the Great Tone Split does provide strong evidence for subgrouping languages and dialects in the Kam-Sui branch.
4.3 Divergent tonal development in Sui dialects
In all Sui dialects, odd-numbered yīn tones more often than not occur with voiceless or preglottalised
onsets. 78.3 of the words with yīn tones that we collected from Sandong have voiceless onsets. On the
other hand, only 52.3 of words with yáng tones have voiced onsets, indicating that devoicing is a more common historical sound change in Sui than voicing. Li Fang-kuei 1948:165 commented on this same
fact. However, we see that certain words bear odd-numbered
yīn tones in Yang’an Sui and all dialects of Kam, whereas the same words bear even-numbered yáng tones in all other Sui dialects. There are other
words for which the opposite is the case. Li Fang-kuei 1977a:36 observed a similar phenomenon among Taic languages: Po’ai sometimes exhibits yáng tones where Siamese and Lungchow have y
īn tones, and occasionally the opposite is evident. Li suggests that this tone alternation is probably due to an
alternation of voiced and voiceless initial consonants in Proto-Tai which led to divergent tones when the Great Tone Split occurred. However, he neither reconstructs the proto-onsets for such words at an earlier
stage nor identifies the conditioning factors which could have led to these tone alternations. Pittayaporn 2009 proposes a Proto-Tai sesquisyllabic form with voiced and voiceless elements which subsequently
monosyllabicised resulting in a low register reflex in some dialects and a high register reflex in others. A simpler solution would be to posit a Proto-Tai voiced initial which became voiceless after the tone split
in Po’ai but before the tone split in Siamese and Lungchow.
In Taic, most of these yīn-yáng tone alternations occur on words with voiceless onset reflexes. Unlike
Taic languages, Sui exhibits a set of contrasting yīn-yáng tone alternations which mainly go with voiced
onset reflexes, almost always liquids or nasals. Thus we propose a set of Proto-Kam-Sui voiceless initials which became voiced after the tone split in non-Yang’an Sui but before the tone split in Yang’an Sui and
Kam. In some cases, the voicing may have occurred due to a voiced presyllable in PKS which was subsequently lost.
4.3.1 Shared tonal developments in Yang’an Sui and Kam
Yang’an tonal developments show striking similarities to those seen in Kam dialects. In particular, our evidence indicates that the Great Tone Split occurred at the same point of time in both Kam and Yang’an
Sui in relation to certain onset developments which are shared by both and are unlikely to have happened independently. We therefore suggest that Yang’an “Sui” actually belongs to the Kam branch of
Kam-Sui along with Mulam and Then. Although Yang’an Sui now shares some tonal features with Sandong and Pandong Sui, thus “sounding” more like Sui than Kam to some extent, we show in section
4.5 that this is a result of areal diffusion of phonetic tonal traits which took place after Yang’an Sui speakers had migrated to the Sui area.
Table 4.2 shows the modern reflexes of PKS ʔl-, hl- and l- in the Sui dialects. Yang’an consistently
realises hl- as l- with a yáng tone, in common with Kam, Then and Mulam see Thurgood 1988:191. This suggests that the Great Tone Split occurred after the voicing of the initial in Yang’an Sui, Kam, Then
and Mulam, whereas the tone split occurred before the initial had become voiced in the rest of Sui. Ferlus 1996:274 posits PKS Cl- instead of Thurgood’s hl-, involving an unspecified preconsonant
which was preserved in non-Yang’an Sui until after the tone split.
