Grammaticalization in Squliq Atayal
＊Lillian M. Huang Shih Chien University
Atayal is an Austronesian language spoken in Taiwan. This paper presents a descriptive study of the grammaticalization in Squliq Atayal, one of the two major dialects of the language. Several examples of grammaticalization in the named dialect are examined, including (i) verbs like musa’/mosa’, wal/wayal, nyux/cyux being grammaticalized into auxiliaries designating tense/aspect/mood; (ii) the locative focus (LF) form of the verb malax as used in negative and/or imperative constructions, i.e.
laxi, serving as a negator; (iii) the ‘saying’ verb mha being treated as a complementizer
and an evidential marker; (iv) the 3rd person singular pronoun hiya’ serving as an emphatic marker; and (v) the interrogative word nanu’ functioning as a pause filler. The original meanings of the above-mentioned words are relatively concrete, and the new meanings are more abstract and have more grammatical functions. It is also noticed that these words are all polysemous; that is, their original meanings/functions co-exist with later developed ones.
Key words: Atayal, Austronesian, auxiliary, complementizer, emphatic, evidential marker, filler, grammaticalization, negator, polysemous, pronoun, Squliq, verb
1. Introduction
Grammaticalization has recently attracted much interest and attention. It generally refers to a linguistic process ‘whereby lexical items and constructions come in certain linguistic contexts to serve grammatical functions, and, once grammaticalized, continue to develop new grammatical functions. Thus nouns and verbs may change over time into grammatical elements such as case markers, sentence connectives, and auxiliaries…’ (Hopper and Traugott 1993). The phenomenon of grammaticalization appears to be rather universal; a wide number of languages around the world illustrate examples of grammaticalization. For instance, in Mandarin Chinese, as L. Huang (1988) states, the locative verb za$i ‘be at’ and the motion verb guo$ ‘cross; pass’ may serve as aspectual markers as well as foreground-background indicators; that is, in addition to serving as verbs, za$i may serve as an on-going aspect marker and foreground indicator, and guo$ a perfective as well as background marker. Biq (1991) while examining the personal pronoun ni& ‘you (singular)’ in Mandarin Chinese,
＊
I would like to show my gratitude to the National Science Council for supporting my research on the grammaticalization of Atayal as well as some other Formosan languages (NSC 93-2411-H-003-002-MD). Part of this paper was presented at the International Symposium on
Austronesian Cultures: Issues Relating to Taiwan at Academia Sinica, December 8-11, 2001. I
would like to thank Professors Laurence Reid, Stanley Starosta, Paul Li and Malcolm Ross for their comments and questions during that presentation. I would also like to thank Dr. Elizabeth Zeitoun, Mr. Tali’ Hayung, and in particular, the two anonymous reviewers of this journal for their very detailed and useful comments and suggestions on this paper.
concludes that the named pronoun often carries dramatic, impersonal and metalinguistic functions besides referring to the second person ‘you’. Yang (1992), while investigating the word na$ ‘that’, points out that it is not only a distal demonstrative but has developed into a discourse marker of transition, indicating the speaker’s intention of continuity. Wang, Katz and Chen (2000) illustrate that the speech verb shuo# ‘say’ has various functions according to its syntactic distributions, e.g., providing impersonal quotative or generalized inferential meaning in the sentence initial position, serving as a complementizer when following mental or reportative words to mark an object clause (i.e. in sentence medial position), and functioning as a marker of counter-expectation, mitigation and intensification in the sentence final position. Wang (2002) demonstrates the evolvement of guo$ from a full-fledged motion verb ‘cross; go through; pass by’ to a verbal complement, and an experiential aspect marker. Lily Su (2004) further demonstrates, following Traugott (1989), that the process known as subjectification should take full responsibility for the development of the shift of shuo# as a main verb, to a complementizer, and finally to an epistemic particle. Wang and Huang (2006), in terms of the concept of subjectification and intersubjectification, and Blakemore’s relevance-based framework, present the grammaticalization of three connectives yīnwei ‘because’,
suǒyi ‘so’ and ránhou ‘then’. Similar examples of grammaticalization can also be
found in other Chinese languages spoken in Taiwan, e.g. Min (cf. Chang 1998, 1999, 2002, Li 2002a, 2002b, and Li and Liu 1995) and Hakka (cf. Lai 1999, 2001, 2002, 2003a, 2003b).
In addition to the above-mentioned Chinese languages, several Austronesian languages spoken in Taiwan also present instances of grammaticalization (cf. Lin 2005, ms, Tang 1999, Huang 2001b, and Rau 2002). Take Atayal for example. It is spoken principally in the mountainous area of northern Taiwan, and has two major dialects, namely, Squliq and C’uli’, with the former being regarded as more innovative. Consequently, there are more cases of grammaticalization that can be found in Squliq Atayal. The present paper thus attempts to examine the examples of grammaticalization in Wulai Atayal, a variant of the Squliq dialect spoken in Wulai Hsiang, Taipei Prefecture. The data upon which this paper is based mainly come from the speech of Pastor Temu Pehu and Mr. Li Kuang-ming, among some other speakers, collected on and off between October 1988 and January 2005. A sample of the texts provided by the Atayal speakers is given in Appendix, for readers’ reference.
