Abstract
The varieties of Berber and of Arabic spoken in Morocco have rather similar surface phonologies, e.g. their segment inventories do not differ much. Reading the literature on Moroccan Arabic (henceforth MA) might give the impression that Tashlhiyt and MA have almost identical syllable structures, at least at those levels of representation which are near the terminal representations of the phonological component. One of our aims in the forthcoming discussion is to show that such an impression would be mistaken. Our outline of the syllable structure of MA in surface representations will be detailed enough to allow us to point out the basic differences with Tashlhiyt. It is also meant as an original contribution, both factual and analytic, to the study of MA phonology.
This is a preview of subscription content, log in via an institution to check access.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
Preview
Unable to display preview. Download preview PDF.
Notes
!xess=ek t-fetteš fe=s=sett-a, in the transcription adopted here (see infra). The sentence means ‘you have to inspect at six o’clock’.
For an overview of the history and dialects of Arabic in Morocco, see Colin (1985).
On the syllabification of postpausal consonant clusters, see § 8.3.1.
As defined in § 2.2, a Pword is a sequence which comprises a word together with all the clitics attached to it.
Some authors posit two contrasting unstable vowels e and o. We interpret the rounded unstable vowel as a variant of e due to a neighboring labialized consonant, e.g. we posit Vwell and xwebz where Harrell (1962b) writes koll (‘all’) and xobz (‘bread’). On this analysis, see Heath (1987: 254-263).
By ‘an analogous contrast’ we mean an underlying contrast between vowels linked to one and two skeletal positions. MA actually has a surface contrast between short vowels and long ones. Sequence iy is realized as a long i when it is tautosyllabic, and so is the sequence customarily transcribed as eyy. Such sequences occur in žmiyl=u ‘his little camel’ and in meyyl=u ‘he put him aside’. The two words are homophonous from m onwards; they both end in [mi:lu]. Similarly the following two expressions, the first a sentence and the second a noun phrase, are homophonous from s onwards: ma sewwl-u=ni ‘they did not interrogate me’ (v. sewwel ‘he interrogated’) and lun kas=u w=lun=i ‘the color of his glass and my color’ (color glass=3ms and=color=1s). Both expressions end in [su:luni].
For claims that the full vowels of MA are underlyingly long or geminate, see e.g. Cantineau (1950), Lowenstamm (1991), and also Kouloughli (1978) for an Arabic dialect of Eastern Algeria with facts similar to a certain extent.
An exclamation point prefixed to a form indicates that the form contains dorsopharyn-gealized segments. Like Berber, MA has dorsopharyngealized coronals in the lexical representations and dorsopharyngealization spreads to the neighboring segments at the phonetic level.
This procedure was invoked in § 6.5 for Rifian Berber.
As a rule the definite article /1=/ completely assimilates to a following coronal contoid.
In our transcription’ s=sder’ in (4)g-I a boundary symbol occurs between the two halves of a geminate. As in all other transcriptions in this book, the boundary symbols only provide information about morphemic affiliation at the skeletal tier; they do not imply anything about the organisation of the distinctive features associated with the skeletal positions. The symbol sequence’ s=s’ implies that we are dealing with two heteromorphemic skeletal positions but it does not indicate whether each skeletal position has its own associated feature bundle or whether the two positions share a single feature bundle, in other words it does not say whether we are dealing with two occurrences of simplex /s/ or with a geminate /s/. A consequence of what precedes is that when a symbol ‘=’ or ‘-’ is sandwiched between two occurrences of the same simplex consonant, it does not imply anything about the release of the first consonant. In fact, in all the examples below, two occurrences of the same letter with an intervening ‘=’ or ‘-’ represent a geminate or a sequence homophonous with a geminate. Qualifications will be added when we discuss consonant releases in the next chapter (§ 9.4).
(6)a does not take into account the negative enclitic /=š7, which has no phonological effect on what precedes it (Heath 1987: 243). (7)a disregards certain prefixes which give rise to initial CCCC sequences in Harrell (1962b), as in n-tt-xek-u ‘for us to be scared’ (p. 49), see also pp. 34, 46 and 57 for other instances. Empirical work has yet to be done on the pronunciation of such clusters as well as on their behaviour in versification, to determine their status with respect to syllable structure. The cases mentioned in this footnote will be ignored in the following discussion. Some of the counter-examples to (7)b are initial clusters analogous with those just mentioned, e.g. n-dfen ‘he was buried’ (Harrell 1962b: 34, see also pp. 46 and 57).
Cf. Heath (1987: 243).
Cf. kašef ‘he guessed’.
Cf. !bsel ‘onion’.
The citation forms of verbs are their perfective 3ms forms, which are naked stems.
