ARABIC VOWEL FORMANT FREQUENCIES
                                                                                            Mohammad Al-Anani
                                                                                Department of Phonetics & Linguistics
                                                                                             University of Jordan
                                                      ABSTRACT                                                                  Α                   ∆                 Α∆
                                                                                                                'decided': /b      T/ 'ducks', /fa / 'genius': /f         / 'rough person' /a:/
                     For most Arabic varieties, the articulatory and acoustic                                   : /o:/ /la:n/ 'softened': /lo:n/ 'colour', /ma:t/ 'died': /mo:t/ 'death',
                     characteristics of Arabic vowels have so far received little                                Α               Α                                           Α
                                                                                                                /   :/ : /o:/, /q  :s/ 'measured' : /qo:s/ 'arrow', /T          :r/ 'flew': /To:r/
                     attention. For example, the precise number of vocalic units                                            ΦΑ                  Φ
                     operating in the phonological systems in most varieties of                                 'phase', /       :S/ 'dived': /    o:S/ 'diving', /i:/ : /u:l, /fi:l/ 'elephant':
                                                                                                                                    ℵ                 ℵ
                     spoken Arabic is still unknown. Important acoustic knowledge                               /fu:l/ 'beans', /     i:d/ 'feast', /    u:d/ 'match', /i/: /a/, /sin/ 'tooth':
                                                                                                                                                                                 Α             ΑΣ
                     about the formant pattern of vocalic units, especially acoustic                            /san/ 'legislated' /min/ 'from', :/man/ 'who', / /: /i/, /r                        /
                                                                                                                                 Σ               Α
                     data relevant for the first two formants is not available.  The                            'sprayed': /ru / 'spray', /r       d/ 'an answer', /rud/ 'give an answer'.
                     purpose of this study has therefore been targeted to collection                                The minimal pairs were incorporated in the carrier phrase
                                                                                                                "la:…… wala: ….."
                     of acoustic data that would provide reliable evidence for                                                              judged to have the least effect on vowel
                     scholars and practitioners when describing and comparing                                   quality as it ties and "rhymes" with the following test-words
                     formant frequencies of vocalic units.  The material of                                     and marked by high  frequency of occurrence in spoken
                     investigation consisted of 64 contrastive CV (V) C words                                   Arabic. The words themselves were selected so that they
                     which represent instances containing each of the ten Arabic                                provide clear evidence of steady state positions making it
                     monophthongs. Paired words were read out onto a bi-                                        possible for identification and extraction of formant values.
                     directional microphone connected to Micron computer. The                                   3.2.  Informants.
                     software computer programme used was "Dr Speech, Tiger                                                                Three female and three male first year
                     Electronics, Inc.".                                                                        students from the University of Jordan, (19-20) years old,
                     Formant frequencies (Fo, F1, F2, F3) were measued by LPC                                   informally assessed as having normal speaking and hearing
                     analysis.                                                                                  ability participated in the experiment.
                                                                                                                3.3 Recordings.                                                             laa….
                                                1.  INTRODUCTION                                                                         Paired words in the carrier phrase 
                     Most research workers who dealt with the articulatory and                                  wala…..were read out onto a bi-directional 'Sony' microphone
                     acoustic characteristics of Arabic vowels have focussed                                    connected to Micron computer. The software used was  " Dr
                     attention on Arabic vowel sounds in relation to consonantal                                Speech, Tiger Electronics, Inc". The number of recorded words
                     environment, (Cf. Lehn, Walter, Harrel, R.S.).                                             totaled (64) pairs.  Formant frequencies (Fo, F1, F2, F3 ) were
                     Although some attention has been given to the study of the                                 measured by LPC analysis. In the analysis, a more or less
                     articulatory and prosodic aspects of the complex features of                               stable central segment was determined and framed as being
                     "emphasis" (cf. EL-dalee, 1984), factors that contribute to the                            characteristic of each vowel including differences between
                     phonetic identities of vowel sounds and their acoustic                                     formant frequencies of vowels occurring in "emphatic" and
                     correlates have not been given sufficient consideration.                                   "non-emphatic' contexts.
                                      2.  OBJECTIVES OF THE STUDY                                                                              4.     RESULTS
                                                                                                                4.1.  Fundamental frequency.
                     The purpose of the study is to collect articulatory and acoustic                                                                       Table 1 shows the summary
                     data that would provide reliable evidence for practitioners in                             statistics of (Fo) data of female-male group. The female group
                     communicative skills when describing or comparing vocalic                                  has a higher average  (Fo) than the male group. There is
                     units. As variation of formant frequency is closely linked to the                          marked difference in the values of fundamental frequency in
                     widely reported basic dichotomy of "emphasis":"non-                                        relation to the presence of a preceding or following  "emphatic"
                     emphasis", differences between the average frequency values                                consonants.
