LANGUAGE AND LINGUISTICS 13.1:171-192, 2012 2012-0-013-001-000299-1 The Characteristics of the Burmic Family of Tibeto-Burman* David Bradley La Trobe University The history of the various names used for the Burmic branch of TibetoBurman (TB) and its components are outlined. Some salient phonological developments which differentiate Burmic from other TB subgroups and some further internal developments are briey exemplied. The morphology reconstructible for Burmic, which is much less extensive than that found in many other TB branches, is discussed. The main syntactic characteristics of Burmic, typical of verb-nal TB, are given, along with the reconstructed forms for various syntactic markers. Some examples of lexical differences providing evidence for Burmic as a separate group within TB and for subgrouping within Burmic are presented. The paper also provides the Burmic forms for the Swadesh 100-word list in an appendix, giving reconstructed etyma and forms from Burmese and from representative languages in the three main branches of Ngwi. Examples are cited below with reference to the numbers of the forms in this appendix. Key words: Tibeto-Burman, Burmic subgroup, Ngwi, phonological, grammatical & lexical reconstruction 1. Terminology The names used for this component of TB differ, but there is general agreement in all classications of TB and/or Sino-Tibetan (ST) ! Shafer (1966-1974), Benedict (1972), Bradley (1997), van Driem (1997), Matisoff (2003) and so on!about its status as a major and cohesive group within TB and/or ST. There is also general agreement that it has two main subgroups. For the whole group, Shafer (1966-1974) used the term Burmic, Benedict (1972) preferred Burmese-Lolo, and Burling (1967) and many subsequent scholars use the term Lolo-Burmese. Chinese scholars use the term Mian-Yi, the rst derived from the Chinese term for Burmese, ultimately derived from the !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! * I am very pleased to thank Academia Sinica and the organizers for convening the June 2010 International Symposium on Sino-Tibetan Comparative Studies in the 21st Century, at which this article was rst presented. I also thank the many participants who gave valuable feedback at the symposium and subsequently the anonymous referees who have provided detailed comments.! David Bradley Burmese autonym Mran as reected in Myanmar, and the second from Yi, the post1950 replacement said to have been chosen by Chairman Mao to replace the former Chinese term Lolo and based on another term formerly used to refer mainly to Dai groups including peoples now classied as Zhuang and Buyi in China. The two subgroups are most often called Burmish and Loloish, following Shafer. Chinese scholars call the former the Mian Branch and the latter the Yi Branch. Recognizing the negative view of the term Lolo held by some speakers and scholars in China, the blended term Yipho was coined by Benedict at a conference in Xichang, Liangshan, Sichuan in 1992, combining the new Chinese term Yi and the Nosu form of the widespread TB male sufx, a cognate derived from Burmic *!pa2. Instead of the problematic exonymic terms Loloish, Yi Branch or Yipho, I prefer to use a term derived from what was the probable autonym of these groups, Ngwi, a cognate of the widespread TB etymon for ‘silver’ (Bradley 2005:164-166), which is not attested within Ngwi in its original meaning. 2. Phonology The clearest and most strongly-attested innovative characteristics of Burmic are phonological, as laid out in detail in Bradley (1979a). These include the development of a third reconstructed tone category in non-stop nal rhymes throughout Burmic (Burling 1967) and the presence of prenasalized stop and affricate initials *mb *nd *nts *"c *!g in numerous etyma. Due to phonological developments, the latter is not supported by Burmese forms, in which these have merged with other *voiced initials, for example into the voiceless unaspirated initials of Burmese; this merger is restricted to Burmese itself, and is not present in other Burmish languages. The subsequent redevelopment of voiced stop/affricate initials in Burmese is somewhat irregular but conditioned primarily by medial juncture voicing; for example, 38 ‘head’, written u2 kho! is normally /gã!42/ in spoken Burmese, with the initial of the second syllable voiced and then the rst syllable, the TB cognate, lost. Another example where the rst syllable is still present is 42 ‘mouth’, written pa2 cap, spoken Burmese /b"za!/. Some of these voiced initials are now written with voiced initials, but most are still written with voiceless unaspirated or aspirated initials. Thus, the voicing development must have started before the Burmese orthography stabilized about 800 years ago; some of this voicing is already seen in the earliest Burmese inscription from 1112 AD, as in the noun compound ma!.gri ‘great king’, the postverbal bri perfective marker, and so on, which are written morphophonemically with voiceless initials in modern Burmese. Some grammatical forms such as spoken Burmese 4 ‘this’ and 7 ‘what’ and a few frequent lexical items such as 172 ! The Characteristics of the Burmic Family of Tibeto-Burman ‘market’ /ze42/ are now written with voicing, perhaps reecting lost former rst syllables and the fact that the change started or was most salient in frequent words. *Tone 3 is found in a relatively small proportion of the etyma reconstructed with non-stop nal rhymes: 13 per cent of securely-reconstructed etyma (Bradley 1979a:228) and a slightly higher proportion of less-securely reconstructed etyma. Etyma reconstructed with *Tone 1 and *Tone 2 are both about the same proportion of such etyma, over 40 per cent each. There are a few etyma which differ between Burmish and Ngwi as to their membership in the *Tone 3 category; for example, 59 ‘know’ reects *Tone 3 in Burmic but *Tone 2 in Ngwi; 36 ‘feather’ is the reverse, *Tone 2 in Burmish, *Tone 3 in Ngwi. The etyma exemplied below which reect *Tone 3 in both branches of Burmic are 3 ‘we (inclusive)’, 33 ‘egg’, 73 ‘moon’, 95 ‘full’ and 99 ‘dry’, among many other TB cognates. This is one of the strongest and most consistent Burmic phonological features, absent from languages which have sometimes been subgrouped with Burmic, such as Naxi (Bradley 1975), Bai and Tujia. The prenasalized stops are always homorganic in Burmic, whatever their TB origins. They occur in a large number of solid TB etyma where no obvious TB nasal prex source can be discerned, such as 54 ‘drink’, TB (Matisoff 2003) *da!, Burmic *nda!1, and 95 ‘full’, TB (Matisoff 2003) *bli!