

Page 40: ZemanekPharyngealization

34 Petr ZEMÁNEK

Table XI: Phonetic environment of the alternation˙t/t in Arabic:a) voiced consonants:

Alternation ˙t/t onthe position

R1 R2 R3 Total

consonants:bdrc

glmnh

w/y

(R2)1020030101

(R1) (R3) (∑)2 0 20 1 11 5 60 1 12 0 24 1 53 3 60 0 01 0 10 0 0

(R2)2011022003

51922108114

Total 8 24 11 43

% 80 54,76 78,57 64,18

b) non-voiced consonants

Alternation ˙t/t onthe position

R1 R2 R3 Total

consonants:’t/ ˙t

˙h

˘hsšfq

(R2)00010010

(R1) (R3) (∑)1 2 30 4 41 2 30 2 21 0 12 0 22 0 22 0 2

(R2)10100010

44431242

Total 2 19 3 24

% 20 45,23 21,42 35,82

Page 41: ZemanekPharyngealization

35Pharyngealization and Glottalization

Here, the dominance of the voiced consonants at the contact positions isclearly observable only at R1 and R3 doublets, while at R2 doublets the frequen-cies are almost equal. In the overall picture, the voiced consonants are stillprevailing at the contact position, with the highest frequencies of liquids (l, r) andnasalm. Here, for the case of de-emphatisation to a non-voiced counterpart, theprevalence of non-voiced neighborhood would be expected, which is obviouslynot the case. Nevertheless, similar relations were observed at all the alternationsof emphatic vs. non-voiced non-emphatic.

Phonetic similarity (beside˙t/t) can be observed only atmt ˙h / mt˘h.

Table XII shows the tentative semantic division of the doublets with thealternation of ˙t/t:

Table XII: Semantic division of the doublets of˙t/t in Arabic:

Alternation ˙t/t on the position R1 R2 R3 Total

violence, aggressivity 1 9 3 13

actions connected with move-ment

3 1 2 6

taboo actions and activities 0 2 0 2

spiritual and corporal qualities 1 1 2 4

production of sounds 0 0 2 2

rest 5 9 5 19

Total 10 22 14 46

The semantics of the doublets exhibits again quite a high percentage of rootswith dynamic and emotional meanings.

Altogether, this type of alternation exhibits similar characteristics as thealternation of˙t/d, except that the frequency of occurrence is considerably lowerthan it is the case with the˙t/d doublets.

2.2.2 Uvularq and its alternation with non-voicedk.

Page 42: ZemanekPharyngealization

36 Petr ZEMÁNEK

The data for this type of alternation, collected in KAZIMIRSKI ’S dictionary,are summarized in the following table:

Table XIII: Alternation ofq/k in Arabic.

Alternationq/kon theposition

R1 R2 R3 Total

Number of doublets 20 20 11 51

Total number of rootsin the dictionary

k q

323 413

k q

211 248

k q

201 363

k q

735 1024

% 5,88 4,6 9,47 8,06 5,47 3,03 6,93 4,98

As has been pointed out several times above, the position ofq in our analysisis, from the point of view of pharyngealized emphatics in Arabic, rather compli-cated. It is highly probable that it is linked with the Arabic emphatics historically(glottalization), but synchronically the situation is different. Also the reflects ofthe process of transition from glottalization to pharyngealization or to back arti-culation, which is the case ofq, are in Arabic obscured by the fact that the proto-Semiticg was palatalized in Arabic. These facts, i.e. the alternation caused bypurely phonetic similarity (q/k) and the alternation with some historical back-ground but the traces of which are obscured by the subsequent phonologicalprocess, could make the results of the sets ofq/g andq/k quite close. Never-theless, similar results were obtained only at R2 doublets, where the numbers ofdoublets are almost identical. In case of R1 and R3 doublets, the numbers (bothabsolute and relative) are much higher (more than twice in most cases) at thedoublets with the alternation ofq/g.

In this set, we found also one onomatopoeic root -krkr. Its presence, how-ever, will hardly exert any influence on the results of our analysis.

The following table shows the phonetic environment at the contact positionsof the alternation ofq/k in Arabic:

Table XIV: Phonetic environment of the alternationq/k in Arabic:a) voiced consonants:

Page 43: ZemanekPharyngealization

37Pharyngealization and Glottalization

Alternationq/kon the position

R1 R2 R3 Total

consonants:bdrzc

lmnh

w/y

(R2)0011133211

(R1) (R3) (∑)2 0 22 3 52 1 21 1 25 2 71 1 20 1 11 1 21 0 12 0 2

(R2)0001002001

2544856424

Total 13 27 4 44

% 65 69,23 40 61,97

b) non-voiced consonants

Alternationq/kon the position

R1 R2 R3 Total

consonants:t

¯t

˙hš

˙sf

k/q53

(R2)2121010

(R1) (R3) (∑)0 0 01 1 21 1 20 1 10 1 11 0 10 5 5

(R2)2011020

4353145

Total 7 12 6 25

% 35 30,77 60 38,03

53 This set consists of the roots R1kk/R1qq, i.e. the mediae geminatae roots.

Page 44: ZemanekPharyngealization

38 Petr ZEMÁNEK

At this type of alternation, the dominance of voiced consonants is broken,this time on the position of R3, where we get the dominance of non-voicedconsonants, the only case from all of our sets. But even at the two other positions,the dominance of voiced consonants is not as high as was the case with the alter-nations of emphatics and voiced non-emphatics. As has been said above, thispoints to a purely phonetic character of the alternation, together with the possi-bility of de-emphatization.

The distribution of both voiced and non-voiced consonants in the contactposition is surprisingly equal, most of them ranging in the interval between 4-6(voiced) and 3-5 (non-voiced). Surprisingly, it is laryngealc that appears 8 times,i.e. with the highest score. Relatively frequent is also the group of roots of thetype R1R2R2 (5 occurrences).

The following table shows the semantic division of the alternation ofq/k inArabic:

Table XV: Semantic division of the doublets ofq/k in Arabic:

Alternationq/kon the position R1 R2 R3 Total

violence, aggressivity 2 6 3 11

actions connected with move-ment

1 1 1 3

taboo actions and activities 0 2 1 3

spiritual and corporal qualities 5 2 1 8

production of sounds 1 0 0 1

"pick up" 3 1 0 4

"origine, race" 0 3 0 3

rest 8) 5 4 17

Total 20 20 10 50

Similarly to all the other sets of alternations so far we find the majority ofmeanings connected with some emotional charge and dynamics. Such a type ofsemantic environment seems to be the catalysing one for the alternations, whicheven stresses the phonetic nature of the phenomenon.

Page 45: ZemanekPharyngealization

39Pharyngealization and Glottalization

2.2.3 Emphatic˙sand its alternation with non-voiceds.

We have several times expressed our reservations in regard to including theemphatic˙sinto our analysis. These limitations are based on the problems with theoriginal way of articulation of this sound, which could be, as observed above,affricate or glottalized, or both. Especially in case of affricate articulations, therewould be problems with finding a direct voiced or non-voiced counterpart. Then,of course, the number of doublets should be lower than in the other two cases.

The following table shows the number of doublets with the alternation of˙s/sin Arabic.

Table XVI: Alternation of ˙s/sin Arabic:

Alternation ˙s/son theposition

R1 R2 R3 Total

Number of doublets 30 21 21 72

Total number of rootsin the dictionary

s ˙s

341 229

s ˙s

243 151

s ˙s

280 184

s ˙s

864 564

% 8,8 13,1 8,6 13,9 7,5 11,41 8,3 12,76

In this case, the number of doublets is very close to those with alternation of

˙s/z. In case of R1 doublets the number is even higher, in the two other casesslightly lower. Generally, there is no substantial difference in the number ofdoublets in both cases. This might point to purely phonetic reasons of bothalternations of˙s/zand ˙s/s.

The following table shows the phonetic environment at the contact positionsof the alternation of˙s/sin Arabic:

Page 46: ZemanekPharyngealization

40 Petr ZEMÁNEK

Table XVII: Phonetic environment of the alternation˙s/sin Arabic:a) voiced consonants:

Alternation ˙s/sonthe position

R1 R2 R3 Total

consonants:bdrc

glmnh

w/y

(R2)1222251103

(R1) (R3) (∑)1 1 20 0 03 2 51 0 10 1 11 1 23 3 60 0 02 0 20 1 1

(R2)2041101112

52114478236

Total 19 20 13 52

% 63,3 47,6 65 56,52

b) non-voiced consonants

Alternation ˙s/sonthe position

R1 R2 R3 Total

consonants:’

˙h

˘h

s/ ˙s54˙t

fqk

(R2)10202150

(R1) (R3) (∑)1 0 11 1 21 3 40 2 20 2 23 0 34 3 70 1 1

(R2)01000231

236246152

Total 11 22 7 40

% 36,6 52,4 35,00 43,48

54 Roots of the type mediae geminatae.

Page 47: ZemanekPharyngealization

41Pharyngealization and Glottalization

The proportions of voiced and non-voiced consonants correspond roughly towhat we have observed at the alternations of˙t/t andq/k. At this set, there is, how-ever, one considerable difference - half of the non-voiced contact positions of non-voiced consonants are taken by emphatics - emphatic˙t (4 occurrences) and uvularq (15 occurrences). This might quite clearly mean that in these cases we witness alsothe process of emphatisation, while in the other cases it was by far more probablythe vice versa process, i.e. the de-emphatization.

Within the voiced consonants, the most frequent are the liquids (r, l) and nasalm.

In this set, we found only one case of phonetically interrelated pairs of dou-blets, namely 24.s/ ˙slq - 25.s/ ˙slqmat R1.

The following table shows the results of the tentative semantic analysis of thedoublets with the alternation of˙s/s:

Table XVIII: Semantic division of the doublets of˙s/sin Arabic:

Alternation ˙s/son the position R1 R2 R3 Total

violence, aggressivity 5 3 6 14

actions connected with move-ment

2 6 1 9

spiritual and corporal qualities 6 1 5 12

production of sounds 2 1 1 4

rest 15 10 8 33

Total 30 21 21 72

2.3 Analysis of the control set: Phonologically and phonetically non-related doub-lets:

In this section, we are going to deal with the doublets of non-emphatic, phono-logically and phonetically non-related consonants, which are included here in orderto show the relevance of the method used and to produce comparable data whichwill enable us to judge the reliability of the data presented here.

The doublets for these two sets have been chosen based on purely quantitativecriteria - the first set represents the two most frequent consonants (b = 7,7%, r =5,7%). Such a choice should show us average frequency of a random alternation ofroots in Arabic.

The existence of this type of doublets in Arabic can basically be explained intwo ways: one of them is pure random similarity, the other way is based on the

Page 48: ZemanekPharyngealization

42 Petr ZEMÁNEK

assumption of the originally bi-radical character of the root in Arabic. The secondexplanation seems to us at present to be unjustifiable only based on the presence ofthis type of doublets, but would need a further examination based on additional dataand knowledge of phonetic, phonological and semantic processes involved in thistype of root development. On the other hand, solving the problem of the originalcharacter of the root in Arabic (and both Semitic and Hamito-Semitic) is of crucialimportance for the whole discipline with serious consequences for many of its parts.However, the set of data that we have at our disposal here, does not allow us to enterinto such speculations.55

2.3.1 Doublets with the alternation ofb/rAs has been said above, the inclusion of this set of doublets in our analysis is

meant as an attempt to provide comparative data that could serve to evaluate thevalidity of the sets of doublets with the alternation of non-voiced emphatics withtheir voiced counterparts.

The results of the examination of roots with theb/r alternation as reflected inKAZIMIRSKI ’S dictionary, are summarized in the following table:

Table XIX: Alternation ofb/r in Arabic

Alternationb/ron the position

R1 R2 R3 Total

Number of dou-blets

4 12 11 27

total number ofroots in the dic-tionary

b r

415 349

b r

417 798

b r

384 456

b r

1216 1603

% 0,96 1,15 2,87 1,5 2,86 2,41 2,22 1,68

The table shows that the number of doublets is in this case much lower than inall the preceding cases, but still the percentage is quite high regarding the differencebetween both types of consonants. However, such a frequency can still be ascribedto a random similarity.

The following table shows the phonetic environment at the contact positions ofthe alternation ofb/r in Arabic.

55 For this hypothesis, cf. MOSCATI ET AL. 1964:72-73, for critique VOIGT 1988 and PETRÁ-CEK 1987. The listing of this possibility does not necessarily mean that the data presentedhere can be interpreted this way.

Page 49: ZemanekPharyngealization

43Pharyngealization and Glottalization

Table XX: Phonetic environment of the alternationb/r in Arabic:a) voiced consonants:

Alternations ofb/r on theposition:

R1 R2 R3 Total

consonants:b/rgdz

˙dc

gnhw

(R2)0010010001

(R1) (R3) (∑)0 2 23 1 41 1 20 2 22 0 20 0 00 0 00 0 01 0 10 1 1

(R2)0101111100

2533321112

Total: 3 14 6 23

% 75 58,3 54,55 58,97

b) non-voiced consonants:

Alternation ofb/ron the position:

R1 R2 R3 Total

consonants:’

¯t

˙h

˘hs

˙s

˙tq

(R2)00010000

(R1) (R3) (∑)0 1 10 0 00 2 20 0 02 1 31 1 21 0 11 0 1

(R2)01200011

11413222

Total 1 10 5 16

% 25 41,7 45,45 41,03

Page 50: ZemanekPharyngealization

44 Petr ZEMÁNEK

The analysis of the phonetic environment shows some interesting facts. Theoccurrence of individual consonants is widely distributed, most of them occurringonly once (7 of 18) or twice (5 of 18). The highest frequencies haveg (5x, 3x atthe alternation at the position of R2) and ˙h (4x, equally at R2 and R3 alternations).This might point to the random character of the process, but the inequality heremight be caused by the small number of doublets at the position of R1.

The tentative analysis of the semantics of the doublets results in the followingtable:

Table XXI: Semantic division of the doublets ofb/r in Arabic.

Alternation ofb/r on the position: R1 R2 R3 Total

violence, aggressivity 1 3 4 8

actions connected with movement 2 1 1 4

spiritual and corporal qualities 1 2 1 4

rest 0 6 5 11

Total 4 12 11 27

Again, as was the case with all the preceding sets of doublets, the prevailingones are the dynamic and emotional semantic charges, and the usual most frequentcategories are present here, too. It is very probably the semantic notion that iscommon to the alternations of consonants in Arabic.

