Streams Converging Into an Ocean, 565-583 2006-8-005-022-000014-1

Rethinking the Tsouic Subgroup Hypothesis: A Morphosyntactic Perspective*

Henry Y. Chang

It is generally held that Kanakanavu and Saaroa should be lumped together with Tsou as a Tsouic subgroup (The Tsouic Subgroup Hypothesis: Tsuchida 1976, Blust 1977, 1999). The paper reassesses this well-accepted assumption in the light of six important morphosyntactic innovations observed in Tsou, viz. the Innovative Focus Morphology, the Complete Loss of the PAn Perfective Marker, the Focus Harmony Restriction on SVCs, the NAF-only Causatives, the Obligatory Auxiliary Constraint, and the Emergence of the 3rd Singular Nominative Bound Pronoun. It is found that none of these Tsou innovations are attested in the so-called southern , namely, Kanakanavu and Saaroa, or elsewhere in the Austro- nesian family. This casts doubts on the TSH and raises the question of whether Tsou constitutes a subgroup with Kanakanavu and Saaroa as well as the question of whether Tsou is one of the primary branches of the PAn (cf. Starosta 1985). Both questions will be addressed in the paper.

Key words: Tsouic, subgroup, morphosyntactic innovations, PAn

1. Introduction

Since Tsuchida (1976), it has been generally assumed that Tsou should be grouped

* This paper grew out of an on-going research project entitled Classification and Dispersal of the Austronesians: Anthropological, Genetic, and Linguistic Studies Relating to (NSC 952627-H-001). I am grateful to the project director Paul J.-K Li for inviting me to join the project and providing financial support. This paper is dedicated to him on his seventieth birthday. I am thankful to the following people for reading an earlier version of this paper and giving useful comments: Blust, Jonathan Evans, Paul J.-K Li, Tsai-hsiou Liu, Malcolm Ross, John Wolff, and two anonymous reviewers. A draft of this paper was presented at the Workshop on Internal and External Relationships of , August 31, 2006. I would like to express my gratitude to the audiences there, in particular, Lillian Huang, Li-may Sung, and Stacy Teng for their helpful suggestions. I am also indebted to my former research assistant Julia Y. Su for her help in editing several tables presented in this paper. Needless to say, I am solely responsible for the remaining errors if there are any.

Henry Y. Chang with Kanakanavu and Saaroa as a Tsouic subgroup. Despite some disputes,1 the Tsouic Subgroup Hypothesis (TSH) has been taken as part of the “standard theory” of the Austroneisan subgroupings. This view is reflected in Blust’s (1999:52) remarks:

The fundamental evidence for a Tsouic subgroup has been presented by Tsuchida (1976). Although some writers have questioned the unity of Tsouic (Harvey 1982:90), I accept it as established on the basis of Tsuchida’s extensive documentation.

Nonetheless, Tsuchida (1976) did not explicitly address the TSH issue, neither did he provide firm evidence for it. It seems that he inherited his belief in TSH from Ferrell (1969) and Dyen’s (1963, 1965) lexicostatistical classification and took it for granted. A reassessment of the TSH is in order. Drawing on the reconstructions put forward by several distinguished Austronesianists (Wolff 1973, Ross 1995, 2002, Blust 2003, Li 2006), this paper enumerates six important morphosyntactic innovations in Tsou and indicates that none of these Tsou innovations are attested in Kanakanavu or Saaroa, or elsewhere in the Austronesian family. In light of the findings, the paper will discuss whether the TSH holds true. First things first, let us consider the Tsou innovations one by one.

2. The innovative focus morphology

It has long been observed since Starosta (1985) that main verbs in Tsou are marked with the focus affixes which are used to mark verbs in dependent clauses in other western , as illustrated in (1):

(1) a. mo mo-si ta pangka to emi ’o amo AF2 AF-put OBL table OBL wine NOM father ‘Father puts wine on a table.’

1 Harvey (1982:90) and Starosta (1985, 1995) have questioned the unity of the Tsouic subgroup, but their efforts have not received much attention. I will return to this issue in section 8. 2 Following the Leipzig Glossing Rules, this paper adopts the following abbreviations and conventions: AF: Actor focus ACC: accusative CAUS: causative EXP: experiential FUT: future tense GEN: genitive IF: instrumental focus IPFV: imperfective IND: Indicative LF: locative focus LOC: locative LNK: linker NAF: Non-Actor focus NEG: negation NEUT: neutral NMLZ: nominalization NOM: nominative OBL: oblique OF: object focus PAST: past tense PFV: perfective PF: Patient focus POSS: possessive RED: reduplication REL: relativizer S: singular TNS: tense TOP: topic 1: first person pronoun 2: second person 3: third person pronoun -: affixes =: clitics