/pli!, Burmic *mbli!3; as well as in some innovative etyma, such as ‘buckwheat’ *!ga2. In the Qiangic component of TB, there are also numerous etyma with prenasalized stops, not always homorganic. However, the etyma with prenasalized stops in Qiangic and Burmic are mainly different; the distribution of prenasalized stops in etyma is a criterion for linking the Na languages Naxi, Na and Namuyi with Qiangic rather than Burmic (Bradley 1975, 2008a). There are also innovations within one or other subgroup of Burmic. One wellknown example is the development of a tonal contrast in syllables with stop-nal rhymes in Ngwi, with a higher tone conditioned by a *voiceless initial and a lower tone conditioned by a *voiced initial (Matisoff 1972). In most languages, cognates of the *H stopped tone have a similar pitch value to *Tone 3, most often mid level; while cognates of the *L stopped tone most often have a low pitch value similar to one widespread reex of *Tone 2 in many Ngwi languages (or *Tone 1 in some others). Both *H and *L most often remain distinct from the non-stop nal tones in modern Ngwi languages by retaining a nal glottal stop, creaky or harsh phonation, or by pitch and contour innovations such as becoming the sole high tone (Nosu *L > 55), becoming a new contour tone (some Lahu *L > 35) and so on. Burmish languages other than Burmese show a variety of separate splits in syllables with stop-nal rhymes; Burmese itself has no split, all have a high short tone with a nal glottal stop. Tonal splits in syllables with stop-nal rhymes seen in languages such as Naxi have underlyingly similar conditioning, but differ in detail, in their outcomes and thus in the lexical forms ! 173 David Bradley in each tonal category. For example, the high level tone of Naxi can not be related to any specic tone category in Burmic or Ngwi. There are various phonological criteria for subgrouping within the Burmish subgroup of Burmic, which of course includes Burmese, the earliest literary language within the group, as well as its dialects (Arakanese, Tavoyan, Intha and so on), the closely related language Hpun (Yabu 2003), and the more distantly related North Burmish cluster which includes Tsaiwa/Atsi/Zi, Lawngwaw/Langsu/Maru, Lachik/Lashi, Ngochang/Achang/Maingtha and so on; for details see Bradley (1997). One major change in Burmese and its dialects only is the merger of alveolar and palatal affricates, *ts/*c and so on. A parallel but separate merger is also seen in some Ngwi languages such as Central Ngwi Lahu, Northern Ngwi northern Nisu, Southern Ngwi Mpi and so on, but not in any other Burmish language; this change has also largely diffused into Hpun, but not completely; the last elderly speakers of Hpun still retain a few words with original /ts/ and /tsh/. Burmese has also merged not just prenasalized voiced but also all other originally voiced stops and affricates to voiceless unaspirated. This is one step further than most other Burmish languages, which either retain a voiced series alongside voiceless unaspirated and aspirated series, like Hpun, or have a three-way unaspirated, aspirated and glottalized contrast for stops and affricates. However, there are few phonological innovations which link all of Burmish but are absent in Ngwi. For a thorough phonological reconstruction of Northern Burmish, see Mann (1998). Criteria for subgroups within Ngwi include various characteristic secondary tonal splits, such as in the *L stop-nal category in many Central Ngwi languages such as Lisu, Lahu, and so on. In these languages, *L has a regular low falling glottal stop nal reex, but also a conditioned high reex after certain *s and *! prexed initials, high level in Lisu and high rising in Lahu. Another typical Central Ngwi split is in *Tone 2, with innovative high and low tone contrasts also conditioned by initial consonant environments. 3. Morphology Some of the morphology found in Burmic derives from TB sources (Wolfenden 1929, LaPolla 2003). The main example is the so-called causative prex *s-, a valencyincreasing device as in Lisu /do33/ ‘drink’ from 54 *nda!1, /to44/ ‘give to drink’ from *s-nda!1. This is reected in a relatively small number of verbs in most Burmic languages, in phonological environments where the prex can be reected by distinct initial or tonal outcomes, as in the Burmese h/non-h pairs (Okell 1969:205-208) or in Lahu (Matisoff 1973[1982]:32-34) among many other Burmic languages. This prex is not productive in any Burmic language. 174 ! The Characteristics of the Burmic Family of Tibeto-Burman Other morphology is innovative within Burmic or one of its subgroups. For example, the positive extentive or dimensional verbs show consistent patterns of grammaticalization in Ngwi languages; the verb 10 *Cmya2 ‘many’ also becomes a question word *!"-Cmya3 ‘how many’, a bound nominal form *!Cmya3 ‘quantity’ and an equative or rst syllable of reduplicated adverbial form *Cmya1 ‘equally many/a lot’ with semantic bleaching and prexes leading to different initial and tonal outcomes (Bradley 1995). Parallel processes apply to a small set of verbs including 13 ‘big’ and 14 ‘long’ among others; Ngwi languages differ as to how many positive extentives grammaticalize in this way. This innovation is absent from Burmish. Table 1 shows the positive and negative extentive forms in Southern Lisu (Bradley et al. 2006). The rst nine positive forms grammaticalize; the last two positive forms and the negative forms do not grammaticalize and are included for illustrative purposes. Table 1: Extentive forms in Southern Lisu ‘many’ ‘far’ ‘tall/high’ ‘big’ ‘long (distance)’ ‘long (time)’ ‘thick (inanimate)’ ‘wide’ ‘deep’ ‘heavy’ ‘fat (animate)’ mj"21 $%21 mo44 wu21 '(44 m%44 ('(44) t&u33 h)i44 n*55 li21 ts&%33 ‘few’ ‘near’ ‘low/short’ ‘small’ ‘short (distance)’ ‘short (time)’ ‘thin (inanimate)’ ‘narrow’ ‘shallow’ ‘light’ ‘thin (animate)’ #i55! p&"33 ti55 ø44! zo44! #o55! du55 b"21 (Chinese)! t&*21 lo33 t+ø44 Table 2 shows the corresponding grammaticalized question, adverbial, intensied adverbial, bound nominal, and equative forms in Southern Lisu, which are exclusively based on the positive forms but have bleached meanings which subsume the negative forms as well. The forms are listed in the same order as in Table 1. The question forms mean ‘how many?’, ‘how far?’, and so on. The adverbial forms mean ‘many’, ‘far’, and so on. The intensied adverbial forms mean ‘very many’, ‘very far’, and so on. The bound nominal forms are either bound forms sufxed to a noun stem and meaning ‘quantity of N’, ‘distance of N’, and so on, or abstract nominals plus the sufx /z"44/ and meaning ‘quantity’, ‘distance’, and so on. The equative forms mean ‘exactly as many’, ‘exactly as far’, and so on. Gaps reect the absence of a grammaticalized form; there is no intensied adverbial or equative derived from ‘deep’ nor adverbial forms derived from ‘long (time)’. In ‘long (time)’, the possible extra syllable (‘long (distance)’) in the ! 