Contrary to what the examples in (9) might suggest, CeCC nouns do not all contain an emphatic consonant, nor do they all end in a voiceless coronal obstruent, e.g. qelb ‘heart’, šelħ ‘Ashlhiy’, kehf ‘cave’, šenq ‘hanging’, mesk ‘musk’. On the contrast between CCeC and CeCC in MA nouns, see Amimi and Bohas (1996) and Zeroual (2000).
Durand (1994, 1995/96) also insists that many schwas in the transcriptions do not correspond to any vocoid.
In dubious cases, the facts are sometimes inferred from other cases, rather than ascertained by direct observation, see note 4 in Kaye (1987: 158) for an example.
On the language situation in Morocco, see Boukous (1995).
We thank Chakir Zeroual for answering our questions about the Oujda dialect.
V. Heath (1987: 251, 263-265) for passages where such assumptions are made explicit. V. also Mitc]hell, who writes (1993: 68): ‘may be less of a phonetic segment in a given case than a phonological fiction recognized […] with a view of facilitating the formulation of general structural patterns’.
See Chapter 6.
In our special use of ‘uncontroversial’ and ‘putative’, these adjectives are predicated of vowels, i.e. of phonological objects, but for the sake of convenience we will allow ourselves to predicate them also of letters in transcriptions, e.g. we shall say that, of the two occurrences of ‘e’ in !besl=ek ‘your onion’ ((9)g), the first is putative and the second is uncontroversial.
Italeb, caretaker of a mosque; often acts as a schoolteacher.
u ‘and’ has a free variant w before a consonant cluster.
The initial geminates in !t=taleb and ž=ženna result from the assimilation of the definite article /1=/ to the following coronal: /l=!taleb/, /l=ženna/. Henceforth in all surface sequences of the form #K=K or =K=K in which the two ‘K’s represent the same coronal, the first ‘K’ represents the assimilated variant of the definite article. ‘K=K’ represents a heteromorphemic geminate, i.e. two heteromorphemic skeletal positions linked with the same bundle of distinctive features.
Variant of muħammed, see following line.
On the use of ‘metrical grids’ such as (15) to represent musical rhythm, v. Lerdahl and Jackendoff (1983).
Here and below we omit from the syllabic parses the exclamation points representing emphasis.
The rhythmic pattern of the chant is the same as that of the first line of the well-known French song ‘Au clair de la lune’ (v. § 4.3). Here is that line, with the strong beats indicated by capitals: AU clair DE Ia LU-NE.
Cantineau (1960: 118-119) suggests that there are no complex onsets inside words in the modern dialects of Arabic. In his detailed and insightful discussion of the phonology of a Bedouin dialect of Eastern Algeria, Kouloughli (1978: 104ff., 256ff.) gives several arguments leading to the same conclusion. Some of his evidence is drawn from versification in songs.
qlib is a noun derived from qleb ‘overturn’.
As, e.g., in ma !te-ħmar ma te-sfar ‘you will neither blush nor turn pale’.
!šserm is a noun derived from the verb !šrem ‘tear, rip’.
In very deliberate speech, or if the speaker intends to put special emphasis on the first word, the homophony disappears in example d’, see next chapter (§ 9.2).
See for instance the second syllable in lines 2a and 3a in (30) below.
See for instance the second syllable in line 11b in (31) below.
More precisely, balls made with dried tripe.
With the hollow facing up.
neg 3fs:impf-undermine hope=1p.
fut 3ms-graze carry:aor-3mp=do3ms=dir, ‘he will graze (his skin) and they will carry him on their backs’. In the underlying representation the line begins with /rad i-nšf/. The alternations involving the final consonant of the future marker /rad/ are described in DE (1989: 188-190).
fut 3ms-graze Ieave:aor-3mp=dat3s=do3ms=dir, ‘he will graze (his skin) and they will leave it (m) with him’, nsr is synonymous with nšf in (24).
For general presentations of meltiun, see Al-Jirari (1970), Tahar (1975), Pellat (1987), A1-Malħuni (1990b) and Jouad (1995). Al-Fassi (1997) contains a representative sample of songs.
On the music of melħun, see Aydoun (1994) for a quick overview and for references to more detailed work.
See De Prémare and Alaoui (1989).
Jouad (1995: 304-314) presents ninety odd lines of melħun with their syllabic parses and French translations. DE (1988: 10) is an earlier instance of syllabic scansion of MA verse.
Qays, A1-Yazan (line 11a) and Antar (line 12a) are well-known characters from the literature.
N and D respectively stand for Nucleus and Coda. Z∼ indicates that Z is linked with the first skeletal position in a geminate.
The syllable type intantiated as ħubb in the table happens not to occur in the song cited above, but is attested in others, e.g. ħubb occurs in the first line of the first stanza in a ballad recorded on tape by Toulali which will be cited in § 8.5.6. See exact reference there.