566

Rethinking the Tsouic Subgroup Hypothesis

b. i-si si-a ta pangka to amo ’o emi NAF-3S put-PF OBL table OBL father NOM wine ‘Father put the wine on a table.’ c. i-si si-i ta amo ta emi ’o pangka NAF-3S put-LF OBL father OBL wine NOM table ‘Father put wine on the table.’ d. i-si si-eni ta emi ta amo NAF-3S put-I/BF OBL wine OBL father ‘Father kept wine (for him).’ (Zeitoun 2000a:95)

As in (1), main verbs are inflected for the focus affixes m-, -a, -i, and -eni in Tsou. Apart from the Actor focus (AF) affix m-, the other three are normally attached to verbs in atemporal clauses such as negatives and imperatives in other western Austronesian languages. Compare:

Table 1: Focus markers: Indicative vs. Atemporal (Based on Ross 1995:739)

AF PF LF B/IF Indicative -en -an si- Atemporal (unmarked) -a -i -eni

Take Seediq for example. In declarative sentences, verbs are inflected for the usual focus affixes m-, -un, -an, and si-, as illustrated below:

(2) a. m-imah qsiya (ka) laqi. AF-drink water NOM child ‘The child drinks water.’ b. mah-un na laqi (ka) qsiya. drink-PF GEN child NOM water ‘The child will drink the water.’ c. mah-an na laqi qsiya (ka) sapah purayan. drink-LF GEN child water NOM house kitchen ‘The child drank water in the kitchen.’ d. si-imah na laqi qsiya (ka) kopu. IF-drink GEN child water NOM cup ‘The child used the cup to drink water.’ (Chang 2000:83-84)

By contrast, verbs bear a distinct set of focus affixes in negatives and imperatives, as indicated in (3).

567

Henry Y. Chang

(3) a. ini-mu mah-i ka sino nii NEG-1S.GEN drink-LF NOM wine this ‘I did not drink the wine.’ b. ini mah-ani begu na laqi ka taku NEG drink-I/BF soup GEN child NOM spoon ‘The child did not use the spoon for soup.’ (Chen 1996:44-46)

Given the reconstructions of PAn focus morphology advocated by Wolff (1973) and Ross (1995), Tsou ‘atemporal’ focus affixes are more likely to be innovations rather than retentions. The innovations, however, do not seem to be attested in Kanakanavu and Saaroa. Let’s consider the focus system in Kanakanavu first. According to Mei (1982), there is a three-way distinction in focus marking in Kanakanavu, as indicated below:

Table 2: The focus system in Kanakanavu (Based on Mei 1982)

AF PF LF Neutral -ai Perfective ni- Imperfective , m-3 -un -a(n)

Unlike Tsou, the focus system in Kanakanavu presumably did not undergo any drastic morphological changes. The emergence of the neutral PF suffix -ai in Kanakanavu is not common, but it does not parallel the PF suffix -a in Tsou in its form and function.4 As noted by Tsuchida (1976:51) and Mei (1982:219), -ai is restricted to verbs bearing neutral aspect, as illustrated below:

(4) ulung-ai-kani ’inia sua takuis-ini kava-cumai take.off-OF2-IND5 he jacket-his skin-bear ‘He took off his bearskin jacket.’ (Mei 1982:219)

3 According to Tsuchida (1976:47-48), there are three more AF affixes attested in Kanakanavu: mu-. um-, and -em-. The prefix mu- occurs with a stem beginning with a bilabial or ng, as in mu-a-pucupucu ‘to crumple’; the prefix um- occurs with a stem beginning with a , as in um-a-ala ‘to take’; -em- occurs after the perfective aspect infix -in-, as in c-in-em-e’era ‘to see’. 4 Tsuchida (1976: 49) treats -ai as a marker of special focus and notes that “the function of the verb form here …is not clear.” 5 Contra Mei (1982), Tsuchida (1976:52) identifies the suffix -kani as a marker of quotation rather than indicative mode.

568

Rethinking the Tsouic Subgroup Hypothesis

Meanwhile, imperfective PF and perfective PF are represented by -un and ni- respectively in Kanakanavu, which are quite comparable to the focus affixes widely observed across western Austronesian languages. Compare:

(5) a. tia-kasu itaru-n will-you wait-OF1;IPFV ‘I’ll wait for you.’ (Mei 1982:213) b. ni-pia-pacai cau tutui kill:OF1;PFV person pig ‘Someone killed the pig.’ (Mei 1982:221)

In Kanakanavu, Locative Focus (LF) is represented by the wide-spread LF suffix -an, as in:

(6) na ta-u-canum-a ni-p-ia-pacal-an-aku tutui at place to draw water kill:LF;PFV-I pig ‘It is at the place to draw water that I have killed the pig.’ (Mei 1982:223)

