Abstract
As explained in § 4.1, our analysis of the syllable structure of Imdlawn Tashlhiyt revolves around two theses, the Licit Consonantal Nuclei thesis and the Sonority-Driven Syllabification thesis. The evidence in favor of the Sonority-Driven Syllabification thesis was presented in the last two chapters. In this chapter we present our evidence in favor of the Licit Consonantal Nuclei thesis. According to this thesis, Imdlawn Tashlhiyt has no epenthetic vowels; if the nucleus in a syllable is not a full vowel it is a consonant.
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
See also § 4.1.
For examples of that mapping, v. Dell (1984).
This assertion is valid only for those VTVs which occur between heterorganic consonants. Between homorganic consonants it is another matter, see below § 6.3.3.
Aspinion (1953: 120) likens the ‘e brefs’ of Aštuken Tashlhiyt to French schwa.
Mentioning the midsagittal region allows us to speak of the release of I, which we take to be a noncontinuant. On the relevance of pulmonic egressive airstream, v. Kim (1994).
This is in sharp contrast with the @ vowel of Rifian and Moroccan Arabic, which can occur between two voiceless consonants (v. § 6.5, § 8.2.2), or with fast speech pronunciations such as tkila for tequila in English, where a vowel can occur between t and k in slower speech (Hammond 1997: 34).
Homophonous with /ix t-lkm/ ‘if she reaches’ (if 3fs-reach).
Said, for instance, to a woman whose hair is hanging in front of her eyes.
Kinesthetic observations by ME suggest that sequences /fl, bl, ml/ may be counterexamples. We disregard these pending further research.
On ‘open’ and ‘close’ transitions, see Bloomfield (1933) and Catford (1977).
On articulatory overlap, see for instance Browman and Goldstein (1989, 1990).
Sonorant consonants are as a rule fully voiced in Imdlawn Tashlhiyt. They may devoice when they are subject to prepausal annexation after a voiceless consonant (v. DE 1985), but even in that context the devoicing is only a partial one.
This was already pointed out in DE (1985: 117).
Contextual factors other than the properties of the two consonants are also involved in determining the vowel quality of the intervening VTV. Coleman (2001) shows that the color of VTVs is influenced by neighboring vowels.
The expressions in (6) have the following meanings: (a) I covered for him; (b) I covered; (c) she lined it (clothing); (d) cause to vanish!; (e) he vanished; (f) wring the neck!; (g) wring (someone’s neck) for him!; (h) I wrung his neck; (i) she whipped; (j) she owns it; (k) you pulled; (1) you pulled for him; (m) you drowned; (n) he strangled; (o) she strangled you; (p) she bamboozled him; (q) you learned; (r) exchange!; (s) knead!
Note that /q/ is not included (see below).
The coronal closure in [istttut] is longer than that in the contrasting form [isttut] /is=t#t-ut/ ‘did she hit him?’ (int=do3ms 3fs-hit).
As already stated in § 3.1, we follow the proposals of Clements and Hume (1995) concerning the internal structure of segments.
There exists yet another variant, viz. dd.
But release is only optional in /mra t-ttu/ ‘if she had forgotten’.
Instances of sequences of more than two siblings were given in the text under (9).
In (12)a /d#t/ can also be pronounced as tt (but not as t2t, and in (12)b /k#kw/ can also be pronounced as a geminate kwkw (but not as kkw2kw). These free variants are due to regressive assimilations in phonation type and in rounding to which we will return later. These assimilations are optional in sequences of short consonants straddling a word boundary. When regressive rounding assimilation occurs in /k#kw in (12)b the preceding a is articulated further back. Only then does sentence (12)b become homophonous with the following: /is=akwkw ra-n tigmmi/ ‘the fact is that they even want the house’ (indeed=even want-3mp house).
That is, if they agree for the features [sonorant], [vocoid] and [approximant]. Clements and Hume’s [+vocoid] is the equivalent of [-consonantal] in Chomsky and Halle (1968), and t-vocoid] is the equivalent of [+consonantal]. Vocoids and liquids are [+approximant] whereas the other sounds are [-approximant].
The diagrams in this section represent only those aspects of the structure of segments which are relevant.
See Saa (1995).