2.3.2 Doublets with the alternation of¯t/g

Similarly to the preceding set, i.e. the doublets with the alternation ofb/r, thisset should offer comparative data on the validity of the method used at the alter-nations with non-voiced emphatics. The two consonants (¯t andg) are, accordingto the figures given in HERDAN 1962:54, among to the least frequent at verbalroots, reaching the relative frequencies of 1,8% (¯t) and 1,9% (g).

The number of the doublets with the alternation of the two above mentionedconsonants is summarized in the following table:

Page 51: ZemanekPharyngealization

45Pharyngealization and Glottalization

Table XXII: Alternation of ¯t/g in Arabic

Alternation ¯t/gon the position

R1 R2 R3 Total

Number of dou-blets

3 0 0 3

total number ofroots in the dic-tionary

¯t g

144 222

¯t g

175 169

¯t g

156 112

¯t g

475 503

% 1,08 1,35 0 0 0 0 0,63 0,59

The relative frequency of doublets at the position of R1 is very similar to theaverage percentage at the alternation ofb/r, which might show the average per-centage of random (or non-phonetically conditioned) similarity. The zero occur-rence of doublets at positions of R2 and R3 decreases this similarity, but this is notvery important for our aims. The situation seems to point to the fact that therandom (or non-phonetically conditioned) similarity will probably not reach 3%.

Analysis of the phonetic environment of doublets is this time very simple,with d featuring only once andm occuring twice. The small number of doubletsdoes not represent sufficient data for drawing more general results.

A very similar situation holds for the semantic analysis: Here, we can observethe notion of quantity expressed by all the three doublets, with a small quantity inone case and a big one in the remaining two. In two cases, the meaning is con-nected with water or liquid.

In general it is clear that the occurrence of doublets of such phonetically non-related consonants that we have here in cases of the alternation ofb/r and ¯t/g isrelatively low and their frequency only slightly surpasses 2 per cent.

Page 52: ZemanekPharyngealization

46 Petr ZEMÁNEK

3. Summary and conclusion

3.1 Evaluation of the sets of alternations

In the preceding parts of the present study, we have examined a number ofsets with doublets of emphatic and its voiced counterparts, namely˙t/d, q/gand

˙s/z. The collecting of such types of doublets was based on an assumption that thistype of alternation is phonologically conditioned, influenced by the shift oforiginally glottalized consonants to the pharyngealized ones.

When collecting these types of sets based on phonetic and semantic simila-rity, Arabic (and other Semitic languages as well) can be deceitful since thesemantic similarity of phonetically similar roots is very common there, and alsothe process of collecting the doublets is based mainly on subjective criteria sinceit is sometimes hard to establish hard criteria of what is semantically similar ornot. In order to show comparative data resulting from the same way of data colle-cting we have included also two types of control sets. One of them is based onphonetically similar doublets with the emphatic consonants that we investigatehere, i.e. ˙t/t, q/kand ˙s/s. The other type tries to show the percentage of randomsimilarity of both phonetically and historically non-related consonants. For this,we have chosen two doublets, based on frequency criteria56 - i.e. two phonemesfrom the highest frequency rank and two from the lowest one, namely we werelooking for alternation ofb/r (highest frequency rank) and¯t/g (lowest frequencyrank).

It would seem that the method used when collecting the doublets, i.e. the useof control sets, offers a far more contoured image of the transition from glotta-lized to pharyngealized consonants. It also shows considerable differencesbetween the three types of doublets, although in some cases the borderlines arenot necessarily too sharp.

For a better overview of the results we have set up the following table thatshows the relative frequencies of doublets at the individual types of alternations:

56 The frequency data are taken from HERDAN 1962; they are based on dictionary entries, noton the frequency of graphemes in the text.

Page 53: ZemanekPharyngealization

47Summary and Conclusions

Table XXIII: Overview of alternations

Type ofalternation

R1 R2 R3 Total

˙t/d ˙t d18,86 14,59

˙t d18,5 14,85

˙t d14,95 12,35

˙t d17,89 14,32

˙t/t ˙t t4,71 6,62

˙t t11,00 9,24

˙t t6,54 7,4

˙t t7,34 7,96

q/g q g12,10 14,00

q g8,43 9,43

q g9,91 11,18

q g10,54 12,22

q/k q k4,6 5,88

q k8,06 9,47

q k3,03 5,47

q k4,98 6,93

˙s/z ˙s z11,35 9,28

˙s z17,21 13,33

˙s z14,67 11,34

˙s z14,18 11,22

˙s/s ˙s s13,10 8,8

˙s s13,9 8,6

˙s s11,41 7,5

˙s z12,76 8,3

b/r b r0,96 1,15

b r2,87 1,5

b r2,86 2,41

b r2,22 1,68

¯t/g ¯t g2,08 1,35

¯t g0,00 0,00

¯t g0,00 0,00

¯t g0,63 0,59

When comparing the relative frequencies of the doublets we get clearlyhigher values at the alternation of˙t/d than at the alternation of˙t/t, and at thealternation ofq/g in relation to the numbers at the alternation ofq/k, although inthe latter case we could expect a lower rate.57 The differences between the

57 According to our hypothesis, the counterpart ofq should be voiced stopg, which hasundergone a subsequent change tog in Arabic. Such a shift could, however, obscure theresults at the alternation ofq/g, since the palatalization ofg could diminish the number ofdoublets at the alternation. This could result in smaller difference between the two types ofdoublets, i.e. betweenq/g and q/k. This is, however, not the case here since as our datashow, the difference is relatively high, especially at the positions of R1 and R3, only at R2 therate of both alternations is very similar.

Page 54: ZemanekPharyngealization

48 Petr ZEMÁNEK

alternations with voiced consonants and non-voiced consonants are relativelylarge, the frequencies of the doublets of non-voiced emphatics with the voicedcounterpart are more than twice as high as the frequencies with the non-voicedsounds.

We have observed considerably less difference in the occurrence of thedoublets with emphatic˙s, where the frequencies of alternation both of voicedzand non-voiceds are very similar to each other. The fact that the emphatic˙sbehaves so differently from˙t andq indicates that its development from its proto-Semitic correspondent has not been completely parallel to the development of thetwo other consonants. On the other hand, the number of alternations with sibi-lants might point to its sibilant (not affricate) character, at least in Arabic. Itshould be noted, however, that our data only indicate necessity to further investi-gate this issue.

The frequency of the alternation ofb/r is not comparable to the situation atother alternations, but still is relatively high bearing in mind that these two con-sonants are not related both phonetically and historically. This points to thedelicate problem of drawing conclusions based on randomly collected data,without a systematic approach or at least a sufficiently closed corpus, and theproblematic character of the root in Arabic.

Alternation of ¯t/g in the overall picture is not very frequent, but at the posi-tion of R1 the results are comparable to the frequencies of the alternation ofb/r.No occurrence at the positions of R2 and R3 might be caused by the relativelylow number of roots with these two consonants.

Both these sets of doublets (b/r and ¯t/g) show that the percentage of random(or phonetically non-conditioned) alternation is not very high, only just sur-passing 2 per cent.

The considerable differencies between the three types of alternations (non-voicedemphatic / voicedcounterpart; non-voicedemphatic /non-voicedcounter-part; alternation of phonetically non-related consonants) lead us to the conclusionthat the data we present here can be used for further speculations on the characterof emphatic consonants in Arabic.

It is evident that it is the emphatic˙t that has according to statistical data pre-served most traces of the transition from the original glottalizedt’ to pharyn-gealized˙t. This is, however, a situation that could also be expected according tothe data in FREWOLDU (1984-86), where it is the emphatic˙t that exhibits highestauditory similarity to the glottalizedt’ . According to the proportion of relativefrequencies it could be also the ejective glottalizedk’ that underwent a develop-ment that might be at the beginning similar to the transition oft’ to ˙t, but the loss

Page 55: ZemanekPharyngealization

49Summary and Conclusions

of glottalization in this case has not been accompanied by the pharyngealized co-articulation, but by the shift of articulatory place to the back. The security of suchspeculations is nevertheless somewhat at doubt owing to the lower frequency ofdoublets with the alternation ofq/g (10,44%) as against˙t/d (17,89%).

3.2 Phonetic environment of alternations

As far as the phonetic environment of the alternations is concerned, we canobserve higher occurrence of voiced consonants at the alternation of emphaticwith voiced occlusives.

Table XXIV: Rate of voiced and non-voiced consonants at the alternations

˙t/d q/g ˙s/z ˙t/t q/k ˙s/s b/r ¯t/g

voiced 69,45 72,00 75,49 64,18 61,97 56,52 58,97 100

non-voiced

30,55 28,00 24,51 35,82 38,03 43,48 41,03 0

Such a situation is probably to be expected, and it emphasizes the phoneticcharacter of the transition from glottalized co-articulation to the pharyngealizedone. For a comparison, we have counted the relation between voiced and non-voiced consonants based on data from HERDAN 1962:54, and the resulting data atverbal roots in Arabic are 61,4% of voiced and 38,2% of non-voiced conso-nants.58 The table shows that while the alternations of emphatics with their non-voiced counterparts are relatively close to figures from HERDAN,59 but at thealternations of emphatics with their voiced counterparts, which is the most impor-tant set for our analysis, the rate of voiced consonants is clearly higher than theaverage.

58 One has to keep in mind that the data from HERDAN are not directly comparable with ourdata, since the inclusion of the nominal roots in our data could alter the relations. Themissing 0,4% in the sum of HERDAN’S data is probably caused by rounding off the resultsat individual consonants.59 The situation at the alternation of¯t/g is not representative, having in mind the number ofdoublets found at this alternation. It should be, however, noted, that our data are not directlycomparable to HERDAN’S, since the way of acquiring them was in both cases different.

Page 56: ZemanekPharyngealization

50 Petr ZEMÁNEK

3.3 Semantic analysis of the alternationsIt is relatively difficult to summarize the semantics of the doublets, especially

because of the tentative character of the analysis. There are also different semanticnotions at some alternations, and the results at each set are not directly compara-ble. Nevertheless, the property that is common to all the groups of doublets,60

and also to the alternationb/r,61 is the prevalence of meanings connected withviolence and aggressivity, notions connected with movement, and spiritual andcorporal qualities. These semantic notions can be described as containing dyna-mics and emotionality, and those two clearly prevail in all the groups. It should bealso noted that most of the lists of doublets in Arabic and Semitic62 do exhibitexactly this type of semantic charge. It seems highly probable that the dynamicsand the emotional charge is the semantic environment accompanying most of thevarious types of root alternations in Arabic and Semitic.

The dynamic nature is even strengthened by comparing the proportion ofverbs and nouns occuring in the doublets. The fact that most of the doublets areverbs is a confirmation of the fact that dynamics is one of the features that are veryfrequent with alternations in Arabic.

These facts confirm the phonetic character of this process, which means thatthe transition from glottalized co-articulation to the pharyngealized one in Arabic(Semitic) is best preserved thanks to the phonetic characteristics. The other fea-tures involved in this transition seem to be general features for most of the alter-nations in Arabic.

3.4 ConclusionsAs it has been said several times, there are considerable differencies between

the three types of sets, which clearly divide them into 3 categories. The relativefrequencies of the first set, i.e. emphatics alternating with their voiced counter-parts, are the highest of the three, and especially the frequency of the set ofdoublets with the alternation of˙t/d clearly overpass the other types. The secondset, i.e. doublets with the alternation of emphatics and their non-voiced counter-

60 In our sets, only¯t/g set is excluded from this type of speculation due to the limited numberof doublets and the consequent difficulties with the evaluation of this set from this point ofview.61 We do not gather here the data from the set of¯t/g alternation because the set is so limitedthat it is difficult to draw conclusions from it.62 Cf., e.g., STEINER 1977, EHRET 1989, the PR3 series mentioned in MOSCATI ET AL.1964:72-3, or examples in VOIGT 1988: passim.

Page 57: ZemanekPharyngealization

51Summary and Conclusions

parts, occupies the middle position, and the lowest frequency can be found at thealternations of phonetically non-related consonants. This leads us to the assertionthat the first set does reflect the transition of glottalized consonants to the pharyn-gealized ones. Such an assertion naturally means that the glottalized articulationof emphatics preceded the pharyngealized one. This sequence seems to us con-firmed also by other arguments, the principal being the development of emphasis(pharyngealization) in the dialects of Arabic and in Modern Aramaic (Neo-Assy-rian), where pharyngealization exhibits a clear tendency to expansion in such anextent that it lead several authors to single out the feature of emphasis (pharyn-gealization) as a suprasegmental feature (e.g., HARRIS 1942, ZAVADOVSKY 1981,etc. for Arabic, TSERETELI1982, HOBERMAN 1988, 1989, etc. for Neo-Assyrian).

The current level of our knowledge of the problem does not allow us to offermore exact conclusions about the chronological dating of the transition, but wecan speculate at least about some basic points. Based on the fact that the pharyn-gealized consonants in Arabic are also voiced, we can judge that pharyn-gealization is fully set only in Arabic. In other Semitic languages, we can probablycome up with some points that could be used for further discussion. It seems thatAkkadian did not posses pharyngealized co-articulation. This is based on twofacts, that emphatics in Akkadian did not influence the neighbouring vowels,which is a property characteristic for glottalization,63 and that emphatic soundsin Akkadian could not co-occur in one word, as shown by GEERS(1945).64 Thelatter fact by itself cannot be used for the exclusion of pharyngealization in Akka-dian, although we might think so from the fact that the cooccurrence of emphasisin Semitic changed later on and that the GEERS’ law is no longer valid in youngerSemitic languages. In these, e.g. in Hebrew, we can observe co-occurrence ofqtogether with ˙s and ˙t (cf. AEŠCOLY1939), and the same holds for Arabic. Even apassing glance through Arabic dictionary shows the co-occurrence ofq and the

63 Cf. KNUDSEN 1961; pharyngealized consonants do influence both vowels and consonantsin their neighbourhood, as has been many times shown in acoustic analyses of emphasis inArabic.64 Beside Akkadian it seems that Eblaic conforms to this rule - cf. CONTI 1990:39-41 andCONTI 1993. It should be noted, however, that, according to WEDEKIND (1990:132) Amharicallows combinations of two glottalized consonants in one root - e.g.s’itt’ita = silence, etc.It is difficult to decide whether this is an inherent feature of Semitic languages, or a resultof areal contact with the neighbouring Cushitic languages. Despite this difficulty, it has to besaid that the argument using incompatibility data for emphasis remains for the time being ata level of mere speculations.