                     for vowels occurring in "emphatic" and "non-emphatic"                                      4.2. Formant frequencies
                     contexts will be highlighted.                                                                                                  . Table 2 gives the average values
                                                                                                                for the frequency of F1, F2 and F3 for the   (10) Arabic pure
                                                   3.  PROCEDURE                                                vowels.  The average values reflect the systematic relationship
                     3.1.         Material. The material of investigation consisted of                          between formant frequency and articulatory configuration in
                     64 contrastive CV (V) C words which represent instances                                    that the first four vowel sounds form a progression from a close
                     containing each of the ten Arabic monophthongs. The test                                   front to an open front articulation.  The progression from open
                     words were chosen so that they exhibit the contrastive                                     to back produces a gradual reduction in the frequency of (F1),
                     relationships between       "emphatic" and "non-emphatic"                                  similar to results obtained elsewhere (Cf. Ladefoged, 1962).
                     unitary complexes of C+V. The ten monophthongs, which are                                  The results show clearly the formant patterns in which F1 is
                                                                                                   ι            consistently lower during "emphatic" articulation than the
                     subject to a first two-fold division between front series [ :, i,                          "Non-emphatic". There is good correlation between the
                                                                 Α
                     e:, a, a:  ], and a back series [ :, o:, u, u:] and a threefold                            coarticulatory effect of "emphatic" consonants and the distance
                                                                                        α       Α Α
                     division: close [i: i , u: u, ] Mid [e:,o: ] and Open [ :, a,  :,                  ,       between the first two formants which are far apart in front
                     ] are illustrated by the following commutations:                                           vowels and close together in back vowels.  In relation to F2,
                     /a/: /a:/, /dam/ 'blood': /da:ml 'lasted'; /tam/: 'completed', /ta:m/                      the figures show some variation correlated with "emphatic"
                                                                                Α Α
                     'complete';  /bat/ 'decided': /ba:t/ 'became'; / /: / :/, /Dal/ 'lost                      sulcal articulation vs. "emphatic" plosive articulation. Whereas
                                 Α                                    Α                           Α
                     way': /D :l/ 'one who loses way', /S m/ 'memorized' : /S :m/                               the frequency of F2 is higher for the "emphatic" plosive
                                          Α                                Α
                     'fasted', /a/: / /, /dam/: 'blood', /D m/ 'included', /bat/
                                                                                                    page 2117                                                    ICPhS99          San Francisco
             consonant /T/, the frequency of F2 is lower for the denti-
                                      ∆
             alveolar sulcal fricatives /S/ and / /.
                         Vowel     Speaker 1(fem)    Speaker 2 (fem)   Speaker 3 (fem)  Speaker 4 (male)
                                  Mean      SD      Mean      SD      Mean      SD      Mean      SD
                                 Fo (Hz)    (Hz)   Fo (Hz)    (Hz)   Fo (Hz)    (Hz)   Fo (Hz)    (Hz)
                          [i:]     235      2.18     252      4.28     239      4.86     132      1.93
                          [e:]     207      2.96     240      4.41     212      2.46     137      2.96
                          [a:]     219      3.34     243      5.12     212      2.44     151      4.63
                           Α       203      2.45     240      4.16     213      2.70     140      4.73
                          [ :]
                          [o:]     214      7.57     241      3.87     216      2.78     146     22.79
                          [u:]     228     13.36     254      3.77     218      3.23     143      4.37
                                    Table 1. Fundamental Frequency of Arabic vowels, sustained production
                         Word      Speaker 1(fem)    Speaker 2 (fem)   Speaker 3 (fem)  Speaker 4 (male)
                                  Mean      SD      Mean      SD      Mean      SD      Mean      SD
                                 Fo (Hz)    (Hz)   Fo (Hz)    (Hz)   Fo (Hz)    (Hz)   Fo (Hz)    (Hz)
                         [Si:d]    208     31.43     220      2.85     227      2.82     145     23.87
                          [si:d]   234     12.79     241      4.71     223      5.93     149     11.59
                         [sa:d]    235     32.53     234      4.99     223      6.38     137      4.49
                         [SΑ:d]    192     34.12     217      4.62     217      3.50     136     13.09
                         [Se:d]    204     27.25     237      5.72     219      5.02     141     13.46
                         [ta:b]    219     11.59     228     10.39     224      6.86     134     19.20
                         [TΑ:b]    198     27.40     221      1.56     215      1.89     143     19.20
                         [tu:b]    254     15.74     249      2.30     226      2.45     147     12.73
                         [Tu:b}    229     32.55     215      1.56     230      2.75     157     19.92
                         [to:b]    214     24.22     210      2.80     223      5.49     141     28.09
                         [be:D]    243     23.75     237      8.47     221      3.70     128      3.71
                         [bΑ:D]    189     1135      223     30.06     218      6.26     134     21.92
                                   Table 2. Fundamental Frequency of Arabic vowels within C1VVC2 domain
                          Vowel          Fo                F1                 F2                F3
                                   S1    S2    S3    S1    S2     S3    S1    S2    S3    S1    S2    S3
                            [i:]   240   261   218   268   468   232   3206  2613  2789  3537  3285  3311
                            [i]    248   265   220   399   368   232   2199  2091  2789  3163  3211  3311
                            [e:]   220   258   219   402   429   445   2394  2269  2235  3130  3015  1839
                            [a:]   225   249   216   295   728   297   1095  1980  904   2391  3063  2904
                            [a]    227   256   213   256   559   347   720   1714  1364  2264  3234  2220
                            Α      223   244   204   233   737   444   911   2934  1746  2997  3435  3972
                           [ :]
                            Α      209   246   203   318   750   412   833   2637  1904  2417  3488   844
                            [ ]
                            [o:]   215   248   207   286   358   407   2582  1570  1934  3951  3518  3716
                            [u]    236   256   210   429   316   342   2013  3239  2423  3380  4317  3642
                            [u:]   212   258   205   214   311   285   1389  3030  2037  2513  4558  3446
                         Table 3. Average formant frequencies for the CV (V)C contexts for the female speakers (S1, S2, S3).