It should also be noted that there is no marker of Benefactive/Instrumental focus (B/IF) reported in the previous literature (Tsuchida 1976, Mei 1982, Li et al. 1997). However, Chun-min Wu (personal communication) recently discovered that B/IF is marked by the prefix se-, as in:

(7) se-kun=maku uru aratin isi IF-eat=1S.GEN cooked.rice chopstick this ‘I use this chopstick to eat rice.’ (courtesy of Chun-ming Wu)

Given this finding, Kanakanavu focus system will look more like the regular four-way distinction system:

Table 3: The focus system in Kanakanavu (Revised)

AF PF LF B/IF Neutral -ai Perfective ni- Imperfective , m- -un -a(n) se-

569

Henry Y. Chang

Now let’s turn to Saaroa. Saaroa focus system does not seem to parallel Tsou either. Compare:

Table 4: The focus system in Saaroa (Based on Li et al. 1997)

AF PF LF B/IF Perfective li- general -um-, um-, u- -a -a(na) sa(a)-

Among the focus markers mentioned above, only the PF -a looks like the PF suffix -a in Tsou. Nonetheless, the PF suffix -a in Saaroa is not as productive as it is in Tsou. It is restricted in its occurrence. First, it will be deleted in the presence of the perfective prefix li-, as in (8):

(8) li-ala na cucu’u ka vutukulu PF-take GEN person NOM fish ‘The person took the fish away.’ (Li et al. 1997:279)

According to Paul Li and John Wolff (personal communication), Saaroa’s perfective prefix li- is a counterpart of ni- in Kanakanavu – it indicates PF when the regular PF focus markers are absent. Such an option is not available in Tsou. I will return to the issue in the next section. Second, as noted by Tsuchida (1976: 77), the PF suffix -a does not occur with the future marker a-, as in a-ala ‘take (PF, FUT)’. The restrictions suggest that the PF suffix -a is grammatically peripheral in Saaroa. Leaving it aside, the focus system in Sarroa will be very much the same as Kanakanavu, which departs substantially from Tsou.

3. The complete loss of the PAn perfective marker

Among Formosan languages, Tsou is the only language that does not realize the PAn perfective marker in its nominal and verbal clauses (Starosta 1985), as indicated in the following table (see also Ross 1995, 2002, Zeitoun 2006):

570

Rethinking the Tsouic Subgroup Hypothesis

Table 5: The reflexes of the PAN perfective marker in Formosan languages6

Perfective markers Language Verbal clause Nominal clause (Relative clause/nominalization) Mayrinax Atayal Seediq / Saisiyat Pazih Thao Bunun //-in / Kanakanavu /ni- Saaroa li- li- Paiwan na-(AF), (NAF) Kavalan ni-/ (=ti) Amis -- (tu)7 Puyuma -- (la) Rukai -- (-nga) 8 Tsou -- (-cu) --

With respect to the realization of the PAn perfective marker, Formosan languages can be classified into three types. In Type I, which includes most Formosan languages, the reflexes can be found in both verbal and nominal clauses. Take Mayrinax Atayal for example:

6 The table is based on H. Chang (2000), Y. Chang (2000), L. Huang (2000), Y. Lin (2000), Zeitoun et al. (1996:51). 7 There is a prefix na- occasionally prefixed to verbs in Amis. At first sight, it seems to look like Paiwan perfective na-. However, Amis na- should not be analyzed as a reflex of the PAn perfective marker on the grounds that it conveys experiential aspect instead of perfective aspect (Tsai-hsiou Liu, personal communication). Compare: (i) a. tayra tu i taypak cingra go PFV LOC Taipei he ‘He has gone to Taipei.’ b. na-tayra i taypak cingra EXP-go LOC Taipei he ‘He had been to Taipei.’ 8 As noted by an anonymous reviewer, the infix is not productive in all the Rukai dialects.

571

Henry Y. Chang

(9) a. maniq cu’ aqih na’ naniqun eat ACC bad GEN food ‘He ate spoiled/bad food.’ b. patiq write ‘word, handwriting’ (L. Huang 2002:215-216)

This is also the case in Paiwan:

(10) a. su-cauv=aken katiaw? 2S.GEN-cover=1S.NOM yesterday ‘Did you cover me with a quilt yesterday?’ b. cavu-cavu=aken tua cavu RED-cover=1S.NOM OBL millet-cake ‘I am making the millet cake.’