Page 58: ZemanekPharyngealization

52 Petr ZEMÁNEK

other emphatic consonants, with the exception of˙z.65 This means that we witnessa change in the incompatibility rules, which can be the result of the change of thearticulatory characteristics of the coarticulation in Semitic, i.e. from the glottalizedcoarticulation to the pharyngealized one. That would also mean that pharyn-gealization in Semitic can be constituted as the main type of coarticulation pro-bably later than in Akkadian. In Ethio-Semitic, the glottalization is preserved dueto the areal influence of the Cushitic languages.

The picture that we have here is one of a relatively complicated model ofdevelopment. At the beginning of this process there existed glottalized (ejective)consonants (t’, k’ , s’, etc.) that gradually start to change from one type of secon-dary articulation (glottalization) to another one (pharyngealization). At a certainstage of this transitory process, after the release of the glottal closure, this releaseis in certain circumstances (voiced neighbourhood, dynamic or emotional mean-ing) substituted not only by pharyngealization, but also by voicedness. In thisstage, the doublets with the alternation of emphatic (originally glottalized) andnon-emphatic voiced consonants appear. The new type of coarticulation starts tochange the characteristics of ejectivek’ from a consonant with coarticulatorycharacteristics to a consonant without coarticulation, and the main articulatoryposition is drawn to the back.66

The emphatic˙s stands a little aside from these speculations, due to the factthat its position in the protosystem seems to be slightly different from that of theabove two consonants.67 Data that we collected also show a behaviour that isdifferent from the other two non-voiced emphatic consonants in Arabic.

After the constitution of pharyngealization as the coarticulatory realization ofthe emphatic consonants, the behaviour of the emphatics changes. This is causedmainly by the fact that pharyngealization, unlike glottalization, has a strongtendency to spread, both progressively and regressively, as has been shown inmany studies on their phonetic characteristics in Arabic. This results in the esta-blishment of pharyngealization as the main type of coarticulation and it coversalso other types of coarticulation that existed in proto-Semitic, e.g. lateralizationat ˙d. Uniform series of coarticulated consonants is thereby established in theClassical Arabic series of˙t, ˙d, ˙s and ˙z. The tendency to spread can be observed

65 E.g., cf. in WEHR’S dictionary verbal roots like˙sqb, ˙sqc, ˙sql; ˙tqq, ˙tqs, ˙tqm, and a numberof roots of the type ofq ˙s3 (10 roots),q ˙d3 (5 roots) andq ˙t3 (9 roots).66 For the impossibility of backing coarticulation of back consonants, cf. DELATTRE 1971.67 The possible affricate realization of this consonant in proto-Semitic would mean that thecounterparts of this consonant are not directlys andz, but other consonants.

Page 59: ZemanekPharyngealization

53Summary and Conclusions

even in Classical Arabic itself, where Arabic grammarians68 admit the possibilityof emphatisation (’i ˙tb aq) of r and l. The spread of this type of coarticulation isespecially evident in Modern Semitic languages - i.e. modern Arabic dialects andNeoassyrian.69 It should be noted, however, that coarticulation in all the casesdoes not have to be pharyngealization.70 The case is rather different with ModernHebrew due to the Western style of pronunciation of emphatics.

It is exactly this type of development that shows that pharyngealization issecondary. The development in modern Semitic languages, where emphasis ispreserved, shows that the other direction of development, i.e. from pharyngeali-zation to glottalization, is not very probable, since in all of them the developmentof pharyngealized coarticulation goes in a different direction.

68 According to SCHAADE 1911:14-16.69 In this language, the spread seems to be very strong. According to TSERETELI 1982 andHOBERMAN 1988, it covers most of the phonological system, and that is why they understandit to be a suprasegmental phoneme.70 As has been pointed out, in case of Arabic one meets coarticulation types like labialization(emphaticm, b), "strong nasalization" (emphaticn - cf. MALAIKA 1959), etc. For modernAramaic, HOBERMAN 1988 and 1989 comes with the feature "constricted pharynx", whichwould mean pharyngealization or a coarticulation type close to pharyngealization. Thisshould be, however, in order to arrive at a definite solution, confirmed by acoustic analysisof emphatics in modern Aramaic.

Page 60: ZemanekPharyngealization

54 Petr ZEMÁNEK

3.5 Model of the development of pharyngealization in Arabic/Semitic

Page 61: ZemanekPharyngealization

55Supplement: List of Doublets

4. Supplement:

List of doublets with the alternation of voiceless emphaticand its counterpart (voiced, non-voiced)

Page 62: ZemanekPharyngealization

56 Petr ZEMÁNEK

4.1 Doublets with the alternation of˙t/d:(T = ˙t/d)

4.1.1 Alternation of˙t/d on the position of R1:

1. T’T’ ˙t: pencher, baisser; / d: pencher, aller de côté2. Tbs ˙t: noir; / d: tout ce qui est noir ou se présente comme une masse

noire3. Tbq ˙t: glu pour prendre les oiseaux; / d: glu.4. Tbl ˙t: battre le tambour; / d: frapper quelqu’un à coups redoublés,

une fois après l’autre, avec une bâton.5. T ˙h ˙h ˙t: étendre à plat, comme un tapis; / d: s’étendre, se détendre,

s’élargir, se dilater.6. T ˙hr ˙t: pousser en avant, donner la chase; / d: éloigner, écarter,

chasser.7. T ˙hm ˙t: violent, qui pousse avec violence; / d: pousser violemment

en frappant par derrière.8. T ˙hn ˙t: petit, courtaud, de petite taille; / d: petit de taille et ventre.9. T ˙hw/y ˙t: étendre à plat comme un tapis par terre; / d: étendre comme

un tapis ou une natte (se dit de Dieu, qui a étendu la terre et enfait une surface plane).

10. T˘hr ˙t: homme maigre, chétif et sans vigeur; / d: être petit, chétif et

méprisé comme tel.11. T

˘hm(d

˘hms) ˙t: tirant sur le noir; / d: épais et noir.

12. Trb 1. ˙t: faire de la musique ou chanter pour emouvrir quelqu’un;/ d: habile, qui joue avec art et habileté d’un instrument demusique.

13. Trs ˙t: effacer en frottant; / d: être effacé, s’effacer.14. Tss ˙t: s’enforcer dans intérieur du terres, du pays; / d: cacher un

objet sous un autre ou sous terre.15. Tsc ˙t: parcourir, traverser un pays, s’enforcer dans l’intérieur des

terres; / d: s’enforcer, être enfoncé.16. Tsm ˙t: être effacé; / d: s’effacer, disparaître.17. Tcs ˙t: cohabiter avec une femme; / d: cohabiter avec une femme.18. Tcz ˙t: cohabiter avec une femme; / d: cohabiter avec une femme.19. Tgr ˙t: voy dgr; / d: pousser en donant un coup par derrière.20. Tgy ˙t: orgueilleuse, injuste, méchante, rebelle (femme); / d: mé-

chanceté, mauvaise nature.21. Tf’ ˙t: s’éteindre (se dit de la flamme, du feu); / d: être chaud, con-

tenir la chaleur.

Page 63: ZemanekPharyngealization

57Supplement: List of Doublets

22. Tqq ˙t: bruit produit par le choc d’un corps dur contre une pierre; /d: casser, concasser, battre, frapper à la porte.

23. Tll ˙t: grace, beauté, élégance; / d: bonnes manières, manièresaissées (qui n’excluent pas ni la gravité ni la décence).

24. Tlq ˙t: être renvoyé, laché, mis en liberté (se dit d’une chamellequ’on laisse paitre librement en lui otant les entraves); / d:envoyer, lancer, lacher.

25. Tmm ˙t: malheur, calamité; d: se conduire mal, agir mal.26. Tm

˘h ˙t: être fier; / d: être vil, bas, ignoble.

27. Tms ˙t: ( ˙tmsl) être impuissant à la cohabitation; / d: cohabiter avecune femme.

28. Tnn ˙t: bourdonner (se dit des insectes); / d: bourdonner (se dit decertains insectes, etc.).

29. Tn’ ˙t: rougir, avoir honte; / d: être vil, bas, ignoble.30. TnTn ˙t: bourdonner (se dit des insectes); / d: bourdonner (se dit des

certains insectes).31. Tnfs ˙t: devenir mauvais, méchant; / d: méchant.32. Thš ˙t: être derangé, troublé dans son travail au point de gâter l’ou-

vrage que l’on fait; / d être jeté dans la stupefaction, dans letrouble, perdre la présence d’esprit.

33. Thq ˙t: marcher avec rapidité; / d: presser, faire marcher plus vite,pousser.

34. Thl ˙t: petite quantité de fourrage; / d: peu, petite quantité.35. Thm ˙t: s’épouvanter, et se presser en foule les uns sur les autres sous

l’impression de la peur; / d: foule, multitude, nuée.36. Thw/y ˙t: être habile (dans un art ou un métier); / d: être habile, adroit.37. Twr ˙t: disque, rond; / d: rond, disque.38. Tws ˙t: fouler avec les pieds; / d: fouler le sol avec les pieds.39. Tw/y

˘h ˙t: anéantir quelqu’un; / d: soumettre, assujetir, conquérir.

40. Twl ˙t: richesse, grande fortune, abondance des biens; / d: richesse,opulence.

4.1.2 Alternation of˙t/d on the position of R2:

1. bTT ˙t: faire le commerce de canards; / d: faire un commerced’échange.

2. bT ˙h ˙t: s’étendre, être vaste; / d: vaste, spacieux, ouvert de touscotés.

3. bTg ˙t: même signif. quebdg; / d: être sali d’excréments humains.

Page 64: ZemanekPharyngealization

58 Petr ZEMÁNEK

4. bTg ˙t: même signif. quebdg; / d: être sali d’excréments humains.5. bTn ˙t: qui a un gros ventre; / d: obésité, corpulence.6. ˙hTr ˙t: être jeté par terre; / d: jeter, lancer quelque chose.7. ˙hTm ˙t: brûler; / d: être brûlant, brûler.8.

˘hTT ˙t: chemin; / d: chemin.

9.˘hTf ˙t: marcher d’un pas rapide; / d: marcher à pas menus et rapides.

10.˘hTw/y ˙t: faire un pas, posant le pied sur le sol, marcher; / d: aller d’un

pas rapide.11. rTb ˙t: tendre, délicat; / d: traiter quelqu’un avec doucer, avec bien-

viellance.12. rTc ˙t: cohabiter avec une femme; / d: cohabiter avec une femme.13. rTm ˙t: embarasser, embrouiller quelqu’un de manière qui il lui soit

difficile de se dégager; / d: obstruer, embarasser.14. rTw/y ˙t: sottise, stupidité; / d: ignorance, bêtise.15. sT ˙h/

˘h ˙t: ( ˙h) étendre à plat, faire une surface plane de q.c.; / d: (

˘h) être

étendu, s’étendre.16. sTc ˙t: frapper, battre des mains; / d: être frappé.17. sTm ˙t: fermer (la porte); / d: fermer, barricader (la porte).18. sTw/y ˙t: qui marche d’un pas large (cheval); / d: qui marche à larges

enjambées (chameau).19. šTT ˙t: être injuste et oppresseur envers quelqu’un; / d: traiter quel-

qu’un avec dureté, avec sévérité, maltraiter, persécuter ...20. štw/y ˙t: diminuer, amoindrir, réduire; / d: un peu, petite quantité qui

reste d’un grand nombre.21. cTm ˙t: périr; / d: ne pas exister.22. cTn ˙t: endroit autour d’un abreuvoir où les troupeaux (chameaux ou

moutons) réponsent après avoir bu et avant de revenir au pâtu-rage; / d: être à demeure fixe dans un lieu; continuer de paitreses troupeaux dans les mêmes pâturages.

23. gTl ˙t: vie aisée, exempte de soucis; / d: aisance, bienêtre.24. fTr ˙t: fendre, pourfendre, couper en deux; / d: se briser, être brisé,

cassé en morceaux grandes et petites.25. qTT ˙t: couper, sourtout dans le sens de la largeur; / d: couper ou

déchirer en lanières, en lambeaux, dans le sens de la longuer.26. qTc ˙t: être arreté et empeché de continuer son chemin; / d: arreter

(son cheval) en tirant la bride à soi.27. qTw/y ˙t: marcher d’un pas menu (soit lent soit rapide); / d: marcher

rapidement.

Page 65: ZemanekPharyngealization

59Supplement: List of Doublets

28. lTT ˙t: exciter quelqu’un ou l’aider à refuser à un autre ce qui lui re-vient de droit; / d: se querreller avec quelqu’un, avoir une rixe.

29. lTs ˙t: frapper un corps large et plat avec un autre corps large etplat; / d: frapper, tapper quelqu’un avec la main.

30. lTm ˙t: donner à quelqu’un (sur la joue ou sur le corps) un coup duplat de la main; / d: se frapper le visage.

31. mTT ˙t: tendre et allonger une chose en la tirant avec force; / d: allon-ger en tirant, tirer pour allonger.

32. mT˘h ˙t: salie le nom, la réputation de quelqu’un, médire de lui; / d:

être injuste.33. mTw/y ˙t: tirer, trainer; / d: tirer, extraire.34. nTb ˙t: irriter, exciter l’un contre l’autre, mettre aux prises l’un avec

l’autre; / d: exciter, pousser quelqu’un à faire quelque chose.35. hTT ˙t: hommes qui périsent; / d: (hd’) mourir, être mort.36. hThT ˙t: hennir (se dit d’un cheval); / d: mugir (se dit d’un chameau).37. wT’ ˙t: cohabiter avec une femme; / d: dresser son pénis (se dit d’un

cheval).

