Abstract
Within the Theory of Constraints and Repair Strategies (Paradis, 1988a,b; Paradis & LaCharite, 1993, 1997, 2001, 2002, 2005, 2008, henceforth TCRS), we provide a formal analysis to the Arabic loanwords in seven languages spoken in Ethiopia: Ge'ez, Tigre, Tigrinya, Amharic, Harari, Argobba and Gurage. The analysis draws upon a corpus of 540 loanwords extracted from the works of Leslau (1956a,b,c; 1957a,b,c; 1958; 1963; 1990). The article presents theoretical challenges to the TCRS Loanword Model (Paradis & LaCharite, 1997), in particular to the Threshold Principle which stipulates that an illicit segment should universally undergo less than two repairs to be licensed in the borrowing language; beyond this limit, it will be deleted. The adaptations of Arabic segmental malformations in these seven Ethiopian languages, however, exceed this number totaling in certain cases to six repairs. The article also discusses the Arabic gutturals, [7], [c] and [h], which undergo unpredictable deletion in Amharic and Argobba, showing that the Non-Availability Hypothesis (Paradis & LaCharite, 2001) cannot account for these deletions either. Although the Francophones systematically delete gutturals in Arabic loanwords due to the non-availability of Pharyngeal node in French, the inventories of Amharic and Argobba include the laryngeal [h], the uvular [q] and the glottalized ejectives, thus employing Pharyngeal node plus the features [RTR] and [constricted glottis] as phonologically treatable primitives.