In Type II, which include Amis, Puyuma, and Rukai, the reflexes are attested in nominal clauses but not in verbal clauses. Take Amis and Puyuma for instance:

(11) a. o maan ko safel no mita? O what NOM cook GEN 1P.POSS ‘What is our cooked dish?’ b. ma-safel-ay to no mako ko titi no mangcel PF-cook-AY already GEN 1S.POSS NOM meat GEN muntjac ‘I have cooked the muntjac’s meat.’ (Amis, courtesy of Tung-chiou Huang) (11’) a. amuna saDu na asi tu=TekeL-an but many NOM milk his.NOM=drink-NMLZ ‘But his drinking milk was a lot.’ (Puyuma, Teng 2006:28) b. Tekel-ku la Da enay drink-1SG.NOM already OBL water ‘I drank water (already).’ (Puyuma, Zeitoun et al.1996:28)

As shown in (11-11’a), the PAn perfective marker realizes as the infix in nominal clause. In contrast, the infix is gone in verbal clause and its role has been taken over by the aspectual markers to and la respectively, as illustrated in (11-11’b). In Type III, the perfective marker is not attested at all. This is the case only found in Tsou.

572

Rethinking the Tsouic Subgroup Hypothesis

(12) a. mi-ta=cu bonu ta naveu AF-3S.NOM=PFV eat(AF) OBL rice ‘He has eaten rice.’ b. ’oanu (cf. ana, PF) ‘food’ c. hiafa (cf. hafa, PF) ‘thing carried’ (Tung 1964:612)

As in (12a), the perfective function is carried out by the enclitic =cu in verbal clause. Unlike many other Formosan languages where patient nominals formally resemble perfective PF verbs, Tsou patient nominals are distinct in form from their verbal counterparts, as in the contrast ’oanu ‘food’ vs. ana ‘to eat’ (PF). Accordingly, the patterns of the Formosan realizations can be summarized as follows:

Table 6: The paradigm of the Formosan reflexes of the PAN perfective marker

verbal clause nominal clause Type I Type II -- Type III -- --

It is evident that the complete loss of the PAn perfective marker in Tsou is an innovation and that it is not shared by any other Formosan languages. This is another cue for the particular position of Tsou in the Austronesian family.

4. The focus harmony restriction

As discussed extensively in Chang (2005), verbs in serial verb constructions (SVCs) exhibit the Focus Harmony Restriction (FHR) in Tsou, viz. verbs are required to agree with each other in their focus inflection (see also Starosta 1985:291):

(13) a. mi-’o kaeb-u bon-u ta tacumu AF-1S like-AF eat-AF OBL banana ‘I like to eat bananas.’ b. os-’o kaeb-a an-a ’o tacumu NAF-1S like-PF eat-PF NOM banana ‘I like to eat the bananas.’

573

Henry Y. Chang

This contrasts sharply with other Formosan languages, where the second verbs in SVCs are normally required to be inflected for Actor focus (hence the AF-only Restriction) (L. Huang 1997, Y. Chang 2006a). The FHR is presumably a recent development, given that it is not attested in Formosan languages such as Mayrinax Atayal and Paiwan which are generally believed to be more grammatically conservative. It is likely that the AF-only restriction is common in PAn and the FHR is a Tsou innovation. It is noteworthy that the FHR does not hold of Kanakanavu. In Kanakanavu, the second verb in SVCs is not required to be inflected in the same manner as the first verb. Instead, it is subject to the usual AF-only restriction. For example:

(14) a. ala-ene-kani-kiai taini aluvalu canumu take(GF)-IPFV-is.said-by.him throw.away rapids water ‘She took it to throw away in the rapids of water.’ (Tsuchida 1976:51) b. urupaca-ku aratin ka-kun use(AF)-1S.NOM chopstick RED-eat ‘I use chopsticks to eat.’ c. urupac-n-maku aratin isi ka-kun use-PF-1S.GEN chopsticks this RED-eat ‘I use this chopstick to eat.’ (b-c as courtesy of Chun-ming Wu)

Neither is it observed in Saaroa, as shown below:

(15) saa-p-atu-cici-a-ami ka ’ilicu ka ’urai make-hot:SF-NEUT-is.said OBL evil_spirit NOM oil u-vura isana give:AF-NEUT him:OBL ‘The evil spirit boiled the oil to give him.’ (Tsuchida 1976:76)

Unlike their Tsou counterparts, the second verbs do not agree with the first verbs in their focus inflection in Kanakanavu and Saaroa, as shown in (14-15). The FHR is not observed in the so-called “southern Tsouic” languages.