4.1.3 Alternation of˙t/d on the position of R3:

1. bcT ˙t: éloigner; / d: être éloigné.2. blT ˙t: se coller à la terre, au sol; / d: être à terre, coucher et se coller

fortemment à la terre.3. ¯tlT ˙t: rendre des excréments liquides; / d: rendre des excréments

liquides.4. g’T ˙t: se remplir d’eau, en avoir bu beacoup; / d: boire d’un seul

trait, à pleine gorge et sans humer.5. glT ˙t: tirer le sabre du fourreau; / d: se battre au sabre l’un avec

l’autre.6. ˙hwT ˙t: garder quelque chose, veiller sur quelque chose; / d: (var.¯d)

garder quelque chose, veiller sur quelque chose.7. rbT ˙t: lier, serrer des liens; / d: lier, attacher avec des liens.8. r ¯tT ˙t: se tenir à la même place sans bouger; / d: faire halte.9. rwT ˙t: s’écarter de la ligne droite, en allant à gauche ou à droite; /

d: aller cà et là,tantôt s’avancer, tantôt revenir sur ses pas.10. zrT ˙t: avaler (une bouchée); / d: avaler (une bouchée).11. zlT ˙t: bouchée qui tombe pendant qu’on la porte à la bouche; / d:

avaler une bouchée.12. srT ˙t: aigu, tranchant (sabre); / d: percer, trouer.

Page 66: ZemanekPharyngealization

60 Petr ZEMÁNEK

13. sqT ˙t: languir, faiblir (dans la course - se dit d’un cheval); / d: a-maigrir, rendre maigre (un cheval).

14. š ˙hT ˙t: être absent et très éloigné, emmener par force; / d: (var.¯d)chasser, éloigner quelqu’un.

15. šnT ˙t: rôtir; / d: préparer le repas.16. šyT ˙t: perdre, faire périr, anéantir; / d: périr.17. ˙dn/mT ˙t: ( ˙dn ˙t) avoir deux amants à la fois (se dit d’une femme); / d:

( ˙dmd) avoir deux amants à la fois (se dit d’une femme).18. ˙tw/yT ˙t: grand, long, haute; / d: grand, énorme.19. czT ˙t: voy czd; / d: cohabiter avec une femme.20. clT ˙t: petits, courts (en parlant des ânes); / d: grand.21. gmT ˙t: humer et boire avec avidité; / d: avoir beacoup d’eau.22. frT ˙t: être premier, être supérieur à tous; / d: être seul, unique et

isolé.23. kšT ˙t: oter l’écume du pot; / d: écumer.24. lbT ˙t: se coucher, être couché par terre; / d: être par terre et presque

collé au sol.25. l ˙hT ˙t: pousser, ou repousser avec violence; / d: injuste, méchant.26. lhT ˙t: frapper quelqu’un du plat de la main; / d: repousser quel-

qu’un en lui portant un coup de poing sous les mamelles.27. mcT ˙t: étendre, allonger; / d: tirer avec force, comme quand on veut

arracher ou enlever promptement quelque chose.28. mgT ˙t: long, très grand de taille; / d: gros, long et charnu.29. n ¯tT ˙t: pousser, sortir de dessous terre (se dit d’une truffe, etc.); /

d: pousser, sortir de dessous terre (se dit d’une truffe, etc.).30. nhT ˙t: percer quelqu’un avec une lance; / d: pénétrer, traverser de

part en part.31. hrT ˙t: déchirer quelqu’un en son absence, attaquer à son réputation;

/ d: déchirer quelqu’un en son absence.32. hyT ˙t: s’éloigner; / d: éloigner, écarter quelqu’un.33. w

˘hT ˙t: marcher d’un pas accéléré; / d: marcher vite ou à larges en-

jambées et d’un pas semble à celui de l’autruche.

4.2 Doublets with the alternation ofq/g:(Q = q/g)

4.2.1 Alternation ofq/gon the position of R1:

1. Q’b q: grand, d’un grand capacité; / g: gros, épais.2. Qbb q: couper, abattre; / g: couper, retrancher en coupant.

Page 67: ZemanekPharyngealization

61Supplement: List of Doublets

3. Qbw q: enlever, cueillir quelque chose avec les doigts; / g: lever,percevoir l’impôt, le tribut...

4. Q ¯t ¯t q: attirer à soi en traînant quelque chose; / g: arracher, dera-ciner; arracher (un arbre) du sol.

5. Q ¯tw q: amasser (des richesses); / g: ramasser, réunir.6. Q ˙hf q: enlever, emporter tout; / g: enlever en balayant.7. Qdd q: couper ou déchirer...; / g: couper, séparer, retrancher en

coupant.8. Qds q: être pur, sans tache; être saint; / g: terrain inculte, qui n’a

jamais été labouré.9. Qdc q: frapper sur le nez avec mépris; / g: mutiler quelqu’un en lui

coupant soit le nez, soit les lèvres, soit les oreilles.10. Qd/ ¯df q:( ¯d) jeter (l’eau du vase); / g: (d) jeter, lancer quelque chose.11. Q ¯d ¯d q: couper également les extrèmités; / g: être coupé, enlevé,

arraché.12. Q ¯dc q: saletés, ordures; / g: homme bas et sans caractère.13. Q ¯dm q: se dépêcher, aller vit; / g: rapide dans sa course, dans sa mar-

che.14. Qr ˙h q: blesser quelqu’un (voygr ˙h); / g: blesser quelqu’un.15. Qrd q: branche de palmier dépouillée de ses feuilles (comp.ga-

r ıd); / g: branche de palmier dépouillée de feuilles.16. Qrz q: sol dur, raboteux et inégal; / g: sol qui souffre de la séche-

resse.17. Qrs q: petits cousins (insectes); / g: insecte qui se dévore, qui se

nourrit de quelque plante.18. Qršb q: agé, très-vieux; / g: arriver à l’âge critique, à l’âge de cin-

quante ans, et cesser d’avoir ses règles (se dit de la femme).19. Qrf q: enlever l’écaille, la croûte; / g: enlever, emporter tout en ba-

layant.20. Qrm q: retenir, emprisonner; / g: charger quelqu’un d’un crime, le

lui imputer à faux.21. Qrn q: manger à la fois des dattes fraîches et des dattes séches; / g:

mettre les dattes en tas sur l’aire.22. Qrw/y q: percer quelqu’un avec la lance; / g: pénétrer quelque part.23. Qss q: chercher, rechercher, poursuivre quelque chose avec persi-

stance; / g: fouiller quelqu’un, chercher dans sa poche.24. Qsm q: beauté, élégance des formes; / g: beau.25. Qsn q: avoir les mains duurcies, calleuses à force de travailler; / g:

être dur et fort.

Page 68: ZemanekPharyngealization

62 Petr ZEMÁNEK

26. Qsw q: être sec et durci; / g: être dur au toucher (se dit de la main,ou de quelque outre membre).

27. Qšš q: broyer, écraser en frottant dans le main; / g: brisser, casser.28. Qc ˙t q: chasser, éloigner; / g: éloigner, chasser.29. Qcf q: arracher (un arbre) avec la racine; / g: arracher.30. Qff q: sécher, se dessécher et devenir aride (se dit d’une plante); /

g: sécher, être sec (se dit des vêtements, ...).31. Qfš q: traire (une femelle) avec rapidité; / g: exprimer en serrant

légèrement avec le bout des doigts (p.ex. traire une femelle enserrant légèrement le pis).

32. QfQf q: se dessécher (se dit des végétaux); / g: sécher à peu près en-tièrement (se dit du linge etc.).

33. Qfn q: égorger (une brebis) en portant le couteau sur la nuque; /g: égorger un chameau et en servir les morceaux dans des écu-elles...

34. Qfw q: frapper quelqu’un à la nuque; / g: traiter avec dureté et injus-tice, opprimer, tyranniser.

35. Qll q: grandir (se dit des plantes); / g: être grand, haut, imposant.36. Qlc q: arracher; oter quelque chose de sa place; / g: oter (l’habit, le

vêtement).37. Qlf q: dépouiller d’écorce (un arbre); / g: enlever, oter (p.ex. la

peau ou la bone qui s’atrtache à quelque chose).38. QlQl q: remuer, agiter, secouer; / g: agiter avec les doigts (une cre-

celle, un grelot etc.).39. Qlm q: couper, rogner; / g: couper, retrancher une partie en coupant.40. Qmm q: troupe des hommes; / g: troupe des hommes.41. Qm ˙h q: éloigner, chasser quelqu’un...; / g: partir de toute sa vitesse,

s’éloigner avec rapidité.42. Qm

˘h q: être fier et 1être assis avec aire de suffisance et d’orgueil; /

g: être fier, orgueilleux.43. Qml q: être très-nombreux, pulluler (se dit d’une population nom-

breuse); / g: grande quantité, grande somme.44. Qn ˙h q: plier, courber, cambrer; / g: se pencher, s’incliner.45. Qns q: racine, origine; / g: origine commune, ou même origine,

même genre.46. Qny q: grappes des dattes; / g: dattes fraîches, récemment cueillies.47. Qhr q: vainquer, victorieux; / g: vaincre, avoir le dessus.

Page 69: ZemanekPharyngealization

63Supplement: List of Doublets

48. Qhm q: avoir de l’aversion pour quelque chose; / g: regarder quel-qu’un de travers, la recevoir mal, lui faire une mauvaise mine,une grimace...

49. Qwb q: creuser (la terre); / g: creuser.50. Qwh q: puissance, pouvoir, rang élevé et influence (voyg ah); / g:

rang, dignité, honneurs.

4.2.2 Alternation ofq/gon the position of R2:

1. ¯tQl q: être grosse (se dit d’une femme); / g: femme de grande taille.2. ˙hQQ q: venir chez quelqu’un; / g: aller et venir; marcher.3. ˙hQn q: retenir; / g: attirer quelque chose à soi avec un baton crochu.4. dQl q: cohabiter avec une femme; / g: cohabiter, être en coït, copu-

lation.5. rQf q: trembler, frissonner (du froid); / g: trembler; qui tremble.6. šQQ q: fendre, pourfendre (le bois), casser le bâton; / g: sillonner,

fendre.7. šQ ¯d q: nuire ou porter malheur par son mauvais oeil; / g: faire de

mal, se faire sentir, causer de la peine à quelqu’un (se dit d’unechose).

8. šQc q: nuire à quelqu’un par son mauvais oeil; / g: rendre hostile,malveillant.

9. ˙sQQ q: produire un bruit (se dit du bruit que produit de caméléon);/ g: sonner, produire un cliquetis (se dit du fer qui heurte contrele fer).

10. cQr q: blesser, faire une plaie; / g: fondre sur quelqu’un le sabre,etc., à la main.

11. fQs q: saisir quelqu’un par les cheveux de derrière et le traîner avecviolence; / g: traiter quelqu’un avec dureté et violence.

12. fQc q: accabler ou briser quelqu’un (se dit des malheurs); / g: frap-per, affecter, accabler quelqu’un (se dit d’un malheur qui faitperdre à quelqu’un, quelque chose qui lui est cher).

13. mQc q: avaler avec avidité (un liquide); / g: manger des dattes etboire en même temps, ou ausitôt après, du lait doux ou caillé.

14. nQf q: tirer, extraire une chose d’une autre; / g: tirer, extraire.15. hQr q: grand, long et stupide; / g: plus long et plus gros.16. hQf q: manger d’appétit; / g: avoir faim.17. hQm q: dompter, soumettre, assujétir; / g: fondre avec impétuosité

sur quelqu’un, aborder quelqu’un à l’improviste.

Page 70: ZemanekPharyngealization

64 Petr ZEMÁNEK

18. hQhQ q: marcher d’un pas accéléré et violent; / g: stimuler un cha-meau à la marche en criant.

19. wQb q: se coucher (se dit du soleil); / g: se coucher (se dit du soleil).20. wQm q: traiter quelqu’un avec violence et dureté; / g: frapper quel-

qu’un d’un coup de poing.21. wQy q: qui craint de marcher sur un sol dur, n’ayant pas le sabot

muni d’une semelle (cheval); / g: avoir le sabot usé par delongues marches.

4.2.3 Alternation ofq/gon the position of R3:

1. ’wQ q: se trouver sur une hauter et dominer ce qui est en bas; / g: leplus haut point du ciel.

2. bcQ q: égorger (un chameau); /g: fendre le ventre avec le couteau.3. blQ q: ouvrir brusquement la port, ou l’ouvrir tout entière; / g:

ouvrir.4. bwQ q: entourer, ceindre, clore d’une cloison; / g: envelopper,

accabler, opprimer.5. glQ q: raser la tête; / g: tête.6. ˙hbQ q: péter; / g: péter.7. ˙hdQ q: blesser quelqu’un, causer une lésion à la prunelle; / g: frap-

per.8.

˘hbQ q: péter; / g: lâcher un pet.

9. d ˙hQ q: ne pas retenir, rejeter le sperme; / g: forcer une fille et coha-biter avec elle.

10. drQ q: aller vite, se hâter; / g: aller, marcher, s’avancer.11. dmQ q: entrer brusquement et sans permission chez quelqu’un; / g:

entrer, s’engrener et y tenir.12. rdQ q: voy rdg; / g: s’avancer, marcher pas à pas.13. rhQ q: exciter quelqu’un au mal, à la rebellion; / g: soulever, exciter

la poussière.14. zcQ q: crier à quelqu’un, pousser un cri contre quelqu’un; / g: crier,

pousse un cri.15. zlQ q: glisser sur un terrain glissant; / g: courir légérement et glis-

ser à la surface du sol.16. s ˙hQ q: courir doucement (comp.s ˙hg); / g: courir, aller vite.17. sfQ q: souffleter (les joues); / g: violent souffle du vent.18. snQ q: étoile qui brille d’un vif éclat; / g: lampe.

Page 71: ZemanekPharyngealization

65Supplement: List of Doublets

19. ˙slQ q: frapper quelqu’un avec un bâton, fustiger; / g: frapper avecun bâton, donner à quelqu’un une rincée.

20. cbQ q: homme d’un mauvais caractère; / g: homme qui n’a rien debon en lui.

21. cblQ q: fort, robuste, grand; / g: gros, épais.22. cfQ q: fouetter et disperser (en parlant du vent qui fouette le sol et

disperse ce qui se trouve à sa surface; / g: frapper, battre.23. cwhQ q: 1. long 2. Awhak, nom d’un chameau dont provient une

belle race du chameaux; / g: 1. qui a un long cou, épithète del’autruche, de la gazelle, du chameau; 2. Jeune chamelle.

24. cwQ q: arrêter quelqu’un; retenir quelqu’un ou le détourner de quel-que chose; / g: faire faire halte à quelqu’un, le faire rester dansun lieu.

