Third Mediterranean Meeting on Morphology (MMM3)

Abstract

Stuart Davis (Indiana University)
The Status of the Consonantal Root in Arabic

There is a present controversy among linguists who work on Arabic as to the grammatical status of the consonantal root. The controversy revolves around the issue as to whether the root consonants of the Arabic word comprise a separate lexical morpheme which is underlyingly represented on a separate tier as proposed in the seminal work of McCarthy (1981), or do the root consonants have no formal grammatical status as a morpheme; i.e. they don't comprise an independent entity represented on a separate tier in underlying representation. This question has implications for the larger issue of whether all morphology is word/stem based (Aronoff 1976). If the root consonants of Arabic comprise a morpheme from which other words are based then this poses a challenge for the stem-based/word-based views of morphology where word-formation processes should not be based on entities that are smaller than a stem. There are two trends in current research regarding the lexical status of Arabic root consonants. One trend, based largely on external evidence, argues for the morphemic status of the consonantal root (Prunet et al. 2000, Linguistic Inquiry), and the other trend, largely based on morphophonological and morphosyntactic phenomena argues against the consonantal root as a morphological entity (McOmber 1995, Ratcliffe 1997, Benmamoun 1999, and Gafos 2000). In this paper I reconcile the conflicting views regarding the status of Arabic root consonants by arguing that the consonantal root is a property of the output, not of the underlying input. That is, there is no separate consonantal tier in the underlying representation; nonetheless, the consonantal root is a property of the output word and so can be referenced in output-to-output word formation processes. The bulk of this paper will argue for this position in two ways. First, I will review the types of evidence that have been used by the researchers above to argue for and against the morphemic status of root consonants and show that it is consistent with the view that the consonantal root is a property of the output, not of the underlying input. Second, I present evidence from Arabic hypocoristics (nicknames) supporting this.

If we review the type of evidence cited by researchers for and against the morphemic status of root consonants in Arabic, we would note that the type of evidence cited is very different. The evidence used to argue against root consonants is morphological (word formation) evidence that, from the perspective of a theory like Lexical Phonology/Morphology would involve rules of the early strata. These word formation processes assume a CCVC stem as basic with the vowel being lexically specified. As shown, especially by Benmamoun (1999), this CCVC stem, which is equivalent to the imperfective stem, serves as the basis for the perfective, the imperative and various derived nouns. There is no need to separate out consonants and vowels. On the other hand, the evidence cited in support of the consonantal root is largely external. For example, Prunet et al. (2000) cite evidence from aphasic speech, speech errors, and language games for the lexical morphemic status of the consonantal root. However, it can be shown that all these phenomena involve, what in optimality theory would be output-to-output processes and so do not refer to the deeper Lexical Phonology/Morphology. This is suggestive of the consonantal root being an output property. Thus, a careful review of the evidence in the literature shows a compatibility with the view that the consonantal root is a property of the output, not of the underlying input.

A second argument for the consonantal root being an output property comes from the analysis of Arabic hypocoristics (or nicknames) based on current work by the present author. Consider the hypocoristic data in (1) representing a common pattern in Arabic.

(1)

Full Name	Hypocoristic	Full Name	Hypocoristic
a. Haamid	Hammuud	d. muHammed	Hammuud
b. basma	bassuum	e. ?ibtisaam	bassuum
c. ?amal	?ammuul	f. ¿ayda	¿ayyuud

The data show that hypocoristics reference root consonants from the full name. The consonants in (1d) and (1e) that do not surface in the hypocoristic are affixal or epenthetic. Interestingly, the root consonants that are referenced are not necessarily the same as what would be considered the underlying root. Consider the name [¿ayda] which has the hypocoristic [¿ayyuud] in (1f). The underlying form for the name [¿ayda] is /¿aawida/. The change from /¿aawida/ to [¿ayda] reflects regular (morpho)phonological processes. Note that the hypocoristic is based on [¿ayda] not /¿aawida/. This suggests that hypocoristic formation is an output-to-output word formation process based on the full name that nonetheless references the consonantal root. This supports the contention that the Arabic consonantal root is an output property and not part of the underlying input, and it reconciles the conflicting evidence regarding the status of the Arabic consonantal root. Implications of this are discussed for stem-based/word-based views of morphology.