5. The NAF-only causatives

Morphological causative is generally represented by the affix pa- or its variants across Austronesian languages and the causative affix *pa- is reconstructible to PAn (Wolff 1973). While similar in form to the PAn *pa-, its major reflex in Tsou, i.e., poa-, observes a peculiar restriction – it can only be used in NAF constructions (G. Lin 2002,

574

Rethinking the Tsouic Subgroup Hypothesis

2005, Huang and Huang 2005, C-W. Chang 2006). It follows that causatives prefixed with poa- must take either causees or patients as their subjects and that causers are excluded from the subject position in Tsou. For example:

(16) a. i-si poa-bon u ta mameoi to fou ’o ino-si NAF-3S CAUS-eat(AF) OBL old_man OBL meat NOM mother-3S ‘The old had his mother eat meat.’ a’. *mi-ta poa-bon u ta ino-si to fou ’o mameoi AF-3S CAUS-eat(AF) OBL mother-3S OBL meat NOM old.man b. i-si poa-an-eni ta mameoi ta maaea si simeo NAF-3S CAUS-eat-IF OBL old_man OBL Japanese NOM fat.meat ‘The Japanese had the old man eat fat meat.’ (Starosta 1985:289)

In (16a, c), the subjects are causee and patient respectively. The sentence will be ungrammatical if a causer serves as the subject, as shown in (16a’). As a result, the combination of the AF auxiliary mi/mo/moh and the causative marker poa- is not allowed in Tsou. This restriction is unheard in other western Austronesian languages and, to my knowledge, unattested in any world’s language. Such a peculiar restriction is likely to be an innovation instead of a retention. The innovation is not observed in Kanakanavu and Saaroa. In Kanakanavu, a causer can freely occur as the subject of a causative sentence. For example:

(17) a-pa-kun=ku cuma FUT-CAUS-eat (AF)=1S.NOM Father ‘I will make Father eat something.’ (courtesy of Chun-ming Wu)

Neither is it attested in Saaroa. Compare:

(18) paa-kita ka ina-ku mamaini na sala’a CAUS-see NOM mother-my child OBL road ‘My mother showed the child the road.’ (Li et al. 1997:289)

As shown in (17-18), the causers =ku and ina-ku can serve as the subjects of the causative sentences. In this regard, Kanakanavu and Saaroa also differ from Tsou.

575

Henry Y. Chang

6. The obligatory auxiliary constraint

It is widely observed that almost every sentence must start with an auxiliary in Tsou, which indicates /focus and temporal/modal information (Tung 1964:88-89, Tsuchida 1976, Starosta 1985:184) (the Obligatory Auxiliary Constraint, shortened as OAC). The OAC is observed in the above examples (1), (13), and (16), which differentiates Tsou from other Austronesian languages. For instance, in expressing a situation like ‘Father drank the wine,’ Tsou contrasts with typical Formosan languages like Paiwan in its clause structure. Compare: (to be continued on next page)

(19) The clausal structure in Tsou Aux BP Asp Verb Case Noun Case Noun marker marker (focus, (number, (temporal) (focus) (case, (case, mood) case) visibility, visibility, proximity) proximity) oh -ta -cu im-a to amo ’o emi drink father wine

(20) The clausal structure in Paiwan Verb Case Noun Case Noun marker marker (focus, (case, (case, aspect) noun noun class) class) tekel ni kama a vava drink father wine

Is the Tsou auxiliary system a retention or an innovation of PAn? Starosta (1985) claims that it is a retention:

Aspect-marking auxiliary verbs must have been common in PAn, since they are also very frequent in Atayalic as well as Tsou. (Starosta 1985:184)

However, Starosta’s claim is not well-grounded. First, auxiliaries indicating tense/ aspect/mood (TAM) cannot be unequivocally reconstructed all the way back to PAn, given the lack of regular sound correspondence with respect to the auxiliaries found across Formosan languages, as illustrated below. Compare: (to be continued on next page)

576

Rethinking the Tsouic Subgroup Hypothesis

Table 7: TAM-marking auxiliaries in Formosan languages9

Realis Irrealis Mayrinax hani’an (immidiate), -- AF/NAF Atayal kia’ (remote) wan (perfective) musa’ Wulai Atayal AF/NAF nyux/cyux (progressive) wada (perfective/inceptive) maha Seediq AF/NAF tena (perfective) gisu/gaga (progressive) ’ina (experiential/perfective) ’am AF mang (progressive) Saisiyat NAF ma (progressive) -- AF/NAF kin (repetitive/incremental) -- Pazih AF/NAF -- -- Thao AF/NAF -- -- Bunun AF/NAF -- -- Kanakanavu AF/NAF esi (progressive) tia-/te- Saaroa AF/NAF -- -- Paiwan AF/NAF -- -- Rukai AF/NAF -- -- Puyuma AF/NAF -- --10 Amis AF/NAF -- -- pun (perfective), -- Kavalan AF/NAF yau (progressive) ya (present) -- Yami AF/NAF jana/zana/teika (perfective) mo(h)-, moso (remote) -- AF mi-, mo (immdiate) o(h)- (remote), -- Tsou NAF i- (immdiate) la (habitual) ta-, nto(h), ntoso (remote) AF/NAF te, tena, nte (immdiate)

9 The table is based on various sources: the twelve reference grammars written in Chinese by a number of authors and published by Yuan-liou Publishing Co., the grammatical sketches by Li et al. (1997) and L. Huang (1993), and Ho’s (1990) M.A. thesis. 10 Contra previous authors (H -C. Chang 2000 and L. Huang 2000), I do not identify the irrealis markers (uri/uru= in Paiwan and aru= in Puyuma) as auxiliaries since, unlike ordinary auxiliaries in other Formosan languages, these markers do not attract clitics.