25. gbQ q: donner à boire à quelqu’un, lui faire boire un coup du soir;/ g: humer, boire d’un trait, en humant.

26. g ˙slQ q: ne pas bien cuire les viandes et ne pas y mettre le sel; / g: nepas bien cuire les viandes et ne pas y mettre le sel.

27. gmQ q: être humide (se dit de la terre ou des herbes; / g: humer etboire avec avidité.

28. frQ q: fendre, pourfendre et séparer en deux; / g: fendre, pour-fendre.

29. flQ q: fendre, couper en deux; / g: fendre en deux, pourfendre;partager en deux, rompre.

30. lzQ q: s’attacher, se coller à quelque chose; / g: se coller, s’atta-cher.

31. mlQ q: teter sa mère (se dit d’un petit); / g: saisir avec le bout deslèvres le sein de sa mère (se dit d’un enfant qui se met à teter);teter (sa mère).

32. nbQ q: lâcher un pet léger; / g: cul, derrière (ka ¯dibat nib agatuhu- ila lâché un pet).

33. hmlQ q: fort, robuste, grand; / g: fort, robuste, grand.34. hw/yQ q: long et mince, grand et mince; / g: long, qui a le corps long.35. wšQ q: percer quelqu’un avec une lance; / g: lance.36. whQ q: être brûlant, renvoyer une chaleur brûlante (se dit des

cailloux embrasés par l’ardeur du soleil; / g: brûler.

Page 72: ZemanekPharyngealization

66 Petr ZEMÁNEK

4.3 Doublets with the alternation of˙s/z:(S = ˙s/z)

4.3.1 Alternation of ˙s/zon the position of R1:

1. S’b ˙s: être rempli de boisson, en avoir bu jusqu’à satiété, être gorgéde ...; / z: boire à grands traits en se hâtant.

2. S’m ˙s: se gorger d’eau; / z: dévorer, avaler avec rapidité.3. Sbr ˙s: éclat, morceau de rocher ou de fer; / z: morceau de fer.4. Sbc ˙s: montrer quelqu’un au doigt avec mépris ou en faisant des

reproches; / z: se mettre dans une colére violente contre quel-qu’un, tempêter.

5. Sbn ˙s: éloigner, détourner de sa destination; / z: qui repousse endonnant un coup violent; situé loin et à l’écart des autres (mai-son, etc.).

6. Sdr ˙s: ramener quelqu’un, faire revenir d’un lieu; / z: voy˙sdr.7. Sdg ˙s: coussin, oreiler; / z: voy˙sdg.8. Sdm ˙s: tomber sur quelqu’un, atteindre, frapper quelqu’un; / z:

attaquer, assaillir quelqu’un (se dit des brigands).9. Srr ˙s: serrer et nouer une bourse; / z: fermer, serrer.10. SrSr ˙s: crier (se dit de la voix du pivert); / z: chanter (se dit de l’é-

tourneau).11. Src ˙s: jeter avec force, violemment par terre; / z: ensemencer un

champ de quelque graine.12. Srm ˙s: couper, retrancher en coupant; / z: interrompre, arrêter tout

court, faire cesser, couper court à quelque chose.13. Sctr ˙s: espèce d’origan ou de thym (voyzctr); / z: marjolaine.14. Scfq ˙s: homme vil, méprisable; / z: homme méchant.15. Scq ˙s: cri, vocifération; / z: crier à quelqu’un, pousser un cri contre

quelqu’un.16. Sqr ˙s: faucon; / z: voysqret ˙sqr.17. Sqc ˙s: lâcher un pet; / z: lâcher un gros pet (se dit de l’âne).18. Slm ˙s: couper et arracher avec la racine (le nez, une oreille); / z:

couper, mutiler (p.ex. le nez).19. Smk ˙s: être en colére; / z: irriter, mettre quelqu’un dans une colére

violente contre un autre.20. Shw ˙s: être riche, posséder beacoup de troupeaux, de biens; / z:

prospérer, être dans un état florissant ou riche de verdure.21. Swb ˙s: verser, répandre quelque chose; / z: couler (se dit de l’eau).22. Sw ˙h ˙s: fendre; / z: disperser (ce qui était réuni).

Page 73: ZemanekPharyngealization

67Supplement: List of Doublets

23. Swr ˙s: faire incliner, faire pencher, ou faire pendre; / z: être penché,incliné, aller en biais, aller obliquement.

24. Swc ˙s: séparer deux choses en les écartant; / z: couper pour quel-qu’un une tranche de melon ou de viande.

25. Swg ˙s: mentir, dire une mensonge, faire des mensonges, inventer; /z: être injuste et s’écarter de la vérité dans ce qu’on dit.

26. Sy˙t ˙s: vacarme, tintamarre, bruit, cris confus; / z: crier, pousser descris, des vociférations.

4.3.2 Alternation of ˙s/zon the position of R2:

1. ’Sn/m ˙s: cordes avec lesquelles on attache aux pieux fichés en terreles pans de la tente; / z: s’attacher à quelqu’un et en être insé-parable.

2. bSbS ˙s: rapide, fait accompli rapidement (se dit du voyage que fontles chameaux dans la nuit); / z: vitesse et impétuosité de lacourse, du mouvement.

3. bSq ˙s: Même signif. quebzq; cracher; / z: cracher, jeter la salive.4. bSl ˙s: peler, mettre à nu en ôtant l’envelope extérieure ou l’écorce;

/ z: oter, enlever.5. ˙hSS ˙s: se partager, se distribuer des portions, aller au partage; / z:

tailler, faire une entaille, des entailles, des coches, à quelquechose.

6.˘hSl ˙s: couper un morceau (de viande, etc.); / z: couper, retrancher,

séparer.7. rSS ˙s: plomb; / z: plomb.8. rSf ˙s: rapprocher, joindre (p.ex., les pieds en faisant la prière); /

z: approcher, être près, imminent; s’approcher de quelqu’un.9. šSb ˙s: être sec, desséché, aride; / z: être sec et fané (se dit d’un ra-

meau).10. šSr ˙s: percer d’un coup de lance; / z: percer; porter au coup de

lance à quelqu’un.11. šSw ˙s: être suspendu très-haut dans les airs (se dit d’un nuage); / z:

être elevé.12. cSS ˙s: devenir dur, se durcir; / z: sol dur.13. cSd ˙s: cohabiter avec une femme; / z: cohabiter avec une femme.14. cSf ˙s: souffler avec violence (se dit du vent); / z: siffler (se dit des

démons, que l’on croit entendre siffler dans le désert).

Page 74: ZemanekPharyngealization

68 Petr ZEMÁNEK

15. fSS ˙s: être en suppuration (se dit d’une plaie); / z: saigner ou sup-purer (se dit d’une plaie).

16. lSS ˙s: s’attacher et se coller fortement (syn.lzz); / z: se coller et êtrejoint (se dit des choses).

17. lSb ˙s: être collé sur les os; / z: s’attacher fortement, se coller àquelque chose.

18. lSq ˙s: être collé, se coller, s’agglutiner à quelque chose; / z: s’atta-cher, se coller à quelque chose.

19. mSS ˙s: sucer; / z: sucer.20. mSd ˙s: intensité du froid; / z: froid.21. nS’ ˙s: stimuler à la marche de sa voix (une bête de somme); / z:

pousser, exciter quelqu’un à quelque chose.22. nSb ˙s: fredonner, chantonner (se dit des palefreniers qui fredonnent

certains airs); / z: crier (en parlant de la voix propre aux ga-zelles).

23. hSS ˙s: casser, briser; / z: (hz’) briser, casser.24. hSr ˙s: repousser quelqu’un et éloigner; / z: éloigné, repoussé à

coups de batons.25. hSm ˙s: 1. homme fort, robuste; 2. lion; / z: 1. fort, robuste; 2. lion.26. wSm ˙s: casser, feler (p.ex. une vase en bois ou par en terre); / z:

briser, casser, rompre.

4.3.3 Alternation of ˙s/zon the position of R3:

1. ’bS ˙s: être agile et rapide à la course; / z: faire un bond, un saut,s’élancer pour courir (se dit d’une gazelle).

2. b˘hS ˙s: crever l’oeil et l’arracher avec les chairs et la graisse qui

l’entoure; / z: crever l’oeil.3. bl’S ˙s: se sauver, s’enfuir; / z: fuir.4. blS ˙s: prendre, reprendre tout sans rien laisser; / z: accepter, prend-

re, recevoir etc. quelque chose de quelqu’un.5. bhS ˙s: éloigner, écarter quelqu’un de quelque chose; / z: repousser,

éloigner.6. trS ˙s: être ferme, solide; / z: être dur, rude au toucher.7. g’S ˙s: boire, avaler à grands traits; / z: être suffoqué en buvant d’un

seul trait.8. gnS ˙s: mourir; / z: mourir.9.

˘hrS ˙s: réparer, arranger; / z: mettre en ordre, arranger (ses affaires).

10. rfS ˙s: pousser, frapper avec le pied; / z: frapper, battre.

Page 75: ZemanekPharyngealization

69Supplement: List of Doublets

11. rqS ˙s: sauter, sautiller; / z: sauter, bondir.12. šmS ˙s: contracté, ratatiné, ridé; / z: se contracter, se renfronge (se dit

du visage).13. clS ˙s: déranger l’estomac (se dit de l’indigestion qui affecte esto-

mac); / z: douleurs en ventre.14. cwS ˙s: être grave, ardu, difficile (se dit d’un affaire, d’un événe-

ment); / z: être difficile, dur à faire (se dit d’un événementfâcheux, d’une chose).

15. frS ˙s: couper, fendre en deux; / z: séparer une chose d’avec unautre.

16. fw/yS ˙s: s’en aller et s’engager dans l’intérieur des terres; / z: s’enaller, fuir.

17. qrS ˙s: pincer quelqu’un (en serrant la chair avec le bout desdoigts); / z: voyqr ˙s.

18. qlS ˙s: sauter, faire un saut; / z: sauter, faire un saut.19. qnS ˙s: proie tuée ou prise à la chasse; / z: voyqn ˙s.20. krS ˙s: broyer, écraser le fromage aigrelet et tendre (’aqi ˙t); / z: man-

ger habituellement beacoup de fromage’aqi ˙t, en être grandamateur.

21. l ˙hS ˙s: acculer quelqu’un, le réduire à l’extremité, le mettre dansl’embarras; / z: presser quelqu’un, demander jusqu’à se rendreimportun.

22. lkS ˙s: frapper quelqu’un d’un coup de poing; / z: frapper quelqu’und’un coup de poing à la poitrine, sur les mâchoires ou sur lecorps pour le repousser; repousser.

23. mrS ˙s: serrer, presser avec les doigts (le sein, la mamelle); / z: pres-ser légerement et sans causer de doleur avec le bout des doigts.

24. mlS ˙s: échapper, s’échapper, être délivré; / z: se soustraire à quel-que chose et s’en délivrer, échapper à ...

25. ngS ˙s: troubler, gâter, p.ex. le plaisir, la jouissance ...; / z: semer,exciter, fomenter les discordes parmi les hommes.

26. w˘hS ˙s: ne donner qu’une petite quantité, que fort peu; / z: petit

nombre.27. whS ˙s: fouler avec violence; / z: fouler fortement avec les pieds le

sol.

Page 76: ZemanekPharyngealization

70 Petr ZEMÁNEK

4.4 Doublets with the alternation of˙t/t:

4.4.1 Alternation of˙t/t on the position of R1:

1. tbn ˙t: Intelligent, habile; / t: être intelligent, fin et rusé.2. t

˘hm ˙t: Voy ta

˘h um. Limites; / t: mettre une borne, une limite à quel-

qu’un.3. trr ˙t: couper, retrancher en coupant; / t: couper, retrancher.4. tr

˘hn ˙t: estragon; / t: voy˙tarh un.

5. tff ˙t: petite quantité, un peu; / t: peu, petite quantité.6. tll ˙t: être humide, légerement humecté par la pluie ou par la rosée

(se dit du sol); / t: se couvrir de moiteur.7. tltl ˙t: agiter, secouer; / t: agiter, secouer.8. tlc ˙t: gravir une montagne, monter, s’élever sur une hauter, monter

en haut; / t: hauteur, élévation, monticule.9. ˙tn

˘h ˙t: se charger l’estomac de mets gras, et en avoir une indige-

stion; / t: souffrir de l’indigestion.10. twr ˙t: tourner autour de...; / t: faire le tour, circuler au tour d’un

point.

4.4.2 Alternation of˙t/t on the position of R2:

1. ’td ˙t: II. raffermir, consolider; / t: corde avec laquelle on lie lespieds d’une vache pour qu’elle se laisse traire.

2. btt ˙t: bouteille en cuir; / t: bouteille (voyba ˙t ˙tatun)3. btr ˙t: fendre, percer, p.ex. un ulcère; / t: couper, retrancher en cou-

pant, enlever.4. ˙ht’ ˙t: cohabiter avec une femme; / t: cohabiter avec une femme.5. rtm ˙t: IV. Se taire, ne pas desserrer les dents (voy.rtm); / t: m a

ratima bi-kalimatIl n’a pas dit un seul mot.6. stm ˙t: mer; / t: mer.7. štr ˙t: partager en deux parties égales; / t: couper, disséquer.8. štc ˙t: voy štc; / t: être triste ou agité, troublé par la faim ou quelque

maladie.9. gtt ˙t: plonger quelqu’un dans l’eau la tête la première; / t: plonger

quelqu’un dans l’eau la tête la première.10. gtrf ˙t: marcher avec fierté, se donner des airs; / t: voyg ˙trf.11. ft’ ˙t: casser, briser; / t: casser.12. ftr ˙t: fendre, pourfendre, couper en deux; / t: voyf ˙tr.

Page 77: ZemanekPharyngealization

71Supplement: List of Doublets

13. qtt ˙t: couper, surtout dans le sens de la largeur; / t: couper quelquechose, rogner (dans le sens de la longueur).

14. qtr ˙t: plage, région (du ciel ou de la terre); / t: région du ciel (voyqu ˙tr).

15. ltt ˙t: serrer sa queue et la tenir, pour ainsi dire, collée contre lesfesses; / t: lier fortement, serrer.

16. lt ˙h ˙t: taper quelqu’un, lui donner un léger coup sur le dos; / t:frapper, taper quelqu’un avec la main.

