At first glance, the Lao tone system seems to be a typical outcome of a messy tone split. Before the split, there were three tones (A, B and C) on syllables not ending in a plosive ('unchecked syllables', or 'live syllables' for those of use happy to calque the Thai terminology). Those ending in a plosive ('checked syllables', or 'dead syllables' in native terminology) can be treated as having tone D. Before the split, consonants could be split into 4 groups, numbered by Gedney as
- Voiceless aspirated stops, voiceless fricatives, and voiceless (or preaspirated) resonants. The written consonants corresponding to them are called 'high' consonants.
- Voiceless unaspirated stops
- Glottal stops and preglottalised consonants. The written consonants corresponding to Groups 2 and 3 are called 'middle' consonants.
- Voiced stops, fricatives and resonants. The written consonants corresponding to them are called the 'low' consonants.
The contrast of contrastive voicing was transferred to the tone and former voiced stops became voiceless aspirates. Preglottalised stops became voiced stops, and preglottalisation was lost. Just to complicate matters, the tones deriving from D also split according to the length of the vowel.
This process also occurred in Central Thai, which was and is a neighbour of Lao.
The correspondence sets are identified by old tone letter, consonant group number, and S or L for vowel length where relevant.
The reflexes of consonant groups 2 and 3 in Lao can be distinguished from those of consonant groups 1 and 4 in Lao - they are the unaspirated plosives plus /j/, while the others are aspirated stops, fricatives and resonants other than /j/. The distinction of middle consonants and non-middle consonants therefore applies to spoken Lao as well as to written Lao. In writing, old tone marks, which were abandoned in Lao and then reintroduced, supplement the distinction of high and low consonants to enable the tones to be recorded when writing the inherited vocabulary.
There are four tones on dead syllables - D123S, D4S, D1L (phonetically the same as C1) and D4L (phonetically the same as C4). Whether D123S and D4S are allophones of tones on live syllables is a matter of convenience.
With the exception of some dialects in Thailand, there are five or six tones on live syllables. The six tones are A1, A23, A4, B (split into B123 and B4 in parts of Thailand), C1 and C234. The tone of A23 may merge with other tones - typically A1 or A4.
So far, we have a straightforward picture, similar to Thai in principles but different in details. However, I have three problems with the descriptions.
The first problem comes when looking for phonemic contrasts. Regularly developed inherited vocabulary readily provides minimal pairs between the five A1, A4, B, C1 and C234 tones. However, it will not provide minimal pairs between the A23 tone and any of A1, A4 and C1, for the former occurs with initial middle consonants and the latter occur with initial non-middle consonants. Interestingly, two new tone marks were borrowed from Tai for use with middle consonants - mai ti for the A4 tone and mai catawa for the A1 tone. That implies a phonemic contrast with the A23 tone. Now, for the '6-tone' Vientiane accents, where A23 is a phonetically distinct tone from A1 and A4, the A23 tone (described as 'low') and the C1 tone (usually described as 'low falling') are similar if one ignores the glottalisation of the latter, so perhaps A23 and C1 are allophones. FWIW, the Becker books, which present a single accent, characterise both A23 and C1 as being the 'low tone'. Can anyone add more information on this situation?
As far as Vientiane speech is concerned, phonetically the A23 tone is merging into the A1 tone, with varying degrees of completeness.
Nick Enfield, who treats the merger of A23 and A1 as complete, presents a different picture. According to him, if one ignores atonal syllables, a live syllable with a middle-consonant initial must have one of the tones A1, B and C234. Furthermore, a dead syllable with a middle-consonant initial must have tone D1S if its vowel is short and tone D1L if its vowel is long. Finally, a dead syllable with a short vowel can only have tone D1S or D4S, and a dead syllable with a long vowel can only have tone D4S or D4L. While this is a fair description of the regularly developed inherited vocabulary, how reasonable is it for the entire vocabulary to follow such a straightjacket? It can be expressed as certain tone contrasts being neutralised on syllables starting with middle consonants. Is he just following a tradition, seen in Thai phonology, of ignoring the awkward exceptions? For Thai, I weep when I see people explaining why such real words as แร่ด /rɛ̂t/ 'pretentious; slutty' and ก๊าซ /káːt/ or /káːs/ 'gas (typically dispensed from a bottle)' do not exist.
I have one final question. In Vientiane, does D23L have the same tone as D1L or D4L? Almost everything I have seen, including p37 of Enfield's a Grammar of Lao, lead me to expect it to have the same tone as D1L. However, http://www.seasite.niu.edu/lao/lao3.htm shows that according to Crisfield, Hartmann and Enfield, it has the same tone as D4L. Hartmann seems to have been involved in the preparation of that page, so it doesn't appear to be a simple typographical error. As both the pair C1 and D1L and the pair C4 and D4L have what can be described as the same tone, and the two tones can be described as low falling and high falling, then if the contrast truly is neutralised for middle initial consonants, perhaps it is not unsurprising that the tone of D23L should switch from that of C1 to that of C234. However, for dead syllables, it does appear to be the first known case outside of southern Thailand of such a different tone split for short and long vowels. So, is the claim of a D1L v. D234L split for 'the Ventiane accent' actually correct?