See Dell and Tangi (1992: 158-159).
… unless there has been an assimilation in phonation type or in secondary labiality. Release is incompatible with assimilation between sibling stops, see DE (1996a: 385-388).
Contrary to the generalization stated at the beginning of this paragraph, there are a few contexts in which Fusion merges a geminate with a simplex sibling. For instance, in 2nd person imperfective forms, the prefix sequence /t-tt-/ must be realized simply as tt. Similarly, /dd-t/ must be realized as tt in /t-bidd-t/ ‘you stood up’ (tbitt). In all such cases, however, Fusion involves the loss of a skeletal slot, and its outc]ome abides by NO-TREBLE (see below).
u-king with=dir 3ms-come.
As formulated in (19), NO-TREBLE is too restrictive, for it excludes languages in which a nasal borrows its primary articulation from a geminate, for example /n+bb/ > mbb. A more adequate formulation is given in DE (1996a: 383).
We account for the optionality of release in sequences such as /tt#t/ by assuming that Fusion (13), which is an obligatory rule, is blocked by NO-TREBLE (19), and that SIBLING-RELEASE (11) is optional. A reviewer has pointed out an alternative: Fusion would be optional in sequences such as /tt#t/, and SIBLING-RELEASE would be obligatory in all contexts. We do not retain this alternative because it would force us to give up restriction (19), which enables us to link up the distribution of releases with the blockage of rules of complete assimilation (see Chapter 3) and with certain facts about the lexical distribution of adjacent identical consonants, on which v. below in § 6.4.1. Yet other evidence in support of (19) is provided by the behavior of the causative prefix before sibilant-initial kernels, v. DE (1996a: 381-385).
Impf 2-eat:impf-2mp.
On the others, see § 6.4.1.
The items in (21)a are all borrowings from Arabic. They all have a free variant with i after the second consonant, e.g. zmmim, xmmim, fnnin, etc]. The free variation between i and zero is also found in native verbs, viz. in the biconsonantal verbs where both consonants are obstruents, e.g. b(i)dd’ stand up’, bb(i)z ‘pound’, kk(i)s ‘remove’. In all such verbs one of the consonants is a geminate.
The uninterrupted triple m in zmmm sounds longer than the uninterrupted double m in tllmm ‘you (mp) spun’ (from /t-llm-m/).
In our 1985 article we stated that a simple consonant cannot immediately precede its geminate counterpart in a lexical entry, see (48)b p. 124. The existence of the items in (21)b shows that that assertion is false.
On the Obligatory Contour Principle see McCarthy (1986), Odden (1988) and references therein.
On syllabification and epenthesis in Ath Sidhar Rifian Berber, v. below in § 6.5.
Here are the glosses for the parenthesized forms: (a) id impf; (b) ‘hiding’ (deverbal noun, pluralia tantum); (c) ‘weave!’; (d-f) id p.
On the realization of causative /s-/, v. § 5.4.
In Tashlhiyt as in other Berber dialects, /dd/ is generally realized as tt in emphatic morphemes.
An exception must be made for vcd-vls sequences beginning with /b/, a consonant which is in some instances immune to devoicing in Imdlawn Tashlhiyt. In Imdlawn the extent to which regressive devoicing operates varies with the speakers’ age. It is more pervasive in the language of older speakers such as ME’s father. Even in ME’s less conservative dialect, the vcd-vls sequences in which regressive devoicing is only optional all belong to kernels which have transparent cognates in the local variety of Moroccan Arabic.
On vowel insertion in the imperfective, see § 5.2.
While the initial vowel belongs to the stem in Imdlawn, it is an augment in At Mangellat.
The Berber stratum also contains vestigial pairs which do not fit anymore into productive alternation patterns, e.g. a-bggas ‘belt’ / biks ‘gird!’ (see (24)f). The imperfective stem of biks is tt-bikis, not tt-bigis, which shows that in biks the velar consonant is lexically voiceless in present-day Imdlawn Tashlhiyt.
See e.g. Aspinion (1953).
On Haha Tashlhiyt, v. Ouakrim (1993) and Ridouane (1999). The latter discusses Haha spirantization in some detail.