17. lt˘h ˙t: salir quelqu’un, éclabousser de quelque chose; / t: salir quel-

qu’un d’ordures (voyl ˙t˘h).

18. ltm ˙t: donner à quelqu’un (sur la joue ou sur le corps) un coup duplat de la main, souffleter quelqu’un; / t: frapper quelqu’un(comp.l ˙tm).

19. mtt ˙t: tendre et allonger une chose en la tirant avec force; / t: éten-dre quelque chose en long (p.ex. une corde).

20. mt ˙h ˙t: frapper quelqu’un avec la main; / t: frapper quelqu’un.21. mt

˘h ˙t: frapper quelqu’un avec quelque chose; / t: frapper.

22. htl ˙t: qui tombe coup sur coup à grosses gouttes (pluie); / t: pluiecontinuelle, mais faible.

4.4.3 Alternation of˙t/t on the position of R3:

1. ¯tbt ˙t: IV. voy ¯tbt; / t: IV. établir solidement, fixer.2. ¯d’t ˙t: voy ¯d’t; / t: étrangler en serrant la gorge au point de faire

sentir la langue.3. zmt ˙t: voy zmt; / t: être grave et plein de dignité dans son maintien.4. šmt ˙t: qui grisonne, qui a des cheveux blancs mêlés aux noirs, ou à

moitié blancs; / t: qui commence à grisonner plus que celuimalh uz(v. šm˙t).

5. sw/yt ˙t: (y) vacarme, tintamarre, bruit, cris confus; / t: (w) pousser uncri; rendre, produire un son, un bruit, se faire entendre.

6. glt ˙t: se tromper, commettre une erreur, une faute (en parlant, enécrivant, ou dans le calcul; comp.glt); / t: se tromper, com-mettre une erreur (dans le calcul) (comp.gl ˙t, qui se dit d’unefaute comise en parlant).

7. flt ˙t: ce qui arrive tout à coup, à l’improviste; / t: survenir à l’im-proviste et fondre sur quelqu’un.

8. kct(l) ˙t: (kc ˙tl) se mettre à courir; / t: se dépêcher, aller vite.9. lwt ˙t: cacher, céler; / d: cacher, taire (une nouvelle, un message).

Page 78: ZemanekPharyngealization

72 Petr ZEMÁNEK

10. n ˙ht ˙t: respiration pénible et étouffée; / t: gémissement ou respira-tion pénible accompagnée d’un gémissement.

11. nft ˙t: bouilloner au point de lancer des gouttes d’eau, de cracher(se dit d’une marmite); / t: bouilloner au point de faire jaillir audehors les gouttes du contenu (se dit d’un pot).

12. hbt ˙t: jeter quelqu’un, précipiter en bas; / t: jeter en bas, précipiteren poussant.

13. hrt ˙t: déchirer (une pièce d’étoffe); / t: déchirer son vêtement.14. hwt ˙t: crier, faire du tapage, du vacarme; / t: appeler quelqu’un en

criant.

4.5 Doublets with the alternation ofq/k:4.5.1 Alternation ofq/kon the position of R1:

1. ktc q: être vil; / k: vil, méprisable, de nulle valeur.2. ktl q: ame; / k: ame (comp.qat al).3. k ¯tm q: ramasser, surtout des objects vils, des ordures, etc.; / k:

réunir, ramasser (ce qui était dispersé).4. k ˙h ˙h q: être pur, franc, sans mélange (se dit d’une boisson); / k: pur,

sans mélange, franc, pur-sang (voyq ˙h ˙h).5. k ˙h ˙t q: manquer, ne pas tomber (se dit de la pluie); / k: manquer (se

dit de la pluie). Voy.q ˙h ˙t.6. krkr q: rire aux éclats; / k: rire tout haut et répéter le rire (voy.qrqr).7. kzm q: petit de taille (homme); / k: court, petit, bref.8. kš˙t q: oter une chose de sa place; / k: oter de dessus une chose ce

qui lui sert de couverture, p.ex. la housse du dos du cheval.9. kcnb q: lion; / k: lion.10. kf

˘h q: frapper un corps creux, la tête, un crâne, etc.; / k: porter un

coup de bâton sur la tête de quelqu’un ou lui donner un souffletsur la joue.

11. kll q: totalité, le tout; / k: la totalité, tous.12. kld q: ramasser, recueillir (le lait, l’eau, le vin, etc.) dans un réser-

voir ou dans un vase destiné à cela; / k: entasser, accumuler,amonceler.

13. kls q: église (voy.kal ısa); / k: église.14. km ˙h q: redresser la tête et ne vouloir pas boire (se dit d’un cha-

meau); / k: tirer à soi avec la bride la tête du cheval au pointqu’il redresse la tête.

Page 79: ZemanekPharyngealization

73Supplement: List of Doublets

15. km˘h q: (’aqma

˘ha bi-anfihi) être fier, et être assis avec un air de suf-

fisance et d’orgueil; / k: (kama˘ha bi-anfihi) être fier.

16. kmz q: ramasser et prendre quelque chose du bout des doigts; / k:ramasser quelque chose avec les mains et en faire un tas arron-di.

17. knr q: gros, épais, mal tourné; / k: gros, épais et mal bâti.18. kns q: racine, origine; / k: racine, origine, tête (d’une chose).19. khb q: gris, grisâtre; / k: couler gris poudreux tirant sur le noir

(pariculière au poil des chameaux).20. kw/yc q: se reculer, se retirer; / k: reculer, s’éloigner d’une chose et

s’abstenir par peur.

4.5.2 Alternation ofq/kon the position of R2:

1. bkk q: fendre (un sac); / k: déchirer, rompre, fendre.2. bkc q: s’en aller, s’éloigner dans une contréebuqca; / k: s’éloigner,

s’en aller (le même quebqc).3. ¯tk ¯tk q: parler sottement; / k: être sot, imbécile.4. ˙hkd q: origine, racine, source; / k: origine, racine, source (comp.

ma ˙hqidetma ˙hfid).5. dkk q: casser, concasser; / k: concasser, piler, broyer (voy.dqq).6. dkm q: casser à quelqu’un les dents de devant en le frappant sur le

bouche; / k: porter à quelqu’un un coup sur la poitrine pour lerepousser.

7. rkk q: être mince, fin, délicat, point gros; / k: être très-mince(comp.rqq).

8. rkd q: être couché (sans dormir); / k: être en repos, se tenir tran-quille (comp.rqd).

9. zkk q: rendre les excréments (se dit d’un oiseau); / k: rendre lesexcréments (voy.zqq).

10. ckd q: racine, naissance de la langue; / k: racine de la langue.11. ckr q: racine, base, fondement, origine (voy.cikr); / k: racine,

origine, point de départ.12. ckš q: réunir, rassembler (son troupeau dispersé); / k: ramasser,

rassembler.13. ck ˙s q: être de mauvaise humeur ou de mauvaise caractère; / k: être

méchant, d’un mauvaise caractère.14. ckl q: entrave, particulièrem. corde avec laquelle on attache le bas

du pied du chameau, en lui pliant la jambe au haut de l’épaule;

Page 80: ZemanekPharyngealization

74 Petr ZEMÁNEK

/ k: corde avec laquelle on attache le bas du pied ployé duchameau à la partie supérieure de la jambe.

15. fkk q: disjoindre, séparer, détacher (voy.fkk); / k: séparer, disjoin-dre deux objets.

16. lkz q: donner à quelqu’un un coup de poing sur la poitrine, ou surle corps; / k: frapper quelqu’un d’un coup de poing à la poi-trine, sur les mâchoires ou sur le corps, pour le repousser;repousser.

17. nk ˙h/˘h q: (

˘h) percer à jour, trouer, forer; / k: percer, forer.

18. hkk q: cohabiter avec une femme avec violence; / k: abîmer unefemme à force de cohabiter fréquemment avec elle.

19. wkc q: frapper, affliger quelqu’un par des maux et l’éprouver; / k:frapper, donner les coups dans le pis de la brebis qu’on trait,pour en faire sortir plus de lait.

20. wkn q: nid (d’un oiseau); / k: nid (d’oiseau).

4.5.3 Alternation ofq/kon the position of R3:

1. ’fk q: mentir; / k: mentir, forger un mensonge.2. ˙hzk q: serrer, garrotter (avec une corde); /k: serrer, presser, compri-

mer.3. zmk q: s’emporter, se mettre en colère contre quelqu’un; / k: irriter,

mettre quelqu’un dans une colère violente contre un autre.4. smk q: être haut, grand, d’une belle taille (se dit des plantes, des ar-

bres); / k: être très-haut, grand.5. ˙swk q: V. se salir de ses propres excréments. Voy˙swkV.; / k: se

salir de ses propres excréments ou de son sperme.6. ctk q: être ancien, antique, vieux; / k: être vieux et rougeâtre de

vicillesse (se dit du bois d’un arc).7. cfk q: cafaqa-hucan ’l-amr Il l’empêcha de se livrer à telle et telle

occupation, il l’en retint; / k: empêcher quelqu’un de fairequelque chose, de se livrer à son gré à quelque chose.

8. l ˙hk q: s’attacher à ... et en être inséparable; / k: se coller fortementet tenir à un corps.

9. ntk q: tirer, retirer (p.ex. le seau du puits); / k: tirer violemmentquelque chose à soi, au point de casser.

10. wšk q: se dépêcher, se hâter; / k: aller vite, s’empresser de ...

Page 81: ZemanekPharyngealization

75Supplement: List of Doublets

4.6 Doublets with the alternation of˙s/s:

4.6.1 Alternation of ˙s/s on the position of R1:

1. s’b ˙s: être rempli de boisson, en avoir bu jusqu’à satiété, être gorgéde ... / s: se gorger de quelque chose, de la boisson.

2. sbg ˙s: être long (se dit du tibia); / s: être long et descendre en basjusqu’à traîner par terre (se dit d’une robe, des cheveux, etc.).

3. s˘hb ˙s: crier, vociférer; / s: cri. voy˙s

˘hb.

4. s˘hd ˙s: brûler quelqu’un, l’affecter par l’intensité de la chaleur; / s:

chaud.5. sdd ˙s: montagne; / s: montagne.6. sdg ˙s: coussin, oreiler; / s: voy˙sdg.7. sr ˙t ˙s: chemin, route, sentier. On l’écrit aussi˙sir a¯t; / s: chemin.8. srm ˙s: couper, retrancher en coupant; / s: II. couper en morceaux.9. s ˙tb ˙s: Hospice. On l’écrit aussimas˙taba; / s: voyma ˙s ˙taba.10. s ˙tr ˙s: voys ˙tr; / s: écrire.11. sctr ˙s: espèce d’origan ou de thym (voyzactar); / s: thym (thymus

serpyllum).12. sc ˙t ˙s: voysc ˙t; / s: injecter un médicament dans le nez.13. sgb ˙s: Voysgb; / s: être épuisé de fatigue, de faim et de douleur.14. sgl ˙s: voysgl; / s: être petit de taille et avoir des jambes minces, un

corps chétif comme un homme qui est mal nourri et faible.15. sfc ˙s: donner un soufflet, souffleter; / s: souffleter quelqu’un, lui

donner un soufflet, des soufflets.16. sq ˙h ˙s: chauve; / s: chauve.17. sqr ˙s: faucon; / s: faucon (voy˙sqr).18. sqc ˙s: éloquent (orateur); / s: éloquent (orateur, predicateur).19. sql ˙s: polir, fourbir; / s: polir, fourbir (une lame, un corps métali-

que) (voy ˙sql).20. sqlb ˙s: Slaves; / s: Slaves, peuples slaves. Voy˙sqlb.21. slt ˙s: frapper avec le sabre, donner un coup de sabre; / s: couper,

abattre (le nez) avec un coup de sabre, couper, retrancher.22. sl ˙t ˙s: voysl ˙t; / s: être absolu et dur dans le commandement.23. slg ˙s: voyslg; / s: avoir une dent incisive.24. slq ˙s: produire un grand bruit, un fracas, pousser un grand cri; / s:

crier, pousser un cri.25. slqm ˙s: claquer, grincer des dents; / s: faire claquer des dents.26. smm ˙s: boucher (un flacon, etc.); / s: boucher (un flacon, etc.).27. sn˙t ˙s: voy sn˙t; / s: n’avoir pas de barbe, ou n’avoir de la barbe

qu’on menton.

Page 82: ZemanekPharyngealization

76 Petr ZEMÁNEK

28. sw/y˘h ˙s: s’embourber, ne pas pouvoir se tirer d’un bourbier; / s: tom-

ber dans un bourbier et y rester embourbé.29. sw˙t ˙s: voysw˙t; / s: fouet de nerf de boeuf ou de courroises tressées,

cravache.30. sw/yq ˙s: voyswq; / s: pousser devant soi, mener, stimuler à la marche

(une bête de somme, le troupeau).

4.6.2 Alternation of ˙s/son the position of R2:

1. ’sr ˙s: lier, attacher avec la corde; / s: lier, serrer avec des liens,avec une corde, etc.

2. bsq ˙s: même signif. quebzq. Cracher; / s: cracher.3. ˙hsk ˙s: chandelier, flambeau; / s: chandelier, flambeau.4.

˘hsm ˙s: voy ’u

˘hs umsous

˘hsm; / s: anse d’une sacoche.

5. rs˘h ˙s: voyrs

˘h; / s: être ferme, solide, être planté, établi solidement.

6. rsg ˙s: voyrsg; / s: procurer à quelqu’un une vie aisée, comode, dubien-être.

7. rsy ˙s: IV. rester à sa place, n’en pas bouger, être ancré; / s: IV. êtreferme, immobile, être ancré quelque part.

8. csb ˙s: chefs ou principaux personnages d’une tribu, d’une commu-nauté; / s: chef d’une tribu, d’une communauté; le premierparmi les siens.

9. fs˘h ˙s: démettre un membre du corps (voyfs

˘h); / s: disloquer, dé-

mettre un membre du corps, p.ex. le bras, causer une luxation.10. fs ˙t ˙s: ordure des ongles; / s: rognure d’ongle.12. fsl ˙s: jeune arbre transplanté ou rameau enlevé de la tige dùn arbre

pour être planté ailleurs (voyfsl); / s: rejeton de palmier oujeune rameau de palmier détaché de la tige et planté.

13. qss ˙s: suivre quelqu’un pas à pas; être aux trousses de quelqu’un;/ s: chercher, rechercher, poursuivre quelque chose avec persi-stance (comp.q ˙s ˙s).