577

Henry Y. Chang

Second, some auxiliaries in Tsou (e.g., moh and oh) are likely to derive from the PAn motion verb *muSa ‘to go’ through metaphorical extension and the sound change *S > h. The development is not surprising. It has been observed cross-linguistically that auxiliaries originate from lexical verbs (Lightfoot 1979, 1991). Accordingly, contra Starosta (1985), I will regard the Tsou auxiliary system and the OAC as an innovation instead of a retention. It is noteworthy that the innovation is not attested in Kanakanavu and Saaroa, as shown in Table 7 and previous examples (see examples (4-7) for Kanakanavu, and example (8) for Saaroa).

7. The emergence of 3rd singular nominative bound pronouns

Except for Tsou, there is no overt realization of third person singular nominative bound pronoun across Formosan languages (Starosta 1985, L. Huang 1999, Zeitoun 2000a:83). The slot for the third person singular nominative bound pronoun is thus generally left blank in the PAn pronominal systems reconstructed by various authors (Dahl 1973, Blust 1977, Ross 2002, 2006, Li 2006). Along this line of thought, the occurrence of the third person singular nominative bound pronoun in Tsou should be an innovation. For instance:

(21) mi-tai baito to o’oko ’e pasuyai AF-3S see(AF) OBL children NOM Pasuya ‘Pasuya saw children.’

As in (21), the pronominal suffix -ta is co-referential with the third person singular subject pasuya. Patterns like this are not observed in Kanakanavu or Saaroa. In Kanakanavu and Saaroa (and actually also in other Formosan languages), third person singular nominative bound pronoun is not overtly represented.11 Compare:

(22) a. m-u-a-ca +kani ’anupu AF-go-IPFV is_said hunt ‘He goes hunting.’ (Kanakanavu, Tsuchida 1976:52) b. kaaiu na ia, marumuku na kaaiu na that TOP love ACC that ‘He loves her.’ (Saaroa, Li et al. 1997:290)

11 Li (1973:78) first observed this for Rukai.

578

Rethinking the Tsouic Subgroup Hypothesis

Unlike (21), the sentences in (22a-b) do not consist of any bound pronoun while their subjects are interpreted as third person singulars.

8. Summary and concluding remarks

I have enumerated six morphosyntactic innovations which are exclusively attested in Tsou, but not in Kanakanavu or Saaroa, as summarized in the following table:

Table 8: The morphosyntactic innovations in the “Tsouic” languages

Languages PAn Tsou Kanakanavu Saaroa MS innovations Focus morphology *, m-, /m-/mu- /um-/u-, *-en, -a, -ai/ni-/-un, -a, *-an, -i, -a(n), -a(na), *si- -eni se- sa(a)- Perfective marker * -- ni- li- Focus marking of * (AF) varies with the (AF) (AF) the lower verbs higher verb in verb sequences (AF/NAF) Focus alternation *pa- (AF), -- (AF) pa- (AF), pa- (AF), of causative *pa--en (PF), poa--a (PF), pa--un (PF) …(NAF)12 verbs *pa--an (LF), poa--i (LF), *si-pa- (B/IF) poa--eni (B/IF) TAM-marking -- mi-/i- etc. esi, -- Auxiliary te- etc., tia/te- la- 3rd singular -- -ta -- -- nominative bound pronoun

It is also suggested that none of these Tsou innovations are found elsewhere in the Austronesian family. What do the findings tell us about the genetic relationships of Tsou, Kanakanavu, and Saaroa? There are basically two ways of interpreting the findings. One can maintain the TSH by dismissing the findings and argue that Tsou made the

