Stops and affricates

In non-final position, /t/ and /t^h/ are denti-alveolar or, more accurately, apico-dental lamino-alveolar plosives (although not marked in our transcription). The voiced plosive /d/ is apico-alveolar. The alveolo-palatal affricates /tɕ tɕ^h dʑ/ are produced with laminal contact in the anterior of the palatal area. Our introduction of a series of affricates, based on phonetic evidence, is a departure from the usual treatment of these segments in Aslian (and other Mon-Khmer languages) as palatal plosives /c ɟ/ with fricated allophones in non-final position [c^ʑ ɟ^ʑ] (see, for instance, Burenhult Reference Burenhult2001 for Jahai, and Kruspe Reference Kruspe2004 for Semelai, both Aslian).

The aspirated voiceless plosives /p^h t^h tɕ^h k^h/ are infrequent in initial position, and for most speakers are in free variation with the unaspirated stops, e.g. /t^hi/ ~ /ti/ ‘hand’.

In final position, the plosives /p t k/ are always glottalised [] and usually also checked (see below for further discussion). The denti-alveolar plosive may be realised as apico-alveolar in final position, possibly due to the constriction of the co-occurring glottal stop. The alveolo-palatal affricate has no fricative release in final position and is realised as a glottalised alveolo-palatal plosive, e.g. /ɰətɕ/ [ɰⁱ] ‘type of bird’, and also with occasional final aspiration when the syllable onset is a fricative, e.g. /stɕ/ [s^ih] ‘endpoint’.

We adopt a fronted voiceless glottalised palatal plosive [] to represent the basic final allophone here.

We also note the presence of nasal pre-plosion on final plosives /p t k/ and the final affricate /tɕ/ when preceded by a nasal vowel. The plosive is usually realised phonetically as glottalised, e.g. /hã p/ [hã^m] ‘not have, not exist’, but given that in most instances the final oral plosive is less audible it is represented here by a superscript: /mt/ [mⁿ] ‘eye’.

Fricatives

In word-final position the alveolar fricative /s/ has an allophone [], a short pre-palatalised glottal fricative, e.g. /ləpas/ [ləpa] ‘after’. Such an allophone is commonly reported in Mon-Khmer languages elsewhere, e.g. Halang (Cooper & Cooper Reference Cooper and Cooper1966) and Mnong Rolom (Blood Reference Blood1976) spoken in Vietnam. In our language, its onset can have a wide constriction line extending from the palatal to alveolar region. The tip of the tongue is lowered, and the body is raised. Lingual articulation is usually back, but occasionally more fronted extending almost to the apical region. Although primarily realised as [], variation is encountered and it may occur as a spread pre-palatal or alveolo-palatal fricative [] with optional supralaryngeal frication.

Nasals

Nasals exhibit a voicing distinction at each point of articulation: bilabial /m /, apico-alveolar /n /, lamino-palatal /ɲ / and dorso-velar /ŋ /. The voiceless nasals always have a voiced offset; the onset varies between voicelessness and pre-aspiration of the nasal, e.g. /əh/ [^dʑh] ~ [^hdʑh] ‘be heavy’.

At each point of articulation before oral vowels, there are orally released voiced and voiceless post-ploded allophones of the nasals [m^{b b}], [n^{d d}], [^dʑ] and [ŋ^{ɡ ɡ}], except for the alveolo-palatal point which lacks a voiced post-ploded counterpart, e.g. /mo/ [m^boː] ‘too’, /ɔ/ [^bɔː] ‘breakline’, /nəl/ [n^dəl] ‘a bit’, /p/ [^d] ‘to be ripe’, /ŋəʔ/ [ŋ^ɡəʔ] ‘elder (voc.)’, /əʔ/ [^ɡəʔ] ‘to be partially ripe’. The presence of voiced post-ploded offset appears to result from a continuation in supralaryngeal constriction after closure of the nasal passage prior to the onset of the following vowel. This oral post-plosion, particularly in connected or rapid speech, is often, however, difficult to perceive, although the following vowel always remains fully oral, e.g. [ŋəʔ] alongside more careful [ŋ^ɡəʔ] for /ŋəʔ/ ‘elder (voc.)’. The presence of post-ploded nasals in Mah Meri appears to be part of a wider areal phenomenon in the region, including neighbouring Sumatra (see Eades & Hajek Reference Eades and Hajek2006 for details).