14. q(w)sr ˙s: panier en feuilles de palmier ou en osier dans lequel on con-serve les dattes; / s: voyqaw ˙sara.

15. qsqs ˙s: appeler à soi un chien. Voyqsqs; / s: appeler un chien endisantq us q us.

16. qsm ˙s: morceau, fragment; / s: partie, portion (d’une chose divisée).17. lsq ˙s: être collé, se coller, s’agglutiner à quelque chose; / s: se

coller, s’attacher à ...

Page 83: ZemanekPharyngealization

77Supplement: List of Doublets

18. ms ˙h ˙s: (ma ˙sa ˙ha ’ll ahu mara˙daka) que Dieu t’ôte ta maladie!; / s:(masa˙ha ’ll ahu m a bika mincilla ) que Dieu ôte la maladie donttu es atteint.

19. ms˘h ˙s: voy ms

˘h; / s: transformer, changer, métamorphoser quel-

qu’un en quelque chose.20. ms˙t ˙s: voyms˙t; / s: serrer avec les doigts l’orifice d’une outre ou

presser avec la main les boyaux, pour en faire sortir peu à peuquelque chose.

21. hss ˙s: casser, briser; / s: casser, briser en petits morceaux.22. hsm ˙s: casser, briser; / s: casser, fracasser.

4.6.3 Alternation of ˙s/son the position of R3:

1. bcs ˙s: maigreur; / s: chamelle amaigrie et qui n’a plus de lait.2.

˘hrs ˙s: nourriture d’accouchée. Voy sous

˘hrs; / s: accouchée à qui

on donne la nourriture qui convient à son état.3. d ˙hs ˙s: agiter les pieds dans des convulsions, et en remuer la terre,

gigotter (se dit d’un animal qu’on vient d’égorger (comp.d ˙hs);/ s: chercher, fouiller en remouant la terre avec le pied.

4. rfs ˙s: pousser, frapper avec le pied (voyrfs); / s: frapper, pousserdu pied, ruer.

5. šrs ˙s: tirer, traîner; / s: tirer à soi par la bride (sa monture).6. šks ˙s: voyšakis; / s: difficile à vivre.7. crs ˙s: entremetteur. Voymucarris; / s: entremetteur, maquereau.8. cms ˙s: voycam as; / s: malheur, calamité.9. fqs ˙s: casser (un œuf); / s: casser, détruire (se dit d’un oiseau qui

détruit ses œufs).10. qbs ˙s: race, origin (voyqbs); / s: origine, race.11. qrs ˙s: ortie, graine de l’ortie; / s: ortie, et graine d’ortie. Voyqar ı˙s.12. qfs ˙s: lier aux pieds (une gazelle); attacher avec un fil dans la ruche

la reine des abeilles pour l’empêcher de s’envoler; / s: lier lespieds (d’une gazelle), mettre des entraves aux pieds de ...

13. qns ˙s: pontre qui soutient le toit; / s: voyq ani˙sa.14. lqs ˙s: être affecté péniblement, éprouver une vive douleur, une an-

goisse; / s: être affecté, éprouver du chagrin par suite de quel-que chose.

15. lw/ys ˙s: sorte de friandise, appelée autrementf al u¯d; / s: qui aime lesfriandises, les douceurs.

16. mgs ˙s: éprouver des douleurs dans les entrailles; / s: voymg ˙s.

Page 84: ZemanekPharyngealization

78 Petr ZEMÁNEK

17. nbs ˙s: parler. Voynbs; / s: parler, surtout avec volubilité et avecdes gesticulations.

18. hw/ys ˙s: casser, rompre, briser, p.ex. le cou à quelqu’un; / s: concas-ser; briser.

19. wqs ˙s: voy. ’awq as; / s: la queue d’une troupe ou d’une tribu,comme les valets et les esclaves.

20. whs ˙s: fouler avec violence. Voy.whs; / s: fouler aux pieds, avec lespieds.

4.7 Doublets with the alternation ofb/r:

4.7.1 Alternation ofb/r on the position of R1:

1. b/rw˘h b: 1. être las, fatigué - r: 3. être las, fatigué.

2. b/rc ˙s b: se remuer, s’agiter en mouvements convulsifs - r: 1. Agiter,remuer, secouer.

3. b/rdd b: 3. Éloigner, renvoyer quelqu’un et le tenir dans l’éloigne-ment -r: 1. Repousser, éloigner, écarter.

4. b/r˘hw/y b: se calmer, s’apaiser (se dit de la colère) - r: 1. Relaxation,

relâchement (de ce qui n’est plus tendu).

4.7.2 Alternation ofb/r on the position of R2:

1. gbb/rr b: 1. Couper, retrancher en coupant; / r: 6. Fendre la langue àun petit chameau (pour l’empêcher de téter sa mère, en voulantle sevrer).

2. gb/rz b: Couper, séparer une partie du reste; / r: 1. Couper, retran-cher.

3. gb/rw/y b: 2. Ramasser l’eau, la faire couler dans les citernes; / b: Cou-ler (se dit de l’eau, du sang, des larmes).

3. db/rdb/r b: Tambour; / r: 1. Bruit du tambour.4. sbb/rr b: 1. Couper; / r: Couper à l’enfant le cordon ombilical par

lequel il tient à sa mère.5. sb/r ˙h b: 4. Être libre, exempt de quelque chose; / r: 1. Laisser (le

troupeau) aller librement et lui permettre de paître où il veut.6. šb/r ˙h b: Fendre; / r: couper, disséquer, fendre.7. ˙sb/r ˙h b: Clair, évident, manifeste (vérité, etc.); / r: Rendre clair,

exposer clairement, rendre évident.8. ˙db/r’ b: IV. 1. Cacher; / r: Être caché, se cacher.9. ˙db/rg b: Se jeter par terre et se rouler dans la poussière; / r: 3. Jeter,

lancer.

Page 85: ZemanekPharyngealization

79Supplement: List of Doublets

10. ˙tb/rs b: Noir (se dit de toute chose); / r: II. Peindre en noir, noircir(la porte au signe de deuil, etc.).

11. qb/r ˙s b: 1. Prendre quelque chose avec le bout des doigts; prendreune pincée; / r: 1. pincer quelqu’un (en serrant la chair avec lebout des doigts).

12. hb/rz b: Mourir d’une mort subite; / r: Mourir, périr.

4.7.3 Alternation ofb/r on the position of R3:

1. ¯tcb/r b: Qui coule, liquide; / r: Suc vénéneux que distille l’arbresamr.

2. gzb/r b: Portion, lot; / r: Couper (les grappes des dattes mûres pour larécolte) (syn.gazza).

3. ˙hnb/r b: Construire une voûte’zg; / r: construire une voûte, une con-struction voûtée.

4.˘h ˙db/r b: Être vert, verdoyant (se dit des arbres), avoir des plantes

vertes, des rameaux verts (se dit de la terre, des arbres qui secouvrent de verdures); / r: Être vert, verdoyant et d’un aspectriant (se dit des plantes, des céréales).

5. d ˙hb/r b: Repousser, éloigner, donner la chasse; / r: Éloigner, écarter,chasser.

6. š ˙tb/r b: Couper en deux, pourfendre; / r: Partager en deux partieségales.

7. ˙sqb/r b: Taper, frapper quelqu’un de la paume de la main; / r: Frap-per quelqu’un d’un bâton, ...

8. q ˙hb/r b: Agé; / r: Très-agé (homme ou chameau), mais qui conserveencore quelques forces.

9. k ¯tb/r b: Nombreux, abondant; / r: Être nombreux, être en grandnombre.

10. ngb/r b: Enlever l’écorce d’un arbre; / r: Raboter le bois avec unrabot (se dit du mennisier).

11. ngb/r b: 1. Avaler, p.ex. la salive; 2. Boire en humant (se dit des oise-aux); / r: Boire beacoup, avaler beacoup d’eau, etc.

4.8 Doublets with the alternation of¯t/g:

4.8.1 Alternation of¯t/g on the position of R1:

1. ¯t/gdq ¯t: Tomber en abondance, verser ses eaux (se dit de la pluie,d’un nuage, d’un torrent); / g: Se répandre abondamment.

Page 86: ZemanekPharyngealization

80 Petr ZEMÁNEK

2. ¯t/gmd ¯t: Petite quantité d’eau, et qui, pour n’être pas alimentée, tarit;/ g: Avoir fort peau d’eau.

3. ¯t/gmr ¯t: Riche, opulent; / g: Richesse, biens immenses.

4.8.2 Alternation of¯t/g on the position of R2:

No doublets found.

4.8.3 Alternation of¯t/g on the position of R3:

No doublets found.

Page 87: ZemanekPharyngealization

81References

REFERENCES

ABD-EL-JAWAD R., 1986: The emergence of an urban dialect in the Jordanian urban cen-tres. International Journal of the Sociology of Languages 61, 53-63.

AEŠCOLYA.Z., 1939: Incompatibilités de phonèmes en hébreu et en araméen. GLECS 3,54-56.

AL-ANI S.H., 1970: Arabic Phonology. The Hague, Paris.ALI L. & D ANILOFF R.G., 1972: A Contrastive Cinefluorographic Investigation of the

Articulation of Emphatic - Nonemphatic Cognate Consonants. Studia Linguistica26, 81-105.

AMBROSA.A., 1981: Zur Entstehung der Emphase inALLAH. WZKM 73, 23-32.APPLEGATEJ.R., 1970: The Berber Languages. In: SEBEOK T.A.(ed.), Current Trends in

Linguistics. Vol. 6, Linguistics in South West Asia and North Africa. The Hague,Paris, 586-661.

ARO J., 1977: Pronunciation of the "Emphatic" Consonants in Semitic Languages. StudiaOrientalia 47, 5-18.

BALDI S., 1987(preprint): Some Remarks on Arabic Loan Words in Hausa. Paper read at5th International Hamito-Semitic Congress, Wien, 27.9.-2.10.1987.

BANI-YASIN R. & OWENSJ., 1987:The Phonology of a Northern Jordanian Arabic Dialect.ZDMG 137, 297-331.

BEESTONF.L., 1981: Languages of the Pre-Islamic Arabia. Arabica 28, 178-186.BEHNSTEDTP., 1986: Die Dialekte der Gegend von˙Sacdah. Wiesbaden.BENDERM.L., 1975: Omotic: A New Afroasiatic Language Family. Carbondale.BLAKE F.R., 1946: Studies in Semitic Grammar IV. JAOS 66, 212-215.BOMHARD A.R., 1984: Toward Proto-Nostratic: A New Approach to the Comparison of

Proto-Indo-European and Proto-Afroasiatic. Amsterdam.BOMHARD A.R., 1988: The Reconstruction of the Proto-Semitic Consonant System. In:

ARBEITMAN Y.L. (ed.), Fucus. A Semitic/Afrasian gathering in Remembrance ofAlbert Ehrman. Amsterdam - Philadelphia, 113-140.

BROCKELMANN C., 1908: Grundriss der vergleichenden Grammatik der semitischen Spra-chen. I. Band: Lautlehre und Formenlehre. Berlin.

CANTINEAU J., 1960: Études de linguistique arabe. Mémorial Jean Cantineau. Paris.CARDONA G.R., 1968: Per la storia fonologica del "˙sad e" semitico. AION 18, 1-14.CLAASSEN W.T., 1971: The Role of /˙s/ in the North-West Semitic Languages. AION 31,

285-302.COHEN D., 1963: Le dialecte arabe˙hass an ıya de Mauritanie. Paris.COHEN D., 1969: Sur le statut phonologique de l’emphase en arabe. Word 25, 59-69.COHEND., 1970: Le système phonologique du maltais. Aspects synchroniques et diachro-

niques. In: Études de linguistique sémitique et arabe. The Hague 126-149.COHEN M., 1931: Études d’éthiopien méridionel. Paris.COLIN 1930: Notes de dialectologie arabe. Hesperis 10, 91-120.

Page 88: ZemanekPharyngealization

82 Petr ZEMÁNEK

COLIN 1939: Incompatibilités consonantiques dans les racines de l’arabe classique. GLECS24, 61-2.

CONTI G., 1976: Il sistema consonantico egiziano. OA 15, 44-55.CONTI G., 1990: Il sillabario della quarta fonte della lista lessicale bilingua eblaita. Mis-

cellanea eblaita 3, Firenze.CONTI G., 1993: Il sistema grafico eblaita e la legge di Geers. Quaderni del Dipartimento

di Linguistica - Università di Firenze 4, 97-114.DELATTRE P., 1971: Pharyngeal Features in the Consonants of Arabic, German, Spanish,

French, and American English. Phonetica 23, 129-155.DIEM W., 1972: Skizzen jemenitischer Dialekte. Beirut.DIAKONOFF I.M., 1984: An evaluation of Eblaite. In: FRONZAROLI P. (ed.), Studies on the

Language of Ebla. Firenze, 1-10.DIAKONOFF 1987: = D’YAKONOV I.M. - M ILITAREV A.Y.- PORCHOMOVSKY V.Y. - STOL-

BOVA O.V.: Obšceafraziyskaya fonologiceskaya sistema. In: Afrikanskoye isto-riceskoye yazykoznaniye. Problemy rekonstrukcii. Moskva 1987, 9-29.

DIAKONOFF I.M., 1988: Afrasian languages. Moscow.DIAKONOFF I.M., 1991-92: Proto-Afrasian and Old Akkadian. A study in historical phone-

tics. Princeton [Journal of Afroasiatic Languages 4, nr. 1-2].DOLGOPOLSKY A.B., 1973: Sravnitel’no-istoriceskaya fonetika kušitskich yazykov.

Moskva 1973.DOLGOPOLSKYA.B., 1977: Emphatic consonants in Semitic. Israel Oriental Studies 7, 1-

13.DOLGOPOLSKYA.B., 1983: Semitic and Cushitic sound correspondences. In: SEGERTS. -

BODROGLIGETI A.J. (ed.), Ethiopian Studies dedicated to W. Leslau. Wiesbaden,123-142.

EHRETC., 1980: The Historical Reconstruction of Southern Cushitic Phonology and Voca-bulary. Berlin.

EHRETC., 1989: The origin of third consonants in Semitic roots: an internal reconstruction(applied to Arabic). Journal of Afroasiatic Languages 2, 109-202.