Compare for instance Basset and Picard (1948) for Kabyle, Penchoen (1973) for Tamazight, and Aspinion (1953) and Destaing (1920) for Tashlhiyt.
For discussions of nearby dialects with rather similar phonological systems, see Chami (1979), Cadi (1981) and Chtatou (1982).
Oufae Tangi’s father and mother are from the Ath Sidhar area and Berber was her first language. She uses Berber with her parents and with other members of her family of their generation, some of whom are monolingual. She uses Arabic with her sisters and the people outside her family.
They are bare aorist stems. As in Tashlhiyt, such stems are used as 2s imperative forms.
RIGHT-TO-LEFT SCAN is almost identical with the epenthesis rule proposed in Saib (1976: 127) for the Ayt Ndhir variety of Tamazight. On that rule, see Hyman (1985: 68).
On how initial clusters fit in with this general scheme, see DT (1992).
See the references in note 49 for the works on Berber. For those on Moroccan Arabic, the references will be found in Chapter 8.
The angled brackets around the do3ms clitic indicate that it is extrametrical; in view of this it is disregarded by RIGHT-TO-LEFT SCAN, see below.
In Moroccan Arabic, according to Heath (1987: 184), schwa deletion in word-final syllables is blocked by ‘list intonation’, and in Japanese high vowel devoicing is blocked ‘when a final syllable in the devoicing environment must carry a rising intonation’ (Vance 1987: 51).
See Basset and Picard (1948: 9), Mitc]hell (1957: 197-198), Penchoen (1973: 10, 94). Extensive data elicited from Fouad Saa show that a situation similar to that just described also prevails in the Figuig dialect, on which see Kossmann (1994) and Saa (1995).
E.g. Harrell (1962a), Mitc]hell (1993: 62, 64), Shoul (1995: 208).
θ and θ do not have a geminate counterpart. Historically they derive from simplex t and d and still alternate with them, but they must be considered independent phonemes. One of the reasons for this state of affairs is a massive influx of Arabic loans with unspirantized t and d.
The corresponding masculine noun is a-ʔarraz /a-ʔrraz/. As a rule /rr/ surfaces as arr. /r/ is realized as r, ar or a depending on the context. See DT (1993) for a detailed discussion of how these alternations link up with syllabification.
Absorption of an unstable vowel by a neighbouring sonorant has been ascribed to other Berber dialects, see Mitc]hell (1957: 194) and Penchoen (1973: 10, 94), and to Moroccan Arabic, see Mitc]hell (1993: 63, 72) and Heath (1987: 249-253).
The glides of Ath Sidhar Rifian are discussed in detail in DT (1992).
V. also Levin’s (1987) distinction between epenthesis and excrescence. The facts of Ath Sidhar Rifian suggest that when epenthetic vowels and excrescent vocoids coexist in the same language, they need not have different vowel qualities.
As in Tashlhiyt, such nouns begin with a prefix /1-/, see § 2.5.3.1. The prefix assimilates to a following coronal; otherwise it surfaces as r, as do most occurrences of simplex /1/ in Ath Sidhar Rifian. On /r/ and /1/ in Ath Sidhar Rifian see DT (1993).
The causative prefix assimilates to a following sibilant, v. § 5.4.
On extrametricality, see e.g. Hayes (1995) and references therein.
Basing themselves on the limited data in Tangi (1991), DT (1992: 134) stated incorrectly that only coronal obstruents can be extrametrical. Subsequent work with Oufae Tangi has turned up nativized loans like s-s@rk ((27)c) and r-m@sk ((27)e), which end with non-coronals.
The initial /1-/ assimilates to the following coronal, hence /rr/, which is realized as arr.
Kossmann (1995: 80) presents similar facts concerning Ait Said Rifian.
Plural ssbayy. The MA source noun is !sbay-a (p !sbayy).
The fourth is extrametrical. It is the final /-θ/ which marks the feminine in /θ-…-m-<θ>/ ‘2fp’ and in /-n(<θ>/ ‘3fp’.
V. the i-epenthesis mentioned in note 35, as well as that presented in DE (1989: 191-193). Both are optional and morphologically-governed.