The singular form is sekk-a. On the special status of e in skek and other similar plural nouns, see § 9.5.
The masculine form is mqiddes.
For an analysis of word-initial consonant clusters in Ath Sidhar Rifian along the same lines, see DT (1992). Hyman (1985: 68) already proposes that in Ayt Ndhir Tamazight Berber, in which the distribution of schwa is similar to that discussed here, words with initial consonant clusters begin with syllabic consonants.
Here are the translations of the four lines: (a) my heart is wrecked by love; (b) and I describe her radiant silhouette; (c) God will judge those who diverted her; (d) he has handed me a cup of bitterness.
e is also deleted elsewhere than after a vowel, e.g. in the first word in lines 6a, 7a and 8a. These cases will receive a different interpretation in § 9.1.
In support of his distinction, Tahar (pp. 73-74) cites Cantineau (1960: 119), who states that in Maghrebian dialects H syllables are ‘a little longer’ than other syllables.
Contrast that syllable with the eighth syllable (me) in line 10b in (31), which counts as L.
Similar differences were found between Tashlhiyt and Ath Sidhar Rifian, see § 6.5.
We are abstracting away from those cases in which the vocoid is absorbed by an adjacent sonorant (see above in § 8.2.2).
As mentioned earlier, RIGHT-TO-LEFT SCAN is only a temporary device adopted for expository convenience.
Here and below, underlinings indicate nuclei.
A sonority peak is any sequence which is a local maximum of sonority, see § 4.7. We are assuming that MA has the same sonority scale as that proposed for Tashlhiyt in § 4.7. This assumption will be modified later, see § 8.5.3.
In Tashlhiyt the deverbal noun of zrb is zzrb (/1-zrb/).
Harrell (1962a) already proposed that MA schwa does not have any supra-laryngeal features of its own.
MA zerb and Tashlhiyt zrb are homophonous. On such homophonies, v. § 9.3.
See Chapter 7.
See Heath (1987: 288).
More precisely, for any two strings A, B and their concatenation AB, every segment which belongs to a sonority peak in AB also belongs to one in A or B. Let A and B be the strings qfxbsn and mtiw. The sonority peaks in A are fx and n, those in B are m and iw, and those in AB are fx, nm and iw.
This was already the general spirit of Keegan’s (1986) pioneering article.
The following fact may look like a counter-example to some readers. In perfective verbs with stems ending in aC the vowel becomes e when the suffix begins with a consonant: šaf ‘he saw’, šef-na ‘we saw’ (-na is the 1p desinence). But note that *šaf-na would be well-formed, as far as syllable structure is concerned, cf. šaf=na ‘he saw us’ (=na is the dolp clitic). One of our basic assumptions is that the phonotactics of any language is due in part to constraints on the form of words which are blind to morphological structure, and it is constraints of that nature which are at the center of our attention in this chapter. The alternation under consideration here is indeed sensitive to morphological structure: First, suffixes trigger it whereas clitics do not, and yet the difference between suffixes and clitics is otherwise irrelevant to syllabification in MA. Second, the alternation is limited to verbs, e.g. /!far-t=i/ ‘my female rat’, the 1s possessivized form of /far-a ‘female rat’, is pronounced !farti, not *!ferti.
See e.g. Selkirk (1982). For MA, see Keegan (1986).
Line e in (44), which is now irrelevant, has been left out.
On overlapping syllables, see (19) in § 8.3.1
See the text below (33).
Some of these pairs were pointed out by Amimi and Bohas (1996).
The only exceptions are the 3fs suffix -et and the 2s clitic =ek (see infra).
Our assumption about the underlying representations of the CCeC words will be modified later, see § 8.5.5.
The interaction between FinH on the one hand, and SYLL and NoHiatus on the other hand, is doing the same work as the cyclic derivations proposed by Keegan (1986).
frenk came into the language as a borrowing but its origin is not anymore traceable in present-day MA. franc has lost its final velar in modern French. FaithAdapt is irrelevant here. On FaithAdapt see below in § 8.5.3.
(80) extends to all words an observation of Heath (1987: 265) about the shape of CeCC words.
Zeroual (2000) advocates a more differentiated scale, with h between the liquids and the nasals, and between the nasals and the obstruents. His conclusions are based on a detailed survey of the CC clusters in CeCC nouns. Unfortunately, this work became available too late for us to take it into account.
Counterexamples will be taken up later.
tben’ straw’, kfen’ shroud’, gdem ‘heel’, šħem ‘grease’.
žbel ‘mountain’, škel’ shape’, !šžer ‘trees, col’, !šher ‘month’.
!mer ‘length of life’, žmel ‘camel’.
On the Is suffix /-t/ see § 8.5.5.