ESSENO. von, 1979: Allgemeine und angewandte Phonetik. Berlin. (5. Auflage).FEGHALI M.T., 1919: Le parler de KfarcAbîda (Liban-Syrie). Paris.FERGUSONC., 1956: The Emphatic L in Arabic. Language, 32, 446-452.FISCHER W., 1968: Die Position von˙d im Phonemsystem des Gemeinsemitischen. In:

FLEISCHHAMMER M., (ed.), Studia Orientalia in memoriam Caroli Brockelmann,Halle, 55-63.

FRE WOLDU K., 1984-86: Evidence of Auditory Similarity between Tigrinya Ejective /˙t/and Arabic Emphatic /¯t/. Orientalia Suecana 33-35, 123-138.

FRE WOLDU K., 1988: Phonetics and Historical Relationships in Semitic. A study ofEjective and Emphatic Consonants. In: BEYENE T., (ed.), Proceedings of the 8thInternational Conference of Ethiopian Studies. Vol. I. Addis Ababa, Frankfurt, 705-714.

Page 89: ZemanekPharyngealization

83References

GAIRDNER W.H.T., 1925: The Phonetics of Arabic. A Phonetic inquiry and practicalmanual for the pronunciation of Classical Arabic and of one Colloquial (the Egyp-tian). London.

GARBINI G., 1972: Le lingue semitiche. Napoli.GARBINI G., 1984: Le lingue semitiche. (2nd edition) Napoli.GARR W. R., 1986: On Voicing and Devoicing in Ugaritic. JNES 45, 45- 52.GEERSF.W., 1945: The treatment of emphatics in Akkadian. JNES 42, 65- 67.GELB I.J., 1961: Old Akkadian Writing and Grammar. Chicago.GHALI M.M., 1983: Pharyngeal articulation. BSOAS 46, 432-444.GHAZELI S., 1981: La coarticulation de l’emphase en arabe. Arabica 28, 251-277.GIANNINI A. & PETTORINOM., 1982: The Emphatic Consonants in Arabic. Speech labora-

tory Report IV, Istituto Universitario Orientale di Napoli. Napoli.GREENBERGJ. H., 1950: The Patterning of Root Morphemes in Semitic. Word 6, 162-181.GVOZDANOVIC J., 1985: Language System and Its Change. On Theory and Testability.

Berlin - New York - Amsterdam.HARRELL R., 1957: The Phonology of Egyptian Colloquial Arabic. New York.HARRELL R., 1962: A Linguistic analysis of Egyptian Radio Arabic. In: FERGUSONC.,

(ed.), Contributions to Arabic Linguistics. Cambridge, 1-77.HARRELL R., 1965: A Short Reference Grammar of Morrocan Arabic. Washington.HARRIS Z.S., 1942: The Phonemes of Morrocan Arabic. JAOS 62, 309-318.HEINE B., 1982: The Nubi Language of Kibera - an Arabic Creole. (Language and Dialect

Atlas of Kenya, Vol. 3). Berlin.HERDAN G., 1962: The Calculus of Linguistic Observation. London.HOBERMAN R.D., 1988: Emphasis Harmony in a Modern Aramaic Dialect. Language 64,

1-26.HOBERMAN R.D., 1989: Parameters of emphasis: autosegmental analyses of pharyn-

gealization in four languages. Journal of Afroasiatic Languages Vol. 2, 73-97.HODGEC.T., 1983: Afroasiatic: The Horizon and Beyond. JQR 74/2, 137- 158.HODGE C.T., 1987(preprint): Etymological Reassessment. Paper read at the 5th Inter-

national Hamito-Semitic Congress, Wien 27.9.-2.10.1987.JAKOBSON R., 1957: "Mufaxxama": The Emphatic Phonemes in Arabic. In: PULGRAM E.

(ed.), Studies Presented to Joshua Whatmough... The Hague, 105-116.JUCQUOISG., 1966: Phonétique comparée des dialectes moyen-babyloniens du nord et de

l’ouest. Louvain.JUNGRAITHMAYR H. & SHIMIZU K., 1981: Chadic Lexical Roots (A First Evaluation of the

Marburg Chadic Word Catalogue). Vol. 2: Tentative Reconstruction, Grading andDistribution. Berlin.

KÄSTNER1981: Phonetik und Phonologie des modernen Hocharabisch. Leipzig.KAZIMIRSKI A.B., 1860: Dictionnaire Arabe-Français. Le Caire.KHALAFALLAH A.G., 1969: A Descriptive Grammar of Sacidi Egyptian Colloquial Arabic.

The Hague, Mouton.

Page 90: ZemanekPharyngealization

84 Petr ZEMÁNEK

KNUDSENE.E., 1961: Cases of Free Variants in the Akkadianq Phoneme. JCS 15, 84-90.KNUDSEN E.E., 1969: Spirantization of Velars in Akkadian. In: ROLLIG W. (ed. with the

collaboration of DIETRICH M.), Liš an mit˘hurti. Festschrift Wolfram Freiherr von

Soden... gewidmet... Neukirchen-Vluyn, 147-155.KREBERNIKM., 1982: Zu Syllabar und orthographie der lexikalischen Texte aus Ebla. Teil

1. ZA 72, 178-236.KREBERNIKM., 1983: Zu Syllabar und Orthographie der lexikalischen Texte aus Ebla. Teil

2 (Glossar). ZA 73, 1-47.KTU = DIETRICH M. & L ORETZO. & SANMARTÍN J., 1976: Die keilalphabetischen Texte

aus Ugarit. Teil 1 Transkription. Neukirchen-Vluyn.KURYŁOWICZ J., 1973: Studies in Semitic Grammar and Metrics. Wroclaw, London.LESLAU W., 1957: The Semitic Phonetic System. In: KAISER L., (ed.), Manual of Phone-

tics. Amsterdam, 325-329.LADEFOGED1964: A phonetic study of West African Languages. Cambridge.LOPRIENOA., 1977: A proposito delle consonanti dentali e velari in egiziano ed in semi-

tico. AION 37 (N.S. 28), 125-142.MAIZEL ’ S.S., 1983: Puti razvitiya kornevogo fonda semitskich yazykov. Moskva.MALAIKA N., 1959: Grundzüge der Grammatik des arabischen Dialektes von Bagdad.

Wiesbaden.MARRASSINI P., 1976: Semitico e camitosemitico nella linguistica etiopica. OA 15, 333-

344.MARTINET A., 1953: Remarques sur le consonantisme sémitique. Paris.MERLINGEN 1978: Zur Problematik der sogenannten Implosive. Phonetica 36, 241-238.MITCHELL T.F., 1956: An Introduction to Egyptian Colloquial Arabic. Oxford.MOSCATI ET AL. 1964: = MOSCATIS. - SPITALER S. - ULLENDORFFE. - SODENW. von, An

Introduction to the Comparative Grammar of the Semitic Languages. Wiesbaden.NAUMKIN V.V. & PORCHOMOVSKIJV.Ja., 1981: Ocerki po etnolingvistike Sokotry. Mosk-

va.NEWMAN P., 1977: Chadic Classification and Reconstructions. Afro-Asiatic Linguistics, 1-

42.OBRECHTD., 1968: Effects of the 2nd Formant on the Perception of Velarized Consonants

in Arabic. The Hague, Mouton, Paris.OWENSJ., 1985: The Origins of East African Nubi. Anthropological Linguistics, Fall 1985,

229-271.PETRÁCEKK.,1952:ZurArtikulationdessogenanntenemphatischenL inArabischen.ArOr

20, 509-523.PETRÁCEKK., 1955: Die Struktur des semitischen Wurzelmorpheme und der Übergangcain

> gain undcain > r im Arabischen. ArOr 23, 475- 478.PETRÁCEKK.,1971:Die innereStrukturationdesphonologischenSystemsderSchriftarabi-

schen. Travaux Linguistiques de Prague 4, 37-40.

Page 91: ZemanekPharyngealization

85References

PETRÁCEKK., 1975: Le dynamisme du système phonologique protosémitique et les problè-mes de la phonologie chamito-sémitique. In: BYNON J. & TH. (eds.), Hamito-Semi-tica. Proceedings of a Colloquium Held by the Historical Section of the LinguisticAssociation (Great Britain) at the School of Oriental and African Studies, Univer-sity of London, 18-20th March 1970. The Hague, 161-165.

PETRÁCEK K., 1981: Le système de l’arabe dans une perspective diachronique. Arabica 28,162-177.

PETRÁCEKK., 1987: Sur le rôle des modalités sonantiques dans l’élaboration de la racine ensémitique. Arabica 34, 106-110.

PETRÁCEK K., 1988: Altägyptisch, Hamitosemitisch und ihre Beziehungen zu einigenSprachfamilien in Afrika und Asien. Praha.

PETRÁCEKK., 1990: Úvod do hamitosemitské (afroasijské) jazykovedy. I., II. [Introductionto Hamito-Semitic (Afro-Asiatic) Linguistics]. Praha.

PRASSEK.-G., 1972: Manuel de grammaire touaregue (Tahaggart). I-III. Copenhagen.RABIN C., 1952: Ancient West Arabian. London.RODINSON M., 1981: Les nouvelles inscriptions d’Axoum et le lieu de déportation des

Bedjas. Rayd an. Journal of Ancient Yemeni Antiquities and Epigraphy. 4, 97-116.ROMAN A., 1977: Les zones d’articulation d’aprés Al-Chalîl. Arabica 24, 139-144.ROMAN A., 1981: De la langue arabe comme un modèle général de la formation des lan-

gues sémitiques et de leur évolution. Arabica 28, 127-161.ROMAN A., 1981-2: Le rôle des seuills de perception temporelle dans la constitution des

contoïdes glottalisées. Travaux de l’Institut de phonètique d’Aix 8, 205-226.ROMAN A., 1990: Grammaire arabe. Paris.RÖSSLERO., 1971: Das Ägyptisch als semitische Sprache. In: ALTHEIM F. - STIEHL R.,

(eds.), Christentum am Roten Meer. I., Berlin - New York, 263-326.ROTHA., 1979: Esquisse grammaticale du parler arabe d’Abbèché (Tchad). GLECS Suppl.

8, Atlas linguistique du monde arabe (publiée sous la direction de David Cohen).Paris.

ROTH-LALY A., 1972: Esquisse de la phonologie du parler arabe d’Abbéché. GLECS 16,33-79.

RUŽICKA R., 1954: La question de l’existence du g dans les langues sémitiques en généralet dans la langue ougaritienne en particulier. ArOr 22, 176-237.

SCHAADE A., 1911: S ıbawaihi’s Lautlehre. Leiden.SEGERTS., 1984: A Basic Grammar of the Ugaritic Language. Berkeley and Los Angeles.SEMAAN K.I., 1968: Linguistics in the Middle Ages. Phonetic Studies in Early Islam.

Leiden.SIBAWAIH 1889: Le livre de Sibawaihi. Traité de grammaire arabe par... Tome second.

Paris.SODEN W. von, 1952: Grundriss der akkadischen Grammatik. Roma 1952.STEINERR., 1977: The Case for Fricative Laterals in Proto-Semitic. New Haven [American

Oriental Society, 59].

Page 92: ZemanekPharyngealization

86 Petr ZEMÁNEK

STEINER R., 1982: Affricated Sade in the Semitic Languages. New York.STOLBOVA O., 1978: Sravnitel’no-istoriceskoye issledovaniye fonetiki zapadocadskich

yazykov. Avtoreferat kandidatskoj dissertacii. Moskva.STOLBOVA O., 1986: Rekonstrukciya konsonantnoy sistemy zapadocadskich yazykov. In:

Pismennyye pamyatniki i problemy istorii i kul’tury narodov Vostoka. Moskva1980 [1986], 80-115.

SWIGGERSP., 1981: A phonological analysis of˙hars usi consonants. Arabica 28, 358-361.TRUBETZKOY N.S., 1939: Grundzüge der Phonologie. Prag.TSERETELIG.V., 1956: Arabskie dialekty sredney Azii. Tom 1. Tbilisi.TSERETELIK., 1978: Grammatik der modernen assyrischen Sprache (Neu-ostaramäisch).

Leipzig.TSERETELIK., 1982: On one suprasegmental Phoneme in Modern Semitic. JAOS 102, 343-

346.VACHEK J., 1968: Dynamika fonologického systému soucasné spisovné ceštiny. [The

Dynamics of the Phonemic System of Contemporary Czech]. Praha.VILENCIK J., 1930: Welchen Lautwert hatte˙d im Ursemitischen? Orientalische Literatur-

zeitung 33, 89-98.VOIGT R.M., 1979: Die Laterale im Semitischen. Die Welt des Orients 10, 93-114.VOIGT R.M., 1986: A note on the Alleged Middle/Neo-Assyrian Sound Changes’(*š’) >

ss< ˙s>. JNES 45, 53-57.VOIGT R.M., 1988: Die infirmen Verbaltypen des Arabischen und das Biradikalismus-

Problem. Stuttgart.VOIGTR.M., 1989: The Development of the Old Ethiopic Consonantal System. In: BEYENE

T., (ed.), Proceedings of the 8th International Conference of Ethiopian Studies. Vol.2, Addis Ababa, Frankfurt, 633-647.

VOIGT R.M., 1991: On voicing and devoicing in Ugaritic. In: KAYE A.S., (ed.), SemiticStudies. In honor of Wolf Leslau... Wiesbaden, 1619-1619.

WEDEKIND K., 1990: Glottalization Constraints and Ethiopian Counter-Evidence. FoliaLinguistica 24, 127-137.

WEHRH., 1980: A Dictionary of Modern Written Arabic. (Third Printing). Beirut, London.WILLMS A., 1972: Einführung in das Vulgärarabische von Nordwestafrika. Leiden.WOLFF E., 1981: Die Berbersprachen. In: Die Sprachen Afrikas. Band 2. Afroasiatisch.

Hamburg, 171-186.ZAVADOVSKY Ju.N., 1962: Arabskie dialekty Magriba. Moskva.ZAVADOVSKY Ju. N., 1981: Mavritanskiy dialekt arabskogo yazyka (xassaniya). Moskva.ZELTNERJ.-C. & TOURNEUXH., 1986: L’arabe dans le bassin du Tchad. Le parler des Ulâd

Eli. Paris.ZEMÁNEK P., 1990: A propos de la pharyngealisation et de la glottalisation en arabe. ArOr

58, 125-134.