Cluster simplification in several Malay dialects

Alan Lee

University of Pennsylvania

While some dialects of Malay allow nasal+stop clusters in root environments, certain others exhibit a tendency to simplify such clusters.  This simplification moreover follows a certain pattern, descriptively summarized below (see page 2 for data):

(A)            nasal+voiced stop clusters reduce in favor of the nasal.

(B)       nasal+voiceless stop clusters reduce in favor of the voiceless stop. 

            (C)             Not observed:             nasal+voiced stop clusters reducing into the voiced stop

                                                nasal+voiceless stop clusters reducing into the nasal

The above phenomena raise two related questions: why the outcome of the cluster reductions should differ depending on the voicing of the stop, and why the logically possible type (C) reductions are not found in the Malay dialects surveyed.

In this paper, I propose an analysis using Optimality Theory that answers the above questions.  My point of departure is the observation in Pater (1997) that languages which simplify nasal+voiceless stop clusters inevitably delete the nasal (e.g. Kelantan Malay, Venda, Swahili, etc.). To explain this nasal-obstruent asymmetry, Pater proposes an intrinsic ranking of segmental deletion: OBSMAX (a constraint against obstruent deletion) >> NASMAX (a constraint against nasal deletion).  This ranking ensures that the stop and not the nasal is deleted in the presence of a higher-ranking constraint disallowing nasal+stop clusters.  (For this paper, I disallow such clusters using an Identical Cluster Constraint on nasality (ICC[N]) which states that consonant clusters should be identical in nasality).

While ICC[N], OBSMAX >> NASMAX accounts for the type (B) case, it appears problematic for the type (A) case, where the nasal appears to be preserved at the expense of the stop.  Pater's insight can nevertheless be preserved with a more careful analysis.  In the (A) case, it can be shown that there is in fact no segmental deletion.  Instead what we have is segmental fusion between the nasal and the voiced stop, (subscripts are used to indicate corresponding segments, a la Pater (1997)):  /lem₁b₂u/   ->    [lem_1,2u]

Two observations support this segmental fusion account: i) careful phonetic transcriptions of fieldwork data on Malay dialects often show a nasal segment with a slight oral closure when there is an underlying nasal+voiced stop cluster; ii) dialects which show nasal harmony appear does not allow nasal spreading when nasal+voiced stop segmental fusion has taken place, suggesting that the oral stop is active in the phonology, preventing the nasalization of the following vowel.

To explain why segmental fusion takes place with nasal+voiced stop clusters but not with nasal+voiceless stop clusters, I propose an IDENT [VOICE] constraint which militates against segmental fusion if the fused output contains a specification for voice which differs from the underlying representation.  Hence (i) below does not violate IDENT[VOICE] but (ii) does:

(i)     /m + b/  ->   [m]       (ii) /m + p/   ->   [m]

             |         |                   |          |   

            +V        +V                  -V         +V  

For the Malay dialects under discussion, a ranking of ICC[N], IDENT[VOICE], OBSMAX >> NASMAX ensures that the type (A) and (B) patterns emerge (see tableaux on page 2).  I then conclude this presentation by examining possible phonetic and typological support for the pattern of cluster simplification discussed here.

DATA (note /ng/ refer to the velar nasal):

TYPE (A)            nasal+voiced stop clusters reduce in favor of the nasal.

            e.g.,             /lembu/ -> lemu 'cow'

                             /mandi/ -> mani 'to bathe'

                        /panggang/ -> pangang 'to roast'

                Observed or reported for the dialects of Kelantan (Ajid Che Kob 1985), Patani,

Langkawi (Collins 1986), Urak Lawoi, Kedah, Terengganu (Asmah Haji Omar 1995), Ulu Muar Malay (Onn 1980,), and the Bangka subdialects of Gadung, Gunungmuda (Nothofer 1997)

TYPE (B)             nasal+voiceless stop clusters reduce in favor of the voiceless stop.   

            e.g.,              /kampong/ -> kapong 'village'

                       /bintang/ -> bitang 'star'

                           /tongkat/  ->  tukat 'walking cane'

Observed for the dialects of Kelantan (Ajid Che Kob 1985), Urak Lawoi, Terengganu (Asmah Haji Omar 1995), the Bangka subdialect of Kacung (Nothofer 1997)

REFERENCES:

Ajid Che Kob 1985. Dialek Geografi Pasir Mas. Bangi: Penerbit Universiti Kebangsaan Malaysia. 

Asmah Haji Omar 1995. Rekonstruksi Fonologi Bahasa Melayu Induk. Kuala Lumpur: Dewan Bahasa dan Pustaka.

Collins, James 1986. Vokal Sengau Dialek Melayu Kedah (Langkawi). In James Collins,

Antologi Kajian Dialek Melayu, 3-19. Kuala Lumpur: Dewan Bahasa dan Pustaka.

Farid M. Onn 1980.  Aspects of Malay Phonology and Morphology. Bangi: Penerbit Universiti Kebangsaan Malaysia. [Published version of 1976 University of Illinois dissertation]

Nothofer, Bernd 1997. Dialek Melayu Bangka. Bangi: Penerbit University Kebangsaan Malaysia.

Pater, Joe 1997. Austronesian Nasal Substitution and other NC effects. In Kager et al,

The Prosody Morphology Interface, Cambridge: Cambridge University Press.