12 I leave the causative NAF forms blank because there is no relevant data available to me at this moment.

579

Henry Y. Chang innovations after it split off from the subgroup. Under this view, the Tsou innovations will have no bearing on the TSH. What they indicate is simply the fact that Tsou innovated far faster than the other two Tsouic languages. This approach is suggested to me by , Shigeru Tsuchida, and John Wolff (personal communication). The problem with this approach is that there is no firm evidence for the TSH at this point13. On the other hand, we might choose to rethink the long held TSH in the light of the findings. Given that linguistic subgroupings are based on exclusively shared innovations and that Tsou does not share its distinctive innovations with any other Austronesian languages, it is by on means impossible that Tsou is one of the first offshoots from PAn rather than subordinate under a Tsouic subgroup together with Kanakanavu and Saaroa. This approach is adopted by Starosta (1985), where Tsou is argued to be coordinate with non-Tsou as the two primary branches of PAn. We are in favor of Starosta’s approach, but we do not agree with his treatment of all the Austronesian languages other than Tsou as a subgroup on the grounds that there is no single innovation, phonological or morphosyntactic, which is exclusively shared by the non-Tsou. A more desirable hypothesis is that Tsou is conjoined with some other PAn primary branches such as Atayalic, Malayo-Polynesian, etc. in a rake-like manner as conceived in Blust (1999). As to the linguistic position of Kanakanavu and Saaroa, we will not make any inference at this point due to lack of sufficient linguistic data. We will leave it for future study.

13 John Wolff (personal communication) contends that there are two mergers which are exclusively found in the Tsouic languages, viz., the merger of *j, *d, and *c and that of *g and *y. This, however, has not been widely recognized by Austronesianists, pending further verification. .

580

Rethinking the Tsouic Subgroup Hypothesis

References

Blust, Robert A. 1999. Subgrouping, circularity and extinction: some issues in Austronesian comparative linguistics. Selected Papers from the Eighth Interna- tional Conference on Austronesian Linguistics, ed. by Zeitoun Elizabeth and Paul Jen-kuei Li, 31-94. Taipei: Institute of Linguistics, Academia Sinica. Blust, Robert A. 2003. Three notes on early Austronesian morphology. Oceanic Linguistics 42:438-78. Chang, Hsiu-chuan. 2000. A Reference Grammar of Paiwan. Taipei: Yuanliou. (in Chinese) Chang, Henry Yungli. 2000a. A Reference Grammar of Seediq. Taipei: Yuanliou. (in Chinese) Chang, Henry Yungli. 2000b. A Reference Grammar of Kavalan. Taipei: Yuanliou. (in Chinese) Chang, Henry Yungli. 2005. Focus harmony and restructuring in Tsou. Paper presented at AFLA-12. Los Angeles: UCLA. Chang, Henry Yungli. 2006. The guest playing host: adverbial modifiers as matrix verbs in Kavalan. Clause Structure and Adjuncts in Austronesian Languages, ed. by Hans-Martin Gaertner, Paul Law and Joachim Sabel, 43-82. Berlin: Mouton de Gruyter. Chang, Chi-wei. 2006. Causative and resultative constructions in Tsou. : National Chung-Cheng University. Manuscript. Chen, Chie-hui. 1996. A Preliminary Study of Negation in Tkdaya Seediq. Hsinchu: National Tsing Hua University MA thesis. (in Chinese) Dahl, O.C. 1973. Proto-Austronesian. Scandinavian Institute of Asian Studies Monograph 15. Lund: Studentliteratur. Dyen, Isidore. 1963. The position of the Malayopolynesian languages of Formosa. Asian Perspectives 7.1-2:261-271. Dyen, Isidore. 1965. Lexicostatistical classification of the Austronesian languages. Indiana University Publications in Anthropology and Linguistics Memoir 19, supplement to IJAL. Ferrell, Raleigh. 1969. Taiwan Aboriginal Groups: Problems in Cultural and Linguistic Classification. Institute of Ethnology, Academia Sinica, Monograph No. 17. Ho, Arlene. 1990. Yami Structure: A Descriptive Study of the . Hsichu: National Tsing Hua University MA thesis. Huang, Lillian. 1993. A Study of Atayal Syntax. Taipei: Crane. Huang, Lillian. 1997. Serial verb constructions in some Formosan languages. Paper presented at 9-ICAL. Taipei: Academia Sinica.