Table 4.2. PKS ʔl-, hl- and l- reflexes in Sui and Kam yīn and yáng tones
are divided by double lines Sandong
Pandong Yang’an
Kam Gloss
PKS SD
DJ JQ
PD JL
TN BL
‘to wash clothes’
ʔlak
D
lɐk⁷ lak⁷
lɐk⁷ lak⁷
lɐk⁷ lak⁷
lak⁷ sak⁷
‘to rest’ -
lwa⁵ lwa⁵
la⁵ luə⁵
lɯə⁵ la⁵
la⁵ sa⁵
‘soup’ † lu⁵
lu⁵ lu⁵
lu⁵ saːu⁵
‘wind’ hlwum
A
- lom¹
- lom¹
lom¹ lom²
- ləm²
‘back of body’
hlaːi
A
laːi¹ laːi¹
laːi¹ -
laːi¹ laːi²
laːi² laːi²
‘book’ hle
A
lɛ¹ lɛ¹
lɛ¹ le¹
læ¹ lɛ²
lɛ² le²
‘to descend’
hlui
B
lui⁵ lui⁵
lui⁵ lui⁵
lui⁵ lui⁶
- lui⁶
‘to lick’ -
- -
lja¹ -
- lja²
lja² lja²
‘to steal’ hljak
D
ljɐk⁷ ljak⁷
ljɐk⁷ ljak⁷
ljɐk⁷ ljak⁸
ljak⁸ ljək⁸
‘chilli’ m-
hljaːn
B
ljaːn⁵ ljaːn⁵ ljaːn⁵ lien⁵ lin⁵
ljaːn⁶ ljaːn⁶ ljaːn⁶ ‘to forget’
laːm
A
laːm² laːm² laːm² laːm² laːm² laːm² laːm² laːm² ‘behind’
lun
A
lən² lən²
lən² lən²
lən² lən²
lən² lən²
‘belly’ luŋ
A
lɔŋ² lɔːŋ²
lɔːŋ² lɔɐŋ² lɔɐŋ² lɔːŋ²
lɔŋ² loŋ²
† Sui data from Zeng 2004 Table 4.3 gives a set of correspondences which is more difficult to explain. In the words for
‘intestines’, ‘tail’, ‘to sit’ and ‘place’, Yang’an along with all Kam dialects exhibits yīn tones with voiced
initials whereas the rest of Sui for these words exhibits yáng tones with voiceless initials. In both cases, this voicing-tone association is the opposite of what one would expect, since voiced initials originally
triggered yáng tones. Thurgood’s reconstruction does not account for this anomaly. Ferlus 1996:268 suggests PKS
kⁿs- for ‘intestines’ and ‘tail’ based partly on the respective Mulam reflexes kʰɣaːi³ and kʰɣat⁷. In this scenario, the nasal component of the presyllable kⁿ- may have induced the yáng tones in
non-Yang’an Sui after the initial k- was lost. Note that s- h- is a well-attested sound change in Sui see chapter 5, section 5.2.5.3. An s- Kam, l- Sui, h
ɣ- Mulam alternation is also attested in words such as ‘soup’ table 4.2.
Table 4.3. PKS dz-, khj-, kj- y īn and yáng tones are divided by double lines
Sandong Pandong
Yang’an Kam
Gloss PKS
SD DJ
JQ PD
JL TN
BL ‘worm’
dzan
C
hɐn⁴ han⁴
hɐn⁴ han⁴
hɐn⁴ han⁴
hɐn⁴ san⁴
‘snake’ dzuːi
A
hoi² hui²
hui² hui²
hui² hui²
hui² sui²
‘son-in-law’ -
haːu⁴ haːu⁴ haːu⁴ haːu⁴ haːu⁴ haːu⁴ haːu⁴ saːu⁴ ‘intestines’
khjaːi
C
haːi⁴ haːi⁴
haːi⁴ haːi⁴
haːi⁴ laːi³
laːi³ saːi³
‘tail’ khjut
D
hət⁸ hət⁸
xət⁸ hət⁸
hət⁸ lət⁷
lət⁷ sət⁷
‘to sit’ dzuːi
B
hoi⁶ hui⁶
hui⁶ hui⁶
hui⁶ lui⁵
lui⁵ sui⁵
‘place’ -
hən² hən²
xən² hən²
hən² lən¹
lən¹ sən¹
‘to give’ khjaːi
A
haːi¹ haːi¹
haːi¹ haːi¹
haːi¹ haːi¹
haːi¹ sai¹
‘early morning’
khjam
A
hɐm¹ ham¹ hɐm¹ ham¹ ham¹ ham¹ hɐm¹ sam¹ ‘to catch’
kjap
D
hɐp⁷ hap⁷
hɐp⁷ hap⁷
hɐp⁷ hap⁷
hɐp⁷ sap⁷
‘root’ kjaːŋ
A
haːŋ¹ haːŋ¹ haːŋ¹ haːŋ¹ kau³ haːŋ¹ haːŋ¹ saŋ¹
Yang’an Sui, and occasionally Pandong Sui too, exhibits yīn-yáng tone correspondences that match
Kam for a subset of words with nasal onsets. Table 4.4 shows correspondences for PKS n- and ŋj-. For
some of these words, Yang’an Sui and Kam bear yīn tones suggesting a voiceless proto-onset whereas
the rest of Sui and Maonan and Mulam, see ILCRD 1996 bear yáng tones. Ferlus 1996 proposes the following proto-onsets for two of the words: Cm- for ‘louse’ and hnr- for ‘moon’. For ‘dew’, Pittayaporn
2009 posits Proto-Tai C̬.nwa:j
A
, presumably cognate with the Kam-Sui word. Whatever the precise origins, the data indicate that the Proto-Kam-Sui onset had a voiceless or glottal component which was
lost before the tone split induced yáng tones in most of the Sui dialects.