175 David Bradley question form and the absence of the adverbials may be related to the fact that ‘now’ is /"44m%44/, itself perhaps derived from ‘what time?’. Note that in Southern Lisu, as in most other dialects, other positive verbs of dimensional extent, such as /li21/ ‘heavy’ and /ts&%33/ ‘be fat (animate)’, lack the types of forms seen in Table 2 below. The grammaticalized forms other than /n*55/ ‘deep’ all show consistent and parallel tones within each grammatical category, which (where they differ) reect different proto-tones as discussed above. These are /44/ in the question, adverbial, and bound nominal forms, and /55/ in the rst syllable of the intensied adverbial and in the equative. By contrast, the /55/ tone in ‘deep’ is stable, unlike the forms with /21/, /33/, or /44/ in the verb form. Table 2: Grammaticalized forms of extent in Southern Lisu Question Adverbial Intensied Nominal Equative Adverbial ‘many’ "44mj"44! "44mj"44mj"44! "21mj"55mj"44! +mj"44/mj"44z"44!tø44mj"55"!21! "44$%44$%44! "21$%55$%44! +$%44/$%44z"44! tø44$%55"!21! ‘far’! "44$%44! 44 44 "44mo44mo44! "21mo55mo44! +mo44/mo44z"44! tø44mo55"!21! ‘high’ " mo ! "44wu44wu44! "21wu55wu44! +wu44/wu44z"44! tø44wu55"!21! ‘big’ "44wu44! 44 44 "44'(, 44'(, 44! "21'(55'(44! +'(44/'(, 44z"44! tø44'(55"!21 ‘long(d)’ " '( ! ‘long(t)’ "44m%44(-.44)! ! +m%44/m,%44z"44!tø44m%55"!21! 44 44 44 44 44 21 55 44 " t,&u t,&u ! " t&u t&u ! +t&u44/t&u44z"44! tø44t&u55"!21! ‘thick’ " t&u ! "44h)i44h)i44! "21h)i55h)i44! +h)i44/h)i44z"44! tø44h)i55"!21! ‘wide’ "44h)i44! 44 55 44 55 55 55 " n* n* ! ! +n* /–!! tø44n*55"!21! ‘deep’ " n* ! Southern Lisu speakers differ as to how many of the abstract nominal forms with /z"33/ they use or accept; some instead use /dz"33/ in one or more of these. The following examples illustrate the use of these forms. The form /mj"21/ ‘many’ behaves differently, both in that it can be followed by a classier as in (2) and thus can operate as a numeral, and in that its question form is productively used with any verb including with other positive extent verbs as in (3c) and (4c), instead of their grammaticalized forms, and with other verbs which do not grammaticalize in this way as in (6) and (7) and including the negative extent forms of verbs whose positive does grammaticalize, such as in (5). The pattern of (1c) with cognate question form and verb is not preferred, but occurs for some speakers. (1) a. ! 176 ! "44mj"44! t+&i44! how.many reach d/o33! exist "!21?! t&ø33!! mj"44! d/o33!"33! Q this quantity exist DEC The Characteristics of the Burmic Family of Tibeto-Burman b. "44le44! which ! mj"44! d/o33! quantity exist "!21?! t&ø33!! k&%55! mj"21! "33! Q this extent many DEC mj"21! "!21?! ! c. ? "44mj"44! t+&i44! how.many reach many Q ‘How many are there? There are this many.’ ! (2) a. "44mj"44! w"21! d/o33!"!21?! t&ø33!! mj"44! d/o33! "33! how.many CLF.human exist Q this quantity exist DEC b. t&ø33!! k&%55! mj"21! "33! this extent many DEC There are this many.’ ! ! ! ‘How many people are there? d/o33!"33! s"21!w"21! 3 CLF.human exist DEC ‘There are three people.’ c. ! With the positive forms from ‘far’ to ‘thick’ in Table 1, the possible patterns are as follows, as exemplied by ‘far’; any other verb apart from /m%44'(44/ ‘long (time)’ is parallel. (3) a. ! "44$%44! how.far b. "44le44! which ! c. ! t+&i44! reach d/o33! exist "!21?! t&ø33!! $%44! d/o33! "33! Q this distance exist DEC $%44! distance d/o33! exist "!21?! t&ø33!! k&%55! $%21! "33! Q this extent far DEC $%21! far "!21! Q "44mj"44! t+&i44! how.many reach ‘How far is it? It is this far.’ For the verb ‘long (time)’ the pattern is slightly different, as follows: (4) a. ! ! b. "44le44! m%44'(44! which duration c. ! ! "44m%44('(44)! t+&i44! d/o33! how.long(t) reach exist d/o33! exist "!21?! t&ø33! m%44('(44)! d0o33!"33! Q this duration exist DEC "!21?! t&ø33!k&%55! m%44'(44! "33! Q this extent long(time) DEC "44mj"44! t+&i44! m%44'(44! "!21!! how.many reach long(time) Q ‘How long a time is it? It is this long a time.’ 177 David Bradley The different question and answer structures used productively with other verbs including negative extent verbs and verbs which do not have grammaticalized forms can be illustrated with ‘few’, ‘heavy’, and ‘light’ as follows, and is exactly parallel to the patterns already seen in (3c) and (4c). (5) "44mj"44! t+&i44! #i55! "!21! ! how.many reach few Q ‘How few is it? t&ø33! k&%55! #i55! "33! this extent few DEC It is this few.’ (6) "44mj"44! t+&i44! li21! "!21! ! how.many reach heavy Q ‘How heavy is it? t&ø33! k&%55! li21! "33! this extent heavy DEC It is this heavy.’ (7) "44mj"44! t+&i44! lo33! "!21! how.many reach light Q ‘How light is it? t&ø33! k&%55! lo33! "33! this extent light DEC It is this light.’ The following examples illustrate the use of the bound nominal forms in (8), the equative forms in (9) and the periphrastic alternatives as in (10) and (11), based on the verb /wu21/ and its various grammaticalized forms. wu44! d/o21! (8) h)i33! house size big ‘Is it the size of a house? "!21! h)i33! wu44! m"21! d/o33! Q house size not big It is not the size of a house.’ "!21! h)i33! tø33wu55"!21! m"21! k"33! (9) h)i33! tø33wu55"!21! k"35! house equal.size happen Q house equal.size not happen ‘Is it of equal size to a house? It is not of equal size to a house.’ xø55xø55! wu21! (10) h)i33! house like big ‘Is it as big as a house? "!21! h)i33! xø55xø55! m"21! wu21! Q house like not big It is not as big as a house.’ (11) h)i33! t&i21! le35! wu21! "!21! h)i33! t&i21! le35! m"21! wu21! house one same big Q house one same not big ‘Is it the same size as a house? It is not the same size as a house.’ These grammaticalized extentive forms are found in most Ngwi languages, though the exact inventory and structures differ from language to language and even between dialects of a language, as outlined in Bradley (1995). For example, Nisu grammaticalizes eight of the nine verbs also found in Lisu, but not ‘big’; it also grammaticalizes ‘heavy’; Nosu 178 ! The Characteristics of the Burmic Family of Tibeto-Burman grammaticalizes further, replacing seven negative extent verbs with grammaticalized positives including ‘heavy’, but lacks ‘far’ (having replaced the cognate with ‘road’ + ‘long (distance)’) and has two distinct forms for ‘wide’, in two and in three dimensions. Lahu grammaticalizes eight verbs including two forms for ‘wide’ but not ‘deep’, ‘thick’, or ‘heavy’. Some of these have additional structures not found in Lisu; for example, from the verb /v153/ ‘far’ Lahu has the question form /q&a21f133/ ‘how far?’, the nominal form /221f133/ ‘distance’, and other forms including /t+&i33f133(f133)/ ‘this far’, /t+&i33f133f133*21/ ‘exactly this far’, and diminutive /t+&i33f135*21/ ‘only this near’; in addition there is periphrastic /q&a21ma33v153/ ‘how far?’, with the question form derived from /ma53/ ‘many’ and parallel to (3c) above (Matisoff 1973[1982], Bradley 1995). Another innovation restricted to Ngwi is the development of a set of dyadic kin group classiers for groups of relatives (Bradley 2001). For example, Lisu has /p"""l"21/ for a group including a father and his children, /m"55l"21/ for a group of a mother and her children, and /pi55 li21/ for a group of a grandparent and his or her grandchildren, used as classiers with numerals two and above, as in Lisu /#i21 pi55 li21/ ‘a group of two people, one grandparent and one grandchild’. Such forms are found with different degrees of grammaticalization in one or more languages in most branches of Ngwi, but are absent from some Ngwi languages. Of course some other TB languages have dyadic nouns, but not dyadic forms which are solely classiers. One other vestige of TB morphology is the productive nominal prex *!a which occurs throughout Burmic, either as an abstract verb nominalizer or as a formative prex on a variety of nominal forms: kinship terms, body parts, as a dummy head for bound forms, and so on. The Burmese cognate [!"] has the verb nominalization and the kinship use; Ngwi forms show mainly the nominal uses, with a distinct form for verb nominalization derived from *!a!1 like Lahu /221/, Lisu /ji55/, Phunoi /!a#55/, and so on in some languages. In the Ngwi languages, cognates of the *!a prex are most unusual in that the tone is determined by the tone of the following syllable, in different ways in different languages, one of the very few examples of tone sandhi seen in this group. The cognate of the *!a prex is highly productive in some S Ngwi languages such as Akha and Hani, both as a verb nominalizer and as a prex on bound nominals and kin terms, and is also very frequent with nominal heads in other languages such as Lisu. In other languages, the *!a!1 prex has become the more productive one, as with Lahu /!221/ with both pre-verb and pre-nominal uses (87 pages of entries in Matisoff 1988), while Lahu cognates of the *!a prex occur in a substantially smaller number of pre-nominal forms (41 pages of entries in Matisoff 1988). ! 179 David Bradley 4. Syntax Like the majority of TB subgroups, Burmic is consistently verb-nal, and has nearly all the other characteristics usual in a verb-nal language. However, unlike many other TB subgroups (van Driem 1993, LaPolla 2003, DeLancey 2010), there is also no reconstructible nominal agreement morphology of any kind on verbs. In Burmese, an innovative apparent number marking is possible but not obligatory using postverbal /t+a55/, written kra3, which is actually a marker of severality, mutuality, or collectivity of action (Okell & Allott 2001:16-17), not just a plural. This follows all grammaticalized serial verbs such as modals, but precedes the politeness marker, aspect and realis/irrealis markers. Burmic as a whole also has no reconstructible nominal case or other postpositions, though all Burmic languages now have such postpositions. One example of a retained TB syntactic characteristic not reecting verb-nal structure is that the TB negative *ma and negative imperative *ta are consistently preverbal throughout Burmic, with forms reecting the etyma *ma2 and *ta2 respectively. However, the negative imperative cognate has been lost in Burmese itself, another instance where this language with the longest written history within Burmic is nevertheless the most innovative. All Burmic languages have numeral classier systems; where a numeral is present, it must normally be followed by a classier; they follow the head noun. However, the forms of the classiers are often not solid cognate etyma and differ extensively between languages, even within Ngwi and more so between Burmish and Ngwi. Given these lexical differences and the syntactic differences in their behavior when combined with deictics discussed below, it appears that the development of classier systems is a relatively recent areal phenomenon in Burmish, possibly diffused early from Chinese when the order there was N + Numeral + Classier, as is also seen in Bai which has been heavily inuenced by Chinese for two millennia. As the data in the extremely valuable comments on this paper by Dai Qingxia indicate, it is possible to nd some similarities of classier forms between classier systems among Burmic languages, but all of these are cognate nouns or other forms which have only relatively recently become classiers. For example, the usual human classier reects cognates for an etymon ‘man’ *yok which also becomes the usual human classier in many Burmish languages including Burmese, as well as the Lisu form /0o44/ found in some dialects; others have /w"21/. However, the human classier in Lahu, which is closely related to Lisu within Central Ngwi, is /$a53/, directly relatable to the Akha (Southern Ngwi) form /$a21/, and both derived from the noun ‘strength’, reconstructible as *#a2. In Nosu the usual human classier differs according to the number and by dialect. Similarly, the classier for ‘egg’ is an autoclasssier in some 180 ! The Characteristics of the Burmic Family of Tibeto-Burman Burmic languages; however, in Burmese and a few other Burmic languages, it derives from a noun *lum2 meaning ‘round thing’, also used to classify other round objects such as fruit in those languages. Conversely, the classier for round things including fruit is /s%21/ in Lisu, /+i11/ in Lahu and /'i21/ in Akha, all from the noun ‘fruit’, Burmic *si2 which of course has cognates in Burmese and elsewhere in Burmish but is not a classier there. This is also used to classify eggs in Lahu and Akha. Such similarities of recently-developed grammatical forms suggest recent contact between Lahu and Akha, which must postdate the migration of the Akha westward into the Lahu area no more than 500 years ago. The classier for eggs in most Burmish languages is /t'&am/. The general default classier also differs between languages, both in form and in the range of nouns with which it is used. As a general default classier, Burmese uses /khu3/, derived from the noun meaning ‘piece’. Many Ngwi languages now use a cognate of the widespread Tibeto-Burman nominalizer sufx *ma1 as a general classier; this includes Lisu and Nosu but not Akha which uses /hm21/. Note, however, that the use of the general classier is much more widespread in Lisu than in Lahu, Nosu, or Akha; on the other hand, Nosu uses the general classier to classify humans with the numerals one (and two in some dialects). Of other Southern Ngwi languages, Mpi uses /to33/ as its general classier, while Bisoid languages such as Bisu, Phunoi, and Laomian use /ma55/ or /ma33/; thus even within Southern Ngwi the classiers do not correspond, suggesting that the addition of classiers to the syntactic inventory of the Burmic languages is quite recent and substantially postdates the splitup of Burmic and even the splits within Burmish and Ngwi. The behavior of deictics differs between Burmic languages, both in order (before or after the head noun), in constituency (whether some other nominalizing element may or must follow the deictic, as in Lisu or Lahu, or whether a following classier is obligatory, as in Akha, and so on), and in how they combine with a numeral and a classier if both a deictic and a numeral are present in the same NP. The Burmese deictic precedes the noun and does not then require nominalization; the numeral plus classier follows the noun, as in (12). Burmese is diglossic; it is often the case that there are distinct literary and spoken forms for grammatical function words, with many literary forms also found in early inscriptions and thus presumably the earlier forms. (12) t&o22/ho22! lu22! that (lit/sp) person ‘those two people’ n3"! two jau!! CLF.human In Akha, the deictic follows the noun and must then be followed by the appropriate classier for the noun. In Lisu it follows the noun and must be followed by a nominalizer; ! 181 David Bradley if a numeral and classier are also present, the numeral follows the deictic and the classier precedes the nominalizer, as in (13); a nominal case and/or topic marker may then follow the nominalizer. m"44! (13) l"21ts&o33! go33! s"21! w"21! person that 3 CLF.human NOM ‘those three people (on the same level)’ In Black Lahu, the proximal deictic /t+&i33/ precedes or follows the head and may be alone or followed by the nominalizer/relative/genitive /ve33/, with differences in meaning (Matisoff 1973[1982]:52-53, 110-116), but the other deictics precede the head and must be followed by /ve33/. In Yellow Lahu, the proximal deictic /t+&i33/ behaves like other deictics: it precedes the head and is followed by relative/genitive /ve33/ (Bradley 1979b). In all Ngwi languages, but not in Burmese, a deictic with its nominalizer or classier may occur alone without a head noun; in spoken Burmese, a deictic without an overt head must be followed by the dummy noun /ha22/ as in (14a); this is very often fused into one syllable as shown in (14b). (14) a. di22! ha22 this NOM ‘this’ b. da22! this.NOM ‘this’ For more discussion of the forms and syntax of deictics in Burmic, see Bradley (2003). It is shown there that various deictics which distinguish relative height and distance can be reconstructed for Ngwi, while Burmish lacks these. Some of these Ngwi deictics have extra-Burmic TB cognates, implying that the situation in Burmese is the result of loss of earlier deictic contrasts. In Burmic languages, nominal case marking is optional and fairly infrequent. Its presence is determined pragmatically and may be used for disambiguation. Most of the nominal case markers, which follow the entire noun phrase and in some languages such as Lisu may then be followed by a topic marker, are not cognate across Burmic, and have transparent recent origins within subgroups of Burmic. For example, the literary Burmese object marker /!a42/, normatively dative but also used for accusative, is derived from the homophonous noun meaning ‘strength’, while the spoken Burmese (and literary Burmese accusative) object marker /ko22/ is derived from the homophonous noun ‘body’, though now spelled differently. The Central Ngwi object marker, Lisu /t*55/ and Lahu /t&a!21/, is derived from a relative location noun meaning ‘(on) top’. 182 ! The Characteristics of the Burmic Family of Tibeto-Burman The nominal topic markers show some limited cross-language cognates within Ngwi, not always with the same function in every language, such as *!na2 with cognates including the Lisu secondary topic marker /n"21/, the Lahu embedded question marker /na11/, and the Nisu topic marker /no55/. It is often the case that a nominal topic marker is also used as a nonnal clause marker of condition or cause, as in Lisu and Nisu, or for some other kind of embedding, as in Lahu. Nominalizers and relativizers follow the verb in the embedded clause; where the head noun of the relative clause is present, it usually follows the embedded clause, though in Lisu a head noun may precede a light relative (which contains few constituents). There are some nominalizers/relativizers which have TB cognates outside Burmic, but which have quite different functions within Burmic or parts of it. For example, the widespread Ngwi locative nominalizer *gu1 ‘place to Verb’ exemplied by Lisu /gu33/ and Lahu /k121/ has extra-Burmic cognates which are nominal ergative markers (Newar, rGyalrong, etc.). There is no cognate of this form in Burmish; in Burmese, forms derived from *ra1 like written Burmese ra, spoken Burmese /ja22/ are used as locative nominalizers; this form also has a variety of other nominalizing functions in