581

Henry Y. Chang

Huang, Lillian. 2000a. A Reference Grammar of Atayal. Taipei: Yuanliou. (in Chinese) Huang, Lillian. 2000b. A Reference Grammar of Puyuma. Taipei: Yuanliou. (in Chinese) Huang, Lillian. 2002. Nominalization in Mayrinax Atayal. Language and Linguistics 3.2:197-226. Huang, Lillian, Marie Yeh, Elizabeth Zeitoun, Anna Chang, and Joy Wu. 1999. A typological overview of pronominal systems of some Formosan languages. Selected Papers from the Fifth International Conference on Chinese Languages, ed. by H. Samuel Wang, Fengfu Tsao and Chinfa Lien, 165–198. Taipei: Crane. Huang, Shuanfan, and Huei-ju Huang. 2005. Causative and applicative: Their split and syncretism in Formosan languages. Paper presented at Taiwan-Japan Joint Workshop on Austronesian Languages. Taipei: National Taiwan University. Li, Paul Jen-kuei. 1973. Rukai Structure. Taipei: Academia Sinica. Li, Paul Jen-kuei. 2006. Personal Pronouns in Proto-Austronesian. Talk given at The International Conference Celebrating The Publication of the Complete Works of Li Fang-Kuei and The Publication of the Bulletin of Chinese Linguistics. Academia Sinica. Li, Paul Jen-kuei, Cheng-hwa Tsang, Yingkuei Huang, Dah-an Ho, and Chiuyu Tseng. (eds). 1995. Austronesian Studies Relating to Taiwan. Taipei: Institute of History and Philology, Academia Sinica. Li, Paul Jen-kuei, Dah-an Ho, Mei-jing Huang, Elizabeth Zeitoun, and Claire Saillard. (eds). 1997. Austronesian Languages in Kaohsiung County. Kaohsiung: Kaohsiung County Government. (in Chinese) Lightfoot, David W. 1979. Principles of Diachronic Syntax. Cambridge: Cambridge University Press. Lightfoot, David W. 1991. How to Set Parameters: Arguments from Language Change. Cambridge: MIT Press. Lin, Gu-jing. 2002. The Complementation of Tsou: Toward a Cognitive Account of Syntactic Structures. Taipei: National Taiwan University MA thesis. Lin, Gu-jing. 2005. The causee actor: on the causative pattern of Tsou. Paper presented at Taiwan-Japan Joint Workshop on Austronesian Languages. Taipei: National Taiwan University. Lin, Ying-chin. 2000. Pazih. Taipei: Yuanliou. (in Chinese) Mei, Kuang. 1982. Pronouns and verb inflection in Kanakanavu. Tsinghua Journal of Chinese Studies, New Series 14:207-232. Ross, Malcolm. 1995. Reconstructing Proto-Austronesian verbal morphology: evidence from Taiwan. Austronesian Studies Relating to Taiwan, ed. by Paul Jen-kuei Li et al., 727-792. Taipei: Institute of History and Philology, Academia Sinica.

582

Rethinking the Tsouic Subgroup Hypothesis

Ross, Malcolm. 2002. The history and transitivity of western Austronesian voice and voice-marking. The History and Typology of Western Austronesian Voice Systems, ed. by Wouk Fay and Malcolm Ross, 17-62. Pacific Linguistics, 518. Canberra: The Australian National University. Ross, Malcolm. 2006. Reconstructing the case-marking and personal pronoun systems of Proto Austronesian. Canberra: Australian National University. Manuscript. Starosta, Stanley. 1985. Verbal inflection versus deverbal nominalization in PAn; The evidence from Tsou. Austronesian Linguistics at the 15th Pacific Science Congress, ed. by Andrew Pawley and Lois Carrington, 281-312. Pacific Linguistics, C-88. Canberra: The Australian National University. Teng, Stacy F.-C. 2006. The core-oblique distinction in Puyuma. Paper presented at the colloquium of Institute of Linguistics, Academia Sinica. Tsuchida, Shigeru. 1976. Reconstruction of Proto-Tsouic Phonology. New Haven: Yale University dissertation. Tokyo: Study of Languages & Cultures of Asia & Africa, Monograph Series No.5. Tung, T’ung-ho. 1964. A Descriptive Study of the , Formosa. Taipei: Institute of History and Philology, Academia Sinica. Wolff, John U. 1973. Verbal inflection in Proto-Austronesian. Parangal Kay Cecilio Lopez, ed. by Andrew B. Gonzalez, 71-91. Quezon City: Linguistic Society of the . Wu, Joy Ching-lan. 2000. A Reference Grammar of Amis. Taipei: Yuanliou. (in Chinese) Yeh, Marie Meili. 2000. A Reference Grammar of Saisiyat. Taipei: Yuanliou. (in Chinese) Zeitoun, Elizabeth. 1996. The Tsou Temporal, Aspectual and Modal System Revisited. Bulletin of the Institute of History and Philology 63:503-532. Zeitoun, Elizabeth. 2000a. A Reference Grammar of Tsou. Taipei: Yuanliou. (in Chinese) Zeitoun, Elizabeth. 2000b. A Reference Grammar of Bunun. Taipei: Yuanliou. (in Chinese) Zeitoun, Elizabeth. 2000c. A Reference Grammar of Rukai. Taipei: Yuanliou. (in Chinese) Zeitoun, Elizabeth, L. Huang, M. Yeh, A. Chang, and J. Wu. 1996. The temporal/ aspectual and modal systems of the Formosan languages: a typological perspec- tive. Oceanic Linguistics 35.1:21-56. Zeitoun, Elizabeth, and Lillian Huang. 2006. Preliminary remarks on nominalization in Formosan languages. Paper presented at AFLA-13. Hsinchu: National Tsing Hua University.

583