This paper presents a preliminary grammatical sketch of the major grammatical features of Siak Malay, based on several years of fieldwork in the region.

Before embarking on the description of any Malayic variety, it is first necessary to locate it in geographical and sociolinguistic space, in order to differentiate it from other varieties which may be spoken in close proximity. Siak Malay is the variety of Malay spoken in those parts of Riau province near the downstream portions of the Siak river, as delimited by the small towns of Siak Sri Indrapura, Sungai Apit, Sungai Pakning and Bengkalis. Although many of its speakers are ethnic Malays, it is also spoken by most of the non-Malay migrants to the region, consisting primarily of Javanese, Minangkabau, Batak and Chinese. Siak Malay is considered by its speakers to be the local "bahasa daerah", or "regional language"; its status is thus similar to other "bahasa daerah" such as Javanese, Bugis, Dani, and so forth, even though its speakers are aware that it is much closer to the national language, Bahasa Indonesia, than are most other "bahasa daerah". Indeed, Siak Malay is quite distinct not only from Standard Indonesian but also from the local variety of colloquial Indonesian, namely Riau Indonesian. Nevertheless, most speakers of Siak Malay are fluent also in Riau Indonesian, and frequently code-switch between these two varieties of Malay/ Indonesian.

Phonologically, Siak Malay has a number of peculiar features which distinguish it from Riau Indonesian, as also from other varieties of Malay and Indonesian. Siak Malay distinguishes 8 vowel phonemes, with contrasting high-mid and low-mid vowels, e and o vs. E and O. In general, the high-mid vowels are much more common than their low-mid counterparts: e and o are the usual reflexes of Riau Indonesian e and o, eg. beso? ~ beso? "tomorrow"; also word final -o is the reflex of Riau Indonesian word-final -a, eg. mato ~ mata "eye". Low-mid E and O tend to reflect loans from Indonesian, eg. OjEk "motorcycle taxi", or the results of (synchronic or diachronic) phonological processes - for example, Riau Indonesian word-final -ar corresponds systematically to Siak Malay -O, eg. bsar ~ b´sO "big". However, in a few instances, the distribution of high vs. low mid vowels seems unpredictable, either synchronically or diachronically, eg. pEnde? "short". Siak Malay also has a series of nasalized vowels, however the occurrence of these is for the most part phonologically conditioned, occurring in the final syllable of a word if that syllable begins with an underlying glottal stop, eg. pu?ãt "Fuad" (but not aap "Aap" and other similar names whose derivation involves reduplication of a vowel followed by insertion of a phonetic glottal stop). An exception to this phonological conditioning is provided by the high-frequency deictic final particle hã; indeed, in some cases, nasalization spreads backwards from the particle hã, imbuing deictic sentences with what seems to be a kind of nasal prosody.

In terms of its morphosyntactic structure, the most salient characteristic of Siak Malay is the radical underdifferentiation of a large number of grammatical categories familiar from other languages. In this respect, Siak Malay resembles Riau Indonesian, thereby further supporting the claim that structural simplicity is not necessarily indicative of a contact language with a history of radical restructuring. Nevertheless, the grammar of Siak Malay does differ from that of Riau Indonesian in a number of interesting ways. One is the higher frequency of the patient- and agent-oriented markers di- and N- - though their grammatical properties are the same as in Riau Indonesian. Another is the higher frequency and wider functional range of the associative enclitic -nyo. An alterrnative morphophonemic realization of this enlicitc is -n, resulting in forms such as dapatnyo ~ dapatn "get-ASSOC", which poses an interesting puzzle. Except for the above, sequences of the form -CN are impermissble word-finally. However, in the neighboring dialects of the Orang Asli, such sequences occur as prosodically-conditioned phonological alternants of word-final nasals, eg. makan ~ makatn "eat". Although formally quite different, the phonetic similarity plus the geographical proximity of these two phenomena make it hard to believe that they are not somehow related, though until now, I have not been able to construct a plausible diachronic scenario accounting for the relationship.