Burmese. Throughout Burmic, an agentive nominalizer *su1, an instrumental nominalizer *du1 and a general nominalizer *ma3 are found, with widespread cognates elsewhere in TB. Unlike in some other parts of TB, cognates of the ‘male’ sufx *!pa2 and the ‘female’ sufx *ma3 (homophonous with ‘mother’ and with the general nominalizer) are not productively used as clausal nominalizers, though they can be used as verbal nominalizers (‘man/woman who Verbs’) in some Ngwi languages. Cognates of *su1 are also in widespread use in various Burmic languages as remote third person pronouns; indeed, in Burmese the cognate has become the normal third person animate pronoun. Some languages have lost or restricted one or other of these; for example, Lahu has cognates for *ma3 only as a bound nominal sufx, and for *su1 only as a remote third person pronoun, having generalized the form /ve33/ from genitive into a relative in all dialects and into a nominalizer in Black Lahu (Bradley 1979b). Conversely, Lisu has extended *ma3 into its general classier, with a change in tone. In Burmese, the nominalizer *su1 also occurs rst in many compounds with a verb, e.g. written su khui2, spoken /4"#&$42/ su + ‘steal’ > ‘thief’; note that most of these have the rst syllable reduced. The clause-nal markers of some Ngwi languages also show interesting cognate patterns, some of which suggest contact with Dai languages and borrowing across genetic linguistic boundaries. For example, there is a nonnal clause (or nominal) simultaneous temporal marker *ta2, with cognates such as Lisu /th*21/, Lahu /tha53/, Hani /ta21/ and Nisu /tho21/. A similar form /thaa51/ is also found in Southwestern Dai languages like Thai, which has conditional as well as temporal meaning. In Ngwi ! 183 David Bradley languages, this marker is frequently seen grammaticalized into the question word ‘when?’, as in Lisu /!"55!t&*21/, Lahu /q&a21t&a!53/, Hani /xa55ta21/, Nosu /k&%21t&%33/, and so on. 5. Lexicon One of the main characteristics distinguishing Burmic from the rest of TB, as in most instances of subgrouping, is innovative lexical material. Another is irregular phonological innovations in cognate TB material, such as *-k in etyma like 23 ‘tree’ *sik, 52 ‘heart’ *snik, and parallel examples which reect *-! in most other TB subgroups. This was already noted by Matisoff in a footnote in Benedict (1972:79). A third is consistent presence of certain two-syllable compounds, such as ‘ashamed’ *srakL da!1, as in Lisu /-"55 do33/, Lahu /ja!21 t233/, Akha /'a21do55/, Nosu /-o55to33/, and so on. The examples in the Swadesh list below are 23, 37, 39 and 83. It hardly needs to be stated that some of the Swadesh list items are less solid cognates; for example, 78 ‘sand’ is probably a Dai loan, 47 ‘knee’ in most Ngwi languages is just a compound of 46 ‘foot/leg’ and *!tsikL ‘joint’ and also means ‘ankle’ in some languages, and 63 ‘swim’ is unlikely to be cognate among groups in whose original homeland rivers were far away down at the bottom of deep valleys and dangerous to swim in; thus a variety of innovative forms develop, such as Burmese, where ‘swim’ literally means ‘cross water’. Words such as 88 ‘green’ show innovative forms and semantic shifts: the Burmese cognate means ‘brown’, the Burmese form /s$%42/ written cim2 meaning ‘green’ is derived from ‘fresh’, and the Lisu form for ‘green’ is a compound meaning ‘heart fresh’. Other words like 71 ‘say’ have a variety of forms, so taking the most frequent form rather than a possible cognate tends to understate the degree of lexical similarity. I have added a cognate for ‘(rocky) cliff’ because there has been some confusion as to which Burmic form should be compared to TB forms meaning 77 ‘stone’. Of course there are also numerous lexical innovations within the components of Burmic, notably Ngwi and subgroups of Ngwi. For example, most Ngwi languages which have birth-order systems (where each person has a specic name based on their order of birth) have a cognate for *s-likL ‘second-born child’, as in Lisu /a21 le35/ (Bradley 2008b); a possible Burmese cognate is seen in the sufx for mother’s younger brother and sister in Burmese, /le2/, though the rhyme correspondence is irregular. There are also innovative two-syllable cognates such as 35, 38, and 46 within all or part of Ngwi but not Burmish. There are also some possible lexical innovations linking Central Ngwi and Northern Ngwi, as seen in 9, 41, 65, 97, and 99. 184 ! The Characteristics of the Burmic Family of Tibeto-Burman Semantic shifts and generalizations in cognate etyma are also found. For example, Burmish retains the TB etymon *d!ul for ‘silver’ as *C!we1, but Ngwi appears to have shifted the meaning of this etymon into its group autonym, and instead generalized the etymon 90 ‘white’ *plu1 to mean ‘silver’ as well. We have just noted the semantic shift from ‘green’ to ‘brown’ of the Burmese cognate of *!#o1 and its replacement by ‘fresh’. There are also consistent semantic shifts in certain lexical items, replacing TB cognate material found elsewhere in Burmic, in some subgroups. For example, 21 ‘dog’ usually shows a cognate for *kwe2, but in Central Ngwi there is also a form *na2 derived from the verb ‘snatch’ as in Lisu /a55 na21/, alongside the cognate. For ‘re’, the usual cognate *C-mi2 (found in all but the Sal subgroup of TB) is widespread, but some Central Ngwi languages have a replacement form with the *!a prex and the causative form of the verb 84 ‘burn’ *s-dukL, for example Lisu /"55 to55/. Similarly, Burmese has replaced 89 ‘yellow’ but kept the cognate in the meaning ‘gold’. The presence of most of these lexical innovations demonstrates membership of Burmic or the relevant part thereof; their absence is not absolutely criterial, as subsequent lexical innovations or losses may eliminate them. For example, Lahu completely lacks the birth order system so widespread in other Ngwi languages. It is also striking that there is no obvious etymon for 86 ‘mountain’ among the Burmic groups, despite the fact that they traditionally lived in or came from mountainous areas. 