Academic journal article Studies in African Linguistics

The Origin of Mid Vowels in Siwi

Academic journal article Studies in African Linguistics

The Origin of Mid Vowels in Siwi

Article excerpt


Most Berber languages have a vowel system consisting of /a/, /i/, /u/ plus /a/. Some varieties, however, additionally display mid vowels /e/ and/or /o/. Insofar as these vowels appear in inherited vocabulary, their origin poses difficulties: Are they inherited from proto-Berber, or do they derive from later secondary developments? Students of Berber have traditionally assumed the three vowel plus schwa system to be original. However, Prasse (1990) has argued that /e/ existed in proto-Berber and has been preserved as such in Tuareg and in Ghadames, while /o/ did not. Recent documentation of Ghomara Berber (El Hannouche 2008, Mourigh forthcoming) has opened up the possibility that the reflexes of *e (> a) and *i also remain distinct there, strengthening the case for its antiquity.

In Siwi, the Berber language of Siwa in western Egypt, both /e/ and /o/ are phonemically distinct from /i/ and /u/, as first observed by Vycichl (1981:176, 2005:180) and established by Naumann (2012: 272-273, 303-307). They occur in inherited vocabulary as well as loans. Naumann suggests, "The mid vowels /e, o/ seem to have acquired phonological status only recently in Siwi", noting that they are "most typically found in final CVC syllables", but makes no attempt to explain their emergence (2012:272-273). Vycichl (2005:189) proposes derivations for two morphemes with /e/, but makes no attempt to establish regular correspondences involving this vowel. Souag (2013:35) suggests diphthong coalescence and vowel harmony as sources for Siwi mid vowels, but does not demonstrate this or work out the details. The origin of Siwi mid vowels therefore remains an open question. This paper will show that Siwi has preserved the *e reflected in Tuareg and Ghadames only before word-final n; otherwise, it has innovated /e/ and /o/ through a "conspiracy" of several distinct sound changes.

Unless otherwise stated, all Siwi data cited here derives from Souag (ms), a draft lexicon of about 3500 words; forms from Naumann (ms) are specifically cited as such. For other Berber languages, the sources unless otherwise indicated are: Van Putten (2014) and Paradisi (1960a;b) for Awjila; Paradisi (1963) for El-Fogaha; Benamara (2013) for Figuig; Lanfry (1973) for Ghadames; Dallet (1982) for Kabyle; Mourigh (forthcoming) for Ghomara; Oussikoum (2013) for Middle Atlas; Beguinot (1942) for Nefusi; Delheure (1987) for Ouargla; Sarnelli (1924) for Sokna; Prasse et al. (2003) for Tamajeq; Heath (2006) for Tamasheq; Boudot-Lamotte (1964) for Timimoun; Taine-Cheikh (2008) for Zenaga.

1.1.A note on phonology. Siwi [e] and [o] occur only word-medially, never word-initially or word-finally. In Arabic loans, they correspond regularly to Classical Arabic word-internal /ay/ and /aw/, typically realized in regional dialects as /e/ and /ö/. Conversely, the diphthongs conventionally transcribed as [ay]2 and [aw] occur word-finally but not word-internally (with rare exceptions in Arabic loans). In this position, they reflect Classical Arabic /ay/ and /aw/ just as /e/ and /o/ do word-internally (e.g. attaw 'light' < Arabic daw ). Synchronically, the obvious conclusion is that [ay] and [aw] are allophones of /e/ and /o/ respectively, and this is supported by some alternations (see 2).

However, their history appears to be different: [ay] and [aw] correspond regularly to vowel+y, w in languages that have preserved such combinations, rather than to monophthongs: e.g. Siwi 3zmay "sew (palm strips)" = Tamasheq szmsy "sew (clothing, tent)"; Siwi adday "below" = Tamasheq ádday "the lowest"; Siwi iraw "give birth" = Tamasheq arsw "id.", Siwi ayraw 'half-ripe date' = Sokna 3. Siwi's non-final allomorphs [e], [o], on the other hand, correspond not only to diphthongs (in particular /äy/) but also to monophthongs (/i/, /u/) in such languages, a phenomenon requiring explanation. This article will therefore focus on the history of non-final /e/ and /o/. …

Search by... Author
Show... All Results Primary Sources Peer-reviewed


An unknown error has occurred. Please click the button below to reload the page. If the problem persists, please try again in a little while.