Table 4.4. PKS n- , ŋj- yīn and yáng tones are divided by double lines
Sandong Pandong
Yang’an Kam
Gloss PKS
SD DJ
JQ PD
JL TN
BL ‘meat’
naːn
C
naːn⁴ naːn⁴ naːn⁴ naːn⁴ naːn⁴ naːn⁴ naːn⁴ naːn⁴ ‘insect’
- nui²
nui² -
nui² nui²
nui² nui²
ni² ‘to smell’
nun
C
nən⁴ nən⁴
nən⁴ nən⁴
nən⁴ nən⁴
nən⁴ nən⁴
‘flower’ nuk
D
nʊk⁸ nʊk⁸
nʊk⁸ nok⁸
nɔɐk⁸ nʊk⁷ nʊk⁷
nuk⁹ ‘louse
body’ nan
A
nɐn² nan²
nɐn² nan²
nɐn² nan¹
nɐn¹ nan¹
‘to weed’ ne
A
ne² ne²
ne² ne²
næ² nɛ¹
nɛ¹ ne¹
‘moon’ nüaːn
A
njen² niɛn² njaːn² nien² nin²
njaːn¹ njaːn¹ ȵaːn¹ ‘dew’
- ȵi²
ȵi² ȵe²
ni² ȵi²
mɛ¹ mɛ¹
me¹ † ‘last year’ ‡ -
ȵu² -
ȵu² SQ
ȵu² JC
- ȵu¹
YC -
- ‘erhu’ §
- ȵen²
- -
- -
- -
jen¹ ‘to stay’
ŋjaːu
B
ȵaːu⁶ ȵaːu⁶ ȵaːu⁶ ȵaːu⁶ ȵaːu⁶ ȵaːu⁶ ȵaːu⁶ ȵaːu⁶ ‘to be
willing’ -
ȵɔn⁶ -
ȵuɐn⁶ ȵʉən⁶ ȵuən⁶ ȵɔːn⁶ ȵuɔn⁶ ȵon⁶ † Kam word means ‘frost’.
‡ Data from Zeng 2004. § Sui data from Zeng andYao 1996 and author’s field notes.
Words with uvular and simple glottal stop many of which originated in a uvular stop onsets exhibit similar tone alternations. Words with PKS onset
hŋl- bear even-numbered yáng tones with voiceless onsets in Yang’an Sui and Kam dialects, and odd-numbered
yīn tones with voiced onsets in most of the Sui dialects. Correspondences are shown in table 4.5. As with the examples in table 4.3
above, this is the opposite of what one would expect see also table 4.7 below. The l- onset in some Kam dialects for ‘mushroom’ and ‘shoulder pole’ provides the clue, indicating that Yang’an Sui, along with
Kam, had lost the initial h- by the time of the tone split, leading to yáng tones, after which ŋl- became
q- or ʔ- in Yang’an and most Kam dialects. On the other hand, other Sui dialects retained the voiceless
onset hŋl- at the time of the tone split, leading to yīn tones.