6. Conclusion All Burmic languages share a number of characteristic innovations at all levels of linguistic structure, supporting their status as a separate major component of TB; there are also internal innovations within parts of Burmic. On the other hand, the Burmic languages also share a large proportion of cognate structures and forms with other major components of TB as well. There is some lexical evidence of a link with Qiangic, particularly with the Na languages, suggesting an Eastern TB subgroup. This also makes sense in the context of traditional history among the Ngwi groups, who all report that their ancestors came from the north. It remains to be determined whether Bai and Tujia also belong within Eastern TB. The development of classiers within various Burmic languages can be approximately dated. It probably started shortly before the advent of the Burmans in the plains of Burma in about 860 AD as part of the Nanzhao army conquering the Pyu, but after the separation within the Central Ngwi languages, and also after the separation within the Southern Ngwi languages. It seems to have involved various independent parallel developments, mainly from nouns into classiers; the exact outcomes differ greatly even within closely-related subgroups. ! 185 David Bradley Though Burmese inscriptions are the earliest securely-attested Burmic data, dating back to 1112 AD, there are also manuscript materials in various mostly Northern Ngwi languages (Nosu, Nasu, Nisu, Sani) reecting a writing tradition which may possibly antedate this somewhat; for some details see Bradley (2009). However, the phonetic interpretation of these materials is somewhat opaque, as the scripts were originally logographic and are read with modern pronunciation. If the Bailang songs embedded in Han Dynasty materials reect a Burmic language, that would be the earliest available record; otherwise, that honor belongs to the Man Shu, c.870 AD, which has two short Cuan wordlists (Luce 1961), one of which is of a Ngwi language.! 186 ! The Characteristics of the Burmic Family of Tibeto-Burman Appendix: Swadesh List etyma and forms in Burmic languages Underlined forms – noncognate (some showing cognacy within all or part of Ngwi) Burmese forms given in transliteration, even tone (*Tone 1, 22) unmarked Burmic (Bradley 1979a) I *!a1! you *na!1! we (incl) *ni1/do3 this *i1/ti1 that *to1/go1 who? *!"su1! what? *!"je2/'i2 not *ma2! all (various) many *Cmya2! one *ti2/tik two *#i2/#ik big *k-ri2! long *s-ri!1! small *n-yay1! woman *mi2! man *g-yok person *tsa!1! sh *!a2! bird *s-!yak! dog *kwe2! louse *xan2! tree *sik ba!1 seed *yo2/je3! leaf *Cpak root *mje1! (tree) bark *Cguk skin *re1/#d/i1! esh *'a2 blood *swe2! bone *s-ro2! grease/fat *tsi1! egg *u3! horn *kro1! tail *da!1 !mri2! feather *!mwe3/2! 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32. 33. 34. 35. 36. ! Burmese literary/spoken #a! na#! -dui3 i=di! t&o=ho! %ay su! %a! ma3 [m"]! a2 lum2! mya2! tac [t"]! hnac [n3"]! kri2! hra#! #ay! min2 ma3! yok kya2! lu! #a2! h#ak! k&we2! san2! sac pa5! ce3! rwak! mrac! khok! re! sa2! swe2! rui2! c&i! !u3! k&rui! mri2! mwe2! C Ngwi Lisu (S) #w"33! nu33! 5w"33nu21! t&ø33! go33! !"21m"33! !"55'(21! m"21! !"21d/(21! mj"21! t&i21! #i21! wu21! '(44! /o33! z"21m%21z"21! z"21gu21! ts&o33! #w"55! #"35! k&%21/!"55n"21! x%44! s%35-! '(55! pj"21! t+e44! ko!21d/(33! d/(33! xw"21! '(21! wo21to44! ts&%33! fu44! t'&(44! m%55! my44! S Ngwi Akha #a55! n255! a21d%33! h633! t633! a55'u21! a21d/e21! ma21! d221t2!33! mja21! t&i21! #i21! h%21! m2)55! #i55! za21mi21za21! xa21dze55za21! ts255! #a21! a55d/i55! k%21! '*55! s2)21-/a55%255! jø21! pa!21! du21t'i55! xo!33! ba!33xo!33! 'a21d/i55! 'i21! 'a21jø21! tsu55! u!33! t'ø55! d221mi21! t'a!21hm33! N Ngwi Nosu #a33! n%33! ni555o21 ts&(44! a33dz(44! k&a44di33! +i44 !a21! d7(33k%44! #i33! ts&(21! #i21! 7(33! -o33! *55ts(!33 a21m(33! za21bu33z%33! ts&o33! h)%335go21! h)*33ts(!33! 7(33ga33 -%33 s(!33%o33 j*55 t+&i33! #d7i21 #d7(33! #d7(44-%33! -%33 s(33! $o33 ts&(33! t+&i21 d0.33! m3(21! #*33! 187 David Bradley Burmic (Bradley 1979a) (head)hair *!tsam1 kri!1! head *u2 !du2 ear *!na2 !ba!1! eye *Cmyak nose *sna1!! mouth *(C)me2/hnutL tooth *swa2/!cway1 tongue *!lya1! ngernail *si!2! foot *kre1 spak! knee *du2 hand *lak belly *wam2! neck *li!1! breasts *no3 heart *s-ni3/k liver *sin2 drink *nda!1! eat *dza2! bite *Ctsat/Ckuk see *!mra!1! hear *gra2! know *si2/3 sleep *yip die *'e1! kill *Csat swim (various)! y *b-yam1! walk (various) come *la1! lie (down) *ndo2 sit *(C)ni1! stand *!ryap! give *be2 say (various) sun *!ne1! moon *b"la3! star *Cgray1! water *re1! rain *rywa1 stone *k-lok cliff! *Crak! 37. 38. 39. 40. 41. 42. 43. 44. 45. 46. 47. 48. 49. 50. 51. 52. 53. 54. 55. 56. 57. 58. 59. 60. 61. 62. 63. 64. 65. 66. 67. 68. 69. 70. 71. 72. 73. 74. 75. 76. 77. ! 188 ! Burmese literary/spoken c&am pa5! !u2k&o#2! na2! myak! hna! hnut/pv2cvp! cway/swa2! hlya! sa#2! k&re! du&! lak! wam2 la#! nui'! hnac lum2! asa#! sok! sa2! kuik! mra#! kra2! si3! ip se! sat! re ku&! pyam! hlyok la! lay& ne/t&ui5! rap! pe2! pyaw2! ne! la3! kray! re! rwa! kyok kyok t&v3 ram! C Ngwi Lisu (S) wu55ts&ø44! wu55dy33! n"55po33! mj"44! n"44bø44! m621n635! s(21t+&i44! l"44t+&ø44! sø21! t+&i44p&*35! t+&i44ts(55! l*!21p&*35! h)e!21m"44 li44be33! "55t'&(35! #i33m"33! '(21! do33! dz"21! k&o!21! mo33! p"33d/"21! s655! ji!21! '(33! se!21! (s"55/wo44)! %jø33! +ø21! l"33! do21! #i44! h)e!21! g%21! %*44! mi21mi33/b%33! h)"33%"33! ku44z"21! ji33! h)"33! lo44di44! $"55! S Ngwi N Ngwi Akha Nosu ts*55k2)55! 221#*33! u21du21! i33t+i33 21 55 na %2 ! n3a21po33! 33 #233dz(21! mja! n%!33! 55 na m*55! n3a21%i55 33 xa21m* ! k&a21p&i55! s621/d/i55! d0.33 m*21la55! h)a33n*33! 21 s2) ! s(33! 55 +(33p&o55! a21k% ! +(33ts(55/ba21ts(!33! p221ts%!21 a21la!21! lo55 21 u ma!33! h)i55! 55 k221l2) ! ku21li33 a21t'ø55! a44n*33 33 n% ma33! h)*33ma55 21 s(21 'a ts2)21! 55 d2 ! ndo33 21 dza ! dz%33! 21 21 k2! /ts*! 8d0.55! 55 m2 ! $%21mo33 21 ga ! n3a33 si21! d%33d7i33! ju!21! i55! 55 'i ! s(33! 21 s*! ! si55! 55 21 21 i di di ! t+(33! d7i33! z233! 33 t+2) ! -u33! 55 la ! la33 21 33 x2 da ! d7i33 ndo33! 55 n% ! #i33! 21 j2! ! h)i55! 21 %i! ! %(21! 55 * ! h)i21 3 n2) ma33! h)o33bu33 33 33 %a la ! l3o21%o21! 55 a21g% ! t+(33! 55 ji33! i t'u!21! 55 j* ! h)a33! 21 33 xa lo ! lu!33! lo33gu55! va55! The Characteristics of the Burmic Family of Tibeto-Burman Burmic (Bradley 1979a) 78. sand *say2 79. earth *!mre1! 80. cloud *Cdim1! 81. smoke *ko2! 82. re *Cmi2! 83. ash *Ckap Cla1! 84. burn *duk! 85. path *lam2/!ga1! 86. mountain *ka!1! 87. red *!ni1! 88. green *!#o1! 89. yellow *s-rwe1 gold 90. white *plu1 91. black *Cnak 92. night *!rak! 93. hot *tsa1! 94. cold *C/!grak 95. full *mbli!3! 96. new *C'ik! 97. good *!min2 etc.! 98. round *wan2/lo2! 