                          Vowel          Fo                F1                 F2                F3
                                   S4    S5    S6    S4    S5     S6    S4    S5    S6    S4    S5    S6
                            [i:]   173   148   164   287   207   260   908   2416  2459  2956  2989  2912
                            [i]    168   137   150   390   312   278   1922  1836  1911  2782  2712  2776
                            [e:]   168   160   147   435   242   392   2137  2069  2070  2784  2908  2811
                            [a:]   165   137   145   676   709   528   1535  2001  1503  2532  3229  2977
                            [a]    168   139   149   636   368   515   1639  1336  1508  2706  2739  2678
                            Α      149   135   144   679   754   646   1425  2301  1463  2654  3659  3080
                           [ :]
                            Α      156   127   142   603   703   572   995   2502  1149  3322  3395  2933
                            [ ]
                            [o:]   148   126   127   519   312   309   1564  1596  1683  3446  3344  3061
                            [u]    158   142   144   357   296   443   1202  2096  1346  2499  3276  3031
                            [u:]   154   143   161   307   271   287   2237  2197  2553  3687  3400  3523
                         Table 4. Average formant frequencies for the CV (V) C contexts for the male speakers (S4, S5, S6).
                                  NOTES                           1. Brief reading conventions for the symbols used in the transcription
                                                                  are as follows:
                                                           page 2118                          ICPhS99          San Francisco
                 T: voiceless "emphatic" denti-alveolar plosive.
                 D: voiced "emphatic" denti-alveolar plosive.
                 S: voiceless "emphatic" sulcal denti-alveolar fricative.
                 ∆: voiced "emphatic" non-sulcal inter-dental fricative.
                 /i:/ close front spread.
                 /e:/ mid to half-close front spread.
                 /a:/ open front unrounded.
                 /Α:/ back open neutrally rounded.
                 /o:/ mid to half-close back rounded.
                 /w:/ back rounded vowel.
                 2. The "emphatic" consonants /D/, /T/, /∆/, /S/ are distinguished from
                 the "non-emphatic" counterparts /d/, /t/, /∆/, /s/ by lateral expansion of
                 the whole body of tongue accompanied by faucal and pharyngeal
                 constriction; the back of the tongue is raised towards the velum. The
                 "non-emphatic" consonants, on the other hand, are pronounced with
                 contraction of the whole body of tongue.
                                           REFERENCES
                 [1] El-Dalee, M.S. & El-Ani,S,H. 1984. Tafkhim in Arabic.
                 Proceedings of the Tenth Unt. Congress of Phonetic Sciences.
                 Dordrecht: Holland.
                 [2] Fant, G. 1960. Acoustic Theory of Speech Production. The Hague:
                 Mouton
                 [3] Firth, J.R. 1948. Sounds and Prosodies. Transactions of the
                 Philological Society
                 [4]Fry, D,B. 1979. The Physics of Speech. Cambridge: University of
                 Cambridge Press
                 [5]Harrel,  R.S. 1957. The Phonology of Colloquial Egyptian Arabic.
                 New York:Amsterdam, Council of Learned Societies, 25-78
                 [6]Kjell, Norin, 1983. Acoustic Analysis of Fricatives in Cairo Arabic.
                 Working Papers
                                 25, 113-17, Lund University.
                 [7} Ladefoged, P. 1962. Elements of Acoustic Phonetics. Chicago:
                 Chicago University Press
                                                                   Language
                 [8] Lehn, Walter. 1963. Emphasis in Cairo Arabic.           , 39:1 ,
                 29-39
                 [9] Mitchell, T.F. 1960. Prominence and Syllabication in Arabic,
                 BSOAS XXIV, 369-89
                 [10] Ohala, J. J. (eds) 1968. Experimental Phonology. Orlando:
                 Academic Prerss
                                                 Prosodic Analysis
                 [11] Palmer, F.R. (ed) 1970.                     . Oxford: Oxford
                 University press
                                                                                page 2119                                        ICPhS99          San Francisco