Cf. the noun meaning ‘the Spanish, col’, which is !sbelyun, not !seblyun.
For other languages with a prohibition against syllable-final clusters with a rising sonority contour, v. e.g. Cantineau (1960: 114-115, 118-119), Kenstowicz (1986), Kouloughli (1978), Hayward (1988) and Bohas (1999).
This word is distinct from its everyday synonym !ster, see below.
For some discussion of classicized MA, see Heath (1989) and Youssi (1992).
In his account of forms like those in (85), Kaye (1987) relies on two debatable assumptions. One assumption implies that all forms which are counterexamples to generalization (80) are nominalized verbs (‘masdars’). This assumption is incorrect, as implied by our presentation of classicized MA. Here are words which contradict (80) and are not masdars: !setr (85)a, !feqr (85)b, nežm’ star, famous performer’ (cf. CA nažm), esr ‘era’ (cf. CA asr). The author also assumes implicitly that in MA all nominalized verbs are built on (what amounts to) a CeCC template. Actually, the nominalizations of many CCeC verbs are of the form CCuC or CCiC (e.g. dxul ‘entering’, from dxel ‘enter’, šrit ‘plowing’, from šret ‘plow’). As for nouns lacking a full vowel which are related to CCeC verbs, many of them are of the shape CCeC, e.g. fqes ‘disembowelment’ (c]f. fqes ‘disembowel’, teš ‘thirst’ (cf. teš ‘be thirsty’).
On phonological adaptations, see e.g. Yip (1993), Paradis and LaCharité (1997) and references therein. In the case of MA, see Heath (1989). Some of the constraints which are needed specifically to account for moroccanizations do not belong to FaithAdapt. An example of a phenomenon which falls outside the purview of FaithAdapt is the fact that the short vowels of CA are represented by e or zero in MA classicisms, e.g. CA adl ‘justice’ is moroccanized as edl, not adel, despite the fact that adel is a well-formed MA kernel, cf. !batel ‘calumny’ and šabel’ shad’.
The ranking of NoLoneSchwa above SonPeak is needed in order to account for the CCeC words in which the medial C is a sonority peak, e.g. !Išreb ‘he drank’. Candidates.šer.be. and .eš.reb. both meet FinL. .eš.reb., which violates SonPeak, is preferred to .šer.be., which satisfies SonPeak but violates NoLoneSchwa, because NoLoneSchwa is ranked higher than SonPeak.
The reader is referred to our previous discussion of qleb and qelb-u, whose viable candidate sets were given in (63) and (65), and to that of !bežyet and !žeyt-u, for which see (67) and (68).
This is the word given for ‘chess’ in Harrell and Sobelman (1966: 156). It is unknown to ME, who only knows the variant !šetrenž.
On these two desinences see e.g. Heath (1987: 233).
A similar phenomenon is attested in Berber. In the Tamazight dialect of Ait Seghrouchen, according to Guerssel (1977: 274-275), CCeC is obligatorily changed into CeCC whenever the middle C is a sonorant.
CeRC sequences in which R is a sonority peak may not be pronounced CReC, e.g. kelb-na ‘our dog’ does not have a free variant klebna.
Certain nouns have undergone lexical change, e.g. the words meaning ‘muslim’ and ‘earthen dish’ are mselm and mterd in Lmnabha, while they are recorded as meslem and metred in Harrell and Sobelman (1966), who state (p. ix) that their dictionary is based on the speech of educated speakers from Fes, Rabat and Casablanca.
‘I let you go, crushed under the weight of your deeds’.
(a) ‘He became unfair and tyrannical, and reached the depths of sin’; (b) ‘It only remained for him to name himself Kisraa so as to replace Kisraa’ (CA kisraa = Caesar).
Side One of tape TCK684 (no date or place indicated on the cassette’s packaging). The title of the song is ‘Fatma’, like that of the song to which line (102) belongs; it was composed by Dris Ben-Ali Elmalki.
(a) ‘Have you now become deaf?’; (b) ‘At the bottom of the heart he lit up the fire of passion’.
Author information
Authors and Affiliations
Rights and permissions
Copyright information
© 2002 Springer Science+Business Media Dordrecht
About this chapter
Cite this chapter
Dell, F., Elmedlaoui, M. (2002). Syllable Structure in Moroccan Arabic. In: Syllables in Tashlhiyt Berber and in Moroccan Arabic. Kluwer International Handbooks of Linguistics, vol 2. Springer, Dordrecht. https://doi.org/10.1007/978-94-010-0279-0_8
Download citation
DOI: https://doi.org/10.1007/978-94-010-0279-0_8
Publisher Name: Springer, Dordrecht
Print ISBN: 978-1-4020-1077-4
Online ISBN: 978-94-010-0279-0
eBook Packages: Springer Book Archive