99. dry *swe2/gwe3! 100. name *!myi!1! Burmese literary/spoken say2! mre! tim! mi2k&ui2! mi2! pra! tok! lam2! to5! ni! #ui/cim&! wa! hrwe p&ru! nak/ma#3 rak/#a3! pu! e2 pra#3! sac! ko52! wan2! swe3! ma#! C Ngwi S Ngwi Lisu (S) Akha x%21t+i44! xa55'%!21! 44 mi ! mi55! 44 m21dm55! my21ti ! 21 mu21k&u ! m21xø21! mi21dza33! "55to55! 21 33 k&u h)" ! xa21l*55! 21 do! ! do!21! 44 44 d/" gu ! ga55ma33! 44 21 w"21/t+e /d7i !g221d/221! #i44! ne55! #ø55! ni35t'&(!21! 44 '( ! '%55! 44 '( ! '%55! 33 p&u ! pju55! 44 n* ! na!33! 21 h)y"! ! u21t'i!21! ts&a33! tsa55! d7"44! ga!33! 44 %i ! %j2)33! 21 -.! ! '%!21! x"44/d7i33! m%21! 55 lu ! z2)33! g%33! fu44! 44 mjø ! mj2)55! N Ngwi Nosu m3 u44-.33/-a33! mu44! ti33! mu33ku33! mu55! k&u21l3a33 du55! ga21mo21 bo33 n3i33 l 3o21 -.33! -.33! t+&u33! n233! s(33! ts&a33 #go33! d7i21 -.55! x%33! vo33l%33! fu33 m3i33! ! In cases where two Burmese forms are given above, the one before the slash is the cognate but now has a restricted or different meaning, while the one after the slash is the contemporary word with the exact current meaning. Literary versus spoken Burmese forms are separated by an = sign. In many cases, the additional underlined syllables in polysyllabic forms are clearly analyzable and often have cognates elsewhere; for example, the rst syllable in the words for 81 ‘smoke’ is of course 82 ‘re’, the second Burmese form for 42 ‘mouth’ is literally ‘cheek-edge’, in 52 ‘heart’ the second syllable in Burmese means ‘round thing’, and so on. Where relevant, the modern Burmese pronunciation (as opposed to a transliteration of the spelling) is given in phonetic brackets. ! 189 David Bradley References Benedict, Paul K. 1972. Sino-Tibetan: A Conspectus. Cambridge: Cambridge University Press. Bradley, David. 1975. Nahsi and Proto-Burmese-Lolo. Linguistics of the Tibeto-Burman Area 2.1:93-150. Bradley, David. 1979a. Proto-Loloish. London & Malmö: Curzon Press. Bradley, David. 1979b. Lahu Dialects. Canberra: Faculty of Asian Studies, The Australian National University. Bradley, David. 1995. Grammaticalisation of extent in Mran-Ni. Linguistics of the Tibeto-Burman Area 18.1:1-28. Bradley, David. 1997. Tibeto-Burman languages and classication. Tibeto-Burman Languages of the Himalayas, ed. by David Bradley, 1-71. Pacic Linguistics A-86. Canberra: The Australian National University. Bradley, David. 2001. Counting the family: family group classiers in Yi (TibetoBurman) languages. Anthropological Linguistics 43.1:1-17. Bradley, David. 2003. Deictic patterns in Lisu and Southeastern Tibeto-Burman. Language Variation: Papers on Variation and Change in the Sinosphere and in the Indosphere in Honour of James A. Matisoff, ed. by David Bradley, Randy J. LaPolla, Boyd Michailovsky & Graham Thurgood, 219-236. Pacic Linguistics 555. Canberra: The Australian National University. Bradley, David. 2005. Sanie and language loss in China. International Journal of the Sociology of Language 173:161-178. Bradley, David. 2008a. The position of Namuyi in Tibeto-Burman. Paper read at the Workshop on Namuyi, November 23, 2008. Taipei: Academia Sinica. Bradley, David. 2008b[2007]. Birth-order terms in Lisu: inheritance and contact. Anthropological Linguistics 49.1:54-69. Bradley, David. 2009. Language policy for China’s minorities: orthography development for the Yi. Written Languages and Literacy 12.2:170-187. Bradley, David, with Edward R. Hope, James Fish, and Maya Bradley. 2006. Southern Lisu Dictionary. Berkeley: Sino-Tibetan Etymological Dictionary and Thesaurus Project (STEDT), Center for Southeast Asia Studies, University of California. Burling, Robbins. 1967. Proto Lolo-Burmese. Bloomington: Indiana University. DeLancey, Scott. 2010. Towards a history of verb agreement in Tibeto-Burman. Himalayan Linguistics 9.1:1-39. van Driem, George. 1993. The Proto-Tibeto-Burman verbal agreement system. Bulletin of the School of Oriental and African Studies 56.2:292-334. van Driem, George. 1997. Sino-Bodic. Bulletin of the School of Oriental and African 190 ! The Characteristics of the Burmic Family of Tibeto-Burman Studies 60.3:455-488. Dryer, Matthew S. 2003. Word order in Sino-Tibetan languages from a typological and geographical perspective. The Sino-Tibetan Languages, ed. by Graham Thurgood & Randy J. LaPolla, 43-55. London: Routledge. LaPolla, Randy J. 2003. Overview of Sino-Tibetan morphosyntax. The Sino-Tibetan Languages, ed. by Graham Thurgood & Randy J. LaPolla, 22-42. London: Routledge. Luce, Gordon H. 1961. The Man Shu: Book of the Southern Barbarians. Ithaca: Southeast Asia Program, Department of Far Eastern Studies, Cornell University. Mann, Noel W. 1998. A Phonological Reconstruction of Proto Northern Burmic. Arlington: University of Texas MA thesis. Matisoff, James A. 1972. The Loloish Tonal Split Revisited. Berkeley: Center for South and Southeast Asia Studies, University of California. Matisoff, James A. 1973[1982]. The Grammar of Lahu. Berkeley: University of California Press. Matisoff, James A. 1988. The Dictionary of Lahu. Berkeley: University of California Press. Matisoff, James A. 2003. Handbook of Proto-Tibeto-Burman: System and Philosophy of Sino-Tibetan Reconstruction. Berkeley: University of California Press. Okell, John. 1969. A Reference Grammar of Colloquial Burmese. London: Oxford University Press. Okell, John, and Anna Allott. 2001. Burmese/Myanmar Dictionary of Grammatical Forms. Richmond: Curzon. Shafer, Robert. 1966-1974. Introduction to Sino-Tibetan. Wiesbaden: Otto Harrassowitz. Wolfenden, Stuart N. 1929. Outlines of Tibeto-Burman Linguistic Morphology. London: Royal Asiatic Society. Yabu, Shiro. (ed.) 2003. The Hpun Language Endangered in Myanmar. Osaka: Osaka University of Foreign Studies. [Received 26 August 2010; revised 14 April 2011; accepted 23 June 2011] Faculty of Humanities and Social Sciences School of Communication, Arts and Critical Enquiry La Trobe University (Melbourne, Bundoora) Melbourne Victoria 3086 Australia d.bradley@latrobe.edu.au ! 191 David Bradley !"#$%&'()* +,-./01-.23'45.6* David Bradley La Trobe University ! ! "#$%&'()*+),*-./0123456789:;<= >?@ABCDEFG),*H()*+6I*JKLM-*NOP56Q R-;DSTU?"#VWX),*YZ-[7\]*J-YZ[7K%( )*+^R_6I*J`a\bcAdef?gh\ijk),*-lAm Zno\pqr-sYtu\v5./*Zwx-[7yz?"#{j|X DEY}~•-€•\v‚C),*ƒ()*+-D„…†*J\v5), *Q‡ˆ_‰Š*J?"#2‹Œ•ŒXŽ••‘’Y“-),*Y}\i ”k[7-•–Yv5–')—*˜–',*J*-™„lAŠ*Jš›œ •“ž-*a-Y}?"#j52•Ÿ ¡¢£¤‹Œš-Y} ¡? ¥¦Y§()*\)*J\,*J\*N\7Y\mZ\¨"Y} 192 !