The facts which make this statement only a first approximation fall into three categories: (1) the special behaviour of prefix boundaries with respect to Fusion (v. § 6.3.3.3), (2) the erasure of one X slot in certain Pword-internal XXX sequences in which all three X slots are linked to sibling segments (for instance the prefix sequence /t-tt-/ must be realized as tt when /t-/ is one of the PNG prefixes and /tt-/ is the imperfective prefix, v. DE (1989: 193), and (3) the operation of Fusion in sequences of three or more simplex consonants which are siblings, v. the examples illustrating the three-way contrast between t2t:2t, t2t: and t2t2t: in the text below (9) in § 6.3.3.1.
See Clements (1997) for some discussion of multistratal syllabification in Imdlawn Tashlhiyt.
The authors of this book have since carried out a systematic survey of nominal and verbal kernels in Imdlawn Tashlhiyt and found that verbal kernels are subject to maximal size requirements akin to those which constrain nominal kernels.
See for instance Jebbour (1988) for number inflection in nouns (Tiznit Tashlhiyt), DE (1991) for verbal stem formation and DE (1992) for derivational morphology (Imdlawn Tashlhiyt).
The TIRRUGZA derivatives are pluralia tantum nouns in the feminine. See DE (1992) for a list of forms and a discussion of the templatic mapping.
The base nouns in (36)-I have the following meanings: (a) man, (b) free person, (c) guest, (d) dummy in a card game, (e) apprentice, (f) wealthy person.
Imperfective gemination is an exception to this generalization.
Basset (1952: 8).
Basset (1952: 5) and Galand (1953).
Some of the inconsistencies in these transcriptions probably come from variations in transcriptional practice, rather than from variation in the data. See Galand (1953: 230) for examples of inconsistencies in the transcriptions in Destaing (1920).
According to Basset and Picard (1948: 9) ʕ is ‘a vocalic element […] which ranges, depending in particular on tempo, from a well-marked vowel to nothing at all. All vowels are voiced; it may be the case, however, that ʕ could devoice, leaving a mere suspension as its only residue’. Penchoen (1973: 10) writes that what he transcribes uniformly as ʔ ‘may be — phonetically — an [ʕ], the syllabicity of a consonant such as a nasal, lateral or /r/ or even a simple consonant release voiced or not’.
See for instance the rule of schwa epenthesis in Guerssel (1977: 271). As we have seen in § 6.5.1, Rifian Berber follows essentially the same pattern.
Galand (1953: 230) even hints that some transcriptions of Berber may have been influenced by assumptions about syllabification in Arabic.
See also Galand (1988: 213): ‘… most often, the numerous occurrences of [ʕ] found in works by Berberists reflect habits which are alien to Tashlhiyt.’
See Galand (1957) on the school founded in Rabat by General Lyautey in 1912.
In our view the RIPIs are terminal representations of the phonological component.
The meanings of the examples in (40) are the following: (a) it went numb; (b) she even hoarded; (c) she even behaved as a miser; (d) what will happen of you?; (e) for the cockroaches; (f) you drowned; (g) you painted; (h) I locked; (i) he strangled; (j) he strangled him; (k) broken branch; (1) he wrung (someone’s neck) for him; (m) take care of his mother!.
On the deletion of /d/ at the end of the future preverb /rad/, see DE (1989: 188).
As a result of voice assimilation and Fusion, /dt/ surfaces as geminate /tt/, see § 6.3.3.2.
On the reasons why we now prefer to work with orthometric syllabification, v. § 4.1.
ME was unsure whether the form contained one syllable or two.
The IFDQ syllabification of form (41)d violates constraint NoPICOR, which disfavors rimes in which an obstruent nucleus is followed by a coda with the same degree of sonority, see § 4.9.1.
See DE (1996b: 232) for the discussion of a case in point.
Here ‘our account’ refers to our characterization of voiced transitional vocoids in DE (1996a), which is the same as that presented in the earlier sections of this chapter.
t-guni is the bound state form of t-a-guni. The affirmative form corresponding to t-gwni is t-gwna.
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). Vowelless Syllables. 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_6
Download citation
DOI: https://doi.org/10.1007/978-94-010-0279-0_6
Publisher Name: Springer, Dordrecht
Print ISBN: 978-1-4020-1077-4
Online ISBN: 978-94-010-0279-0
eBook Packages: Springer Book Archive