Table 4.5. Tone alternations on words with uvular and glottal onsets y īn and yáng tones are divided by
double lines; A, B or C tone category differences are shaded in grey Sandong
Pandong Yang’an
Kam Gloss
PKS SD
DJ JQ
JL PD
TN BL
‘to carry over one’s
shoulder’ -
ɴɢaːi⁵ ɴɢaːi⁵ ʁaːi⁵ -
ɣaːi⁵ ɴɢaːi⁵ ɴɢaːi⁵ -
‘ramie’ † -
ɴɢaːn¹ - ʁaːn¹ -
- ɴɢaːn¹ -
ʔaːn¹ ‘saliva’
- ɴɢe¹
ɴɢɛ¹ ʁɛ¹
- -
ŋɛ² ŋɛ²
mje², ȵe² ‘mushroom’
hŋla
A
ɴɢa¹ ɴɢa¹
ʁa¹ ʁa¹
ɣa¹ qa²
qa² la², ʔa²
‘shoulder pole’
hŋlaːn
A
ɴɢaːn¹ ɴɢaːn¹ ʁɐːn¹ ʁaːn¹ ɣaːn¹ qaːn² qaːn²
laːn², ʔaːn² ‘young hen’
hŋlaːŋ
B
ɴɢaːŋ⁵ ɴɢaːŋ⁵ ʁaːŋ⁵ ʁaːŋ⁵ ɣaːŋ⁵ qaːŋ⁶ qaːŋ⁶
ʔaːŋ⁶ ‘hai
亥 earthly
branch’ ‡ EMC
ɣəj
B
ɴɢaːi³ - -
ʔaːi³ -
qaːi⁴ -
ʔəi⁴ ‘in inside’
- ɴɢaːu³ ɴɢaːu³ ʔaːu³ ʔaːu³ ʔaːu⁴
qaːu⁴ qaːu⁴
ʔaːu⁴ ‘bitter’
kam
A
qɐm¹ qam¹
qɐm¹ qam² qam² qam² qɐm²
ʔam² ‘sugarcane’
ʔuːi
C
ʔoi³ ʔui³
ʔɔi³ -
ʔoːi⁴ ʔõi⁴
ʔõi⁴ -
‘neck’ -
qɔ⁴ qɔ⁴
qo⁴ qo⁴
qɔ⁴ qo²
qɔ² ʔəu²
‘grandfather’ - qɔŋ⁵
qɔːŋ⁵ qɔːŋ⁵ qɔɐŋ⁵ qɔɐŋ⁵ qɔːŋ³
qɔŋ³ ʔoŋ³
‘crow’ ka
A
qa¹ qa¹
qa¹ qa¹
qa¹ qa¹
qa¹ ʔa¹
‘chicken’ kaːi
B
qaːi⁵ qaːi⁵
qaːi⁵ qaːi⁵ qaːi⁵ qaːi⁵
qaːi⁵ ʔaːi⁵
‘slanted’ § -
qaːi² -
qaːi² - qaːi²
qaːi² -
ʔaːi² ‘excrement’
ke
C
qe⁴ qɛ⁴
qɛ⁴ ȶe⁴
ke⁴ qɛ⁴
qɛ⁴ ʔe⁴
† Sui data from Zeng 2004:274. ‡ Sui data from Zeng 1994:83 and Burkiewicz 2012. This is an old Chinese loanword.
§ Sui data from Zeng 2004:257.
Similar alternations on ‘in’, ‘bitter’ and ‘sugarcane’ indicate that the Proto-Kam-Sui form must have had a voiced element possibly part of a sesquisyllabic form which induced yáng tones in Pandong Sui,
Yang’an Sui and Kam. The ɴɢ- pronounced as ʁ- by some speakers and yīn tone combination in Central
and Eastern Sandong Sui dialects in ‘hai’ and ‘in’ suggest that the Great Tone Split was perhaps not as straightforward as a simple voiceless onset triggering high register and a voiced onset triggering low
register. There may have been other conditioning factors, such as the presence of a presyllable of a certain phonation type, that prevented words from acquiring yáng low-register tones. This possibility was
raised by Edmondson and Yang 1988:163 who suggested that preconsonants may have had “the effect of causing high tones irrespective of their own voicing state… another tonogenetic mechanism in the
history of Kam-Sui”.
More unusual are the contrasting tone categories on ‘neck’ and ‘grandfather’— apparently rare instances of words acquiring different tone categories in different dialects during the original tonogenetic
processes.
3
With these words, as with words showing yīn- yáng tone alternations, Yang’an Sui is entirely
consistent with Kam in the development of its tone categories.
4.3.2 Shared tonal developments in Pandong, Western, Yang’an Sui and Kam