Theoretical Models Testing for Examining Permissible Onset Sequences in Jordanian Urban Arabic

Languages differ as to which sounds are permissible to combine to form onset sequences. The aim of this article is to test two versions of the theory of sonority hierarchy to examine to what extent that they can make correct predictions about permissible onset sequences in Jordanian urban Arabic. It also aims at testing articulator-based feature theory to examine its capability to account for permissible onset sequences in the language in question. Findings of the study have shown that the sonority theory seems to fail to predict some occurring onset sequences in Jordanian urban Arabic and to make wrong predictions about some other missing sequences. Some sequences violate the minimal sonority distance, but they are found in Jordanian urban Arabic while some other sequences satisfy the minimal sonority distance, but they are not found in this language. The results have also shown that the articulator-based feature theory fails to account for permissible onset sequences in Jordanian urban Arabic.


INTRODUCTION
Languages differ as to which word-initial sounds can be adjacent or not. Every language imposes restrictions on which sounds are permissible to combine to make up a wellformed sequence. Some examples of onset sequences from Jordanian urban Arabic (JUA, henceforth) are shown in (1). For the IPA symbols of JUA, see (appendix 1).
(1) Onset Sequences in JUA Onset Sequence Example Gloss [kt] [ktaːb] 'book' [kl] [klaːb] 'dogs' [fṭ] [fṭuːr] 'breakfast' [dm] [dmuːʕ] 'tears' The list in (1) Kenstowicz's (1994) and Gouskova's (2004) two versions of the theory of sonority hierarchy to examine to what extent that the two versions can make correct predictions about permissible onset sequences in JUA. It also aims at testing Duanmu's (2002Duanmu's ( , 2008 articulator-based feature theory (or complex-sound approach) to investigate its capability to account for permissible onset sequences in JUA. I opted for the theory of sonority hierarchy as it has been commonly used in the analysis of sequences IJALEL 9 (5): [27][28][29][30][31][32] of sonority hierarchy. The purpose of this analysis is to test Kenstowicz's (1994) model and examine the extent to which it can make correct predictions about permissible onset sequences in JUA. According to Kenstowicz (1994), four degrees of sonority are proposed for consonants and glides as it is shown in (3) [sl], the sonority distance is 3 -1 = 2, which is higher than the MSD, so the sequence is good. The sonority distance of [sw] is 4 -1 = 3, which exceeds the MSD constituting a good sequence.
The examples in list (4) show issues with the sonority analysis. Firstly, some sequences violate the MSD, but they are found in JUA, such as [sd, sb, sf, kt, kf, kb]. Secondly, some other sequences satisfy the MSD, but they are not found in JUA, such as [lm, ln, rn, rs, wf, wn, ws, wb]. List (5) shows to what extent the sonority-based analysis makes correct predictions in JUA, where the relevant sonority distance (SD) for each sequence is indicated.
Another 85 sequences such as [bt, bṭ, bk] are predicted to be bad, but they occur in JUA. A different version of sonority-based analysis was introduced by Gouskova in 2004. This version will be tested in the following sub-section.

Testing Gouskova's Version of the Theory of Sonority Hierarchy
In this sub-section, I test Gouskova's (2004) model of the theory of sonority hierarchy to examine its capability to make correct prediction about permissible onset sequences in JUA. A finer-grained sonority scale, which is adapted from Jespersen's (1904) scale, is used in this model. The view of this model is that sequences have a universal sonority scale, shown in (6), in which sequences in the same column have the same sonority rise, which appears on the first top row. W in the sonority scale indicates a glide, R an [r]-like sound, L a lateral, N a nasal, Z a voiced fricative, S a voiceless fricative, D a voiced stop, and T a voiceless stop.

ARTICULATOR-BASED FEATURE ANALYSIS OF JUA ONSET SEQUENCES
In this section, I present the main aspects of articulator-based feature theory and I examine the capability of the theory to account for permissible onset sequences in JUA.

Aspects of Articulator-based Feature Theory
The view of articulator-based feature theory is that words are comprised of consonants and vowels which in turn are made of distinctive features. The present theory is different from Chomsky and Halle's (1968) feature theory in the sense that it distinguishes between articulators and features: 'articulators are movable parts in the vocal tract that participate in speech production. A feature is a gesture made by an articulator' (Duanmu, 2002: 6). Duanmu's feature geometry in (8) is based on the works of Clements (1985), Sagey (1986), Ladefoged andHalle (1988), McCarthy (1988), Steriade (1989), Kenstowicz (1994), and Halle (1995 Duanmu (2002: 8) clarifies the reason behind the combination of coronal and dorsal-[-back] for palatals. The first part accounts for alternation between palatals and coronals and 'the feature [-back] accounts for close interaction between front vowels and palatals.' A sound that has one place articulator is called a simple sound whereas a sound which has two or three place articulators is called a complex sound. For Duanmu (2002), when a sound has only one place articulator, this articulator is considered the primary articulator whereas when it has two or more articulators, the degree of closure is considered: the articulator that has greater constriction is considered the primary articulator and the one that has less constriction is considered the non-primary articulator 1 . If articulators have the same degree of closure, all of them are considered primary articulators.
For Duanmu, the two consonants that an onset sequence is comprised of occupy one slot -the onset. The two consonants comprise what is called a complex sound in the current theory. A complex sound can be thought of as a merger of two or more sounds (Duanmu, 2008). Consider the three cases in (10) (Duanmu, 2008: 25). (10) Art a -[-F i ] → ? For Duanmu, the two cases in (10a) and (10b) are accepted where two different articulators that make same or different gestures can do so simultaneously. The same is true for the articulator which makes two different gestures. However, the case in (10c) is unaccepted as a complex sound in Duanmu's current theory, as shown in (11). (11) No Contour Principle (Duanmu, 2008: 174) An articulator cannot make the same feature (F) twice within one sound. A complex sound can have more than one articulator at the same time. However, the No Contour Principle governs possible and impossible complex sounds in a language.

CONCLUSION
In this article I tested Kenstowicz's (1994) and Gouskova's (2004) two versions of sonority theory and Duanmu's (2002Duanmu's ( , 2008 articulator-based feature theory. The aim of the study was to examine to what extent the two versions of sonority theory in question can make correct predictions about permissible onset sequences in JUA. The study also aimed at investigating if the articulator-based feature theory could account for per-missible onset sequences in JUA. Findings have shown that the two versions of the sonority theory seemed to fail to predict some occurring sequences in JUA and to make wrong predictions about some other missing sequences. Some sequences violate the MSD, but they are found in JUA such as [sd] while some other sequences satisfy the MSD, but they are not found in JUA such as [lm]. Kenstowicz's version of the sonority theory makes 166 wrong predictions, where 81 onset sequences such as [ṭn] are predicted to be good, but they do not exist in JUA. The other 85 sequences such as [bt] are predicted to be bad, but they are found in JUA. Gouskova's version of the theory makes 155 wrong predictions, in which 145 sequences are predicted to be good, but they are not found in JUA. Other 10 sequences are predicted to be bad, but they exist in JUA. Findings of the study have also shown that the articulator-based feature theory failed to prove the capability to account for permissible onset sequences in JUA. It has been found that complex sounds which have two primary articulators violate the No Dual Place constraint. The two consonants involved in a complex sound such as [bz] Selkirk (1993), I opted for using the terms primary and non-primary in replacement for Duanmu's terms of major and minor, respectively.