In word-initial position, voiceless nasals are syllabic before a following homorganic stop, e.g. /kiʔ/ [ˈkiʔ] ‘3rd person pronoun’.

Other consonants

/ɾ/ in the onset of the syllable varies from a tap [ɾ] to a light trill [r] to an alveolar approximant [ɹ]. For some speakers, when intervocalic, it occurs in free variation with the velar approximant /ɰ/: /sərk/ [səɾ] ~ [səɰ] ‘to be tall’. In rare syllable-final position, /r/ is a strident trill (not specifically marked here), e.g. /ʲər/ [ʲr] ‘goose bumps’.

/l/ is apico-alveolar. Its voiceless counterpart // is confined to word-initial position: /uh/ [uh] ‘storm’. As with the voiceless nasals, phonetically this segment fluctuates between a voiceless [] or pre-aspirated onset [^ʰ], e.g. /k/ [] ~ [^ʰ] ‘eagle’. However, the lateral offset is always voiced.

The velar approximant /ɰ/ has a labialised allophone [ɰ^w] before rounded vowels, e.g. /ɰɔh/ [ɰ^wɔh] ‘branch’.

Register 1

From an auditory perspective, Register 1 vowels are typically, although not always, characterised by a clear tense voice quality, shorter duration, and lower pitch. They are also subject to fewer phonotactic restrictions: (i) they occur in both final and non-final syllables, (ii) a phonemic oral versus nasal distinction is found for all vowels, with the exception of the central unrounded vowel // and (iii) in word-final position they are found before a larger inventory of coda consonants, when compared to Register 2 vowels.

Register 2

Register 2 vowels give the impression of being laxer, longer and higher in pitch than their Register 1 counterparts. They often have a breathy articulation, which is most clearly audible in the lower vowels, but less so as vowel height increases. Overall, breathy voice in Mah Meri is always realised as slight breathy voice, never strong or heavy.

Register 2 vowels are restricted to the final syllable, for which the only possible coda is a voiceless plosive or voiceless affricate. The consonant in question has a muted articulation and although it may exhibit simultaneous glottalisation or checking like plosives following Register 1 vowels, e.g. /wk/ [w] ‘person (classifier)’, it may also appear unchecked as [w].

The perceptual differences between Register 1 and 2 are often very subtle, even in careful speech. In connected speech, they do not seem to be particularly noticeable. Preliminary acoustic investigation reported by Stevens et al. (Reference Stevens, Kruspe, Hajek, Hoffmann and Mixdorff2006) of register in the speech of our Mah Meri speaker confirms our difficulty in reliably identifying and separating the two registers, even though they are used contrastively in well-established minimal pairs, e.g. /luwat/ ‘mangrove worm’ vs. /luwt/ ‘front’.

Contrary to expectations, duration was not at all a reliable cue across the eight vowel pairs examined, e.g. /u/ was much shorter than // (−86 ms) but /a/ was much longer than // (+91.5 ms). Moreover, although spectral appearance of Register 2 tended to show less defined formants, with some additional weakening of second and higher formants, the pattern was not always consistent. With respect to fundamental frequency, this was found, as predicted, to be higher on average in Register 2 at each of the three measurements points (vowel onset, midpoint, and offset). While the effect was the most consistent of the ones identified and then inspected, it was nevertheless not significant at any point in the vowel (p > 0.05) and also varied in degree across vowel pairs.

Nasality

Nasality is not predictable, given the presence of contrastive nasal vowels, and there is no evidence of nasal spread across glottals, e.g. /mã -ʔətɕ/ [mã -ʔⁱ] ‘to defecate’. Apparent left-to-right spread when the onset of the final syllable is a glottal phoneme were disproved. Syllable cutting tests reveal that the final vowel is phonemically nasal, e.g. /mʔh/ [mʔh] ‘to feel distended’.

It is worth noting that many speakers, particularly women, exhibit a high-pitched nasalised speech style often resulting in the neutralization of the phonemic distinction between oral and nasal vowels in natural speech.

Stress and syllable weight

Word-level stress is non-distinctive and predictable: it is fixed on the final syllable which in turn is always bimoraic (heavy), e.g. /luˈwat/ ‘mangrove worm’. As a result, stressed final vowels in open syllables are lengthened – they are usually fully long, but can also appear optionally as semi-long e.g. /ka/ [ˈkaː] ~ [ˈkaˑ] ‘fish’.