The present research is a first attempt at examining grammaticalization in Squliq Atayal. In what follows, several examples of grammaticalization in the named variant are examined, including (i) verbs like musa’/mosa’1, wal/wayal2, and nyux/cyux being
1
grammaticalized into auxiliaries designating tense/aspect/mood; (ii) the locative focus (LF) form of the verb malax as used in negatives and/or imperatives, i.e. laxi, serving as a negator in imperative constructions; (iii) ‘saying’ verb mha being treated as a complementizer and an evidential marker; (iv) the 3rd person singular pronoun hiya’ serving as an emphatic indicator; and (v) the interrogative word nanu’ functioning as a filler. As will be noted, the new uses/functions of the words under examination co-exist with their original meanings; in other words, these words are all polysemous. Furthermore, as will be shown below, while such original meanings seem relatively concrete, the new meanings are much more abstract and even emptier.
2. Verbs and auxiliaries
This section will present several Atayal verbs illustrating instances of polysemy. In other words, these verbs have been grammaticalized and may function like tense/aspect/mood auxiliaries, while they still retain their functions and serve as main verbs in different contexts. Below I will first give a brief introduction to the properties of sentence structures of Squliq Atayal. I will then examine musa’ and mosa’ designating a motion ‘going’ and futurity/irrealis, wal and wayal indicating a past event ‘went’ and past/realis meaning, and nyux and cyux marking ‘existence, possession, location’ and on-going events, respectively.
2.1 A brief introduction to Atayal syntactic structures
Atayal is a verb-initial language,3 in which the verb ordinarily, but not always precedes the other elements in a sentence and attracts the bound pronouns (if there is any), as illustrated by the following examples with the verbs being boldfaced and the bound pronouns in italics:4
Ministry of Education and the Council of Indigenous Peoples’ Affairs, on December 15, 2005. Most of the letters are the same as their IPA symbols, except for the following: ’=> glottal stop ///; b=>voiced bilabial fricative /B/; g=>voiced velar fricative /ƒ/; ng=>velar nasal /N/; n_g=>consonant cluster /n/ & /g/; y=>palatal glide /j/.
2
In Wulai Atayal, when /l/ appears in the word final position, it is often pronounced as [n]. Thus words like wal, wayal, huzil, ngasal and kneril are often pronounced as [wan], [wayan], [huzin], [ngasan] and [knerin], respectively.
3
Please refer to Huang (1993, 2008) and Rau (1992) for a more detailed discussion of the Atayal structures.
4
The abbreviations and symbols used in this paper are as follows. 1PE: First person plural exclusive; 1PI: First person plural inclusive; 1S: First person singular; AF: Agent focus; BF: Beneficiary focus; DIST: Distal; EMP: Emphatic usage; EVI: Evidential marker; GEN: Genitive; IF: Instrument focus; IMPRF: Imperfective aspect; IRR: Irrealis; LF: Locative focus; LOC: Locative; NAF: Non-agent focus; NEG: Negator; NEU: Neutral; NOM: Nominative; PART: Particle; PF: Patient focus; PRF: Perfective aspect; PROX: Proximal; Q: Question; QUO: Quotative marker; REC: Reciprocal; RED:
(1) a. m-qwas laqi’ qasa krryax AF-sing child that every.day ‘That child sings every day’ b. p-tucing=saku’ huzil
IRR-beat.AF=1S.NOM dog ‘I will beat a/the dog’ c. bhiy-un=mu sayun
beat-PF=1S.GEN Sayun ‘Sayun will be beaten by me’ d. biq-an=saku’=nya’ pila’
give-LF=1S.NOM=3S.GEN money ‘I was given money by him’
e. s-phapuy=mu (qu’) yaba’ BF-cook=1S.GEN NOM father ‘I cooked for Father’
f. s-phapuy=mu (qu’) bagah IF-cook=1S.GEN NOM charcoal ‘I cooked with charcoal’
As for the negative counterparts in the above sentences, the negators ini/ (used for past/realized events, including habitual ones) and iyat (used for future/unrealized events) have to appear sentence initially and are attached with the bound pronouns (if there are any). The verbs will then follow the negators and will be affixed with appropriate focus markers, as presented below:
(1) a’. ini’ pqwas laqi’=mu krryax NEG sing.AF child=1S.GEN every.day ‘My child does not sing every day’
b’. iyat=saku’ p-tucing huzil NEG=1S.NOM IRR-beat.AF dog ‘I won’t beat the dog’
c’. iyat=mu bhiy-i sayun NEG=1S.GEN beat-PF Sayun
‘I won’t beat Sayun; Sayun won’t be beaten by me’
Reduplication; < >: indicating the enclosed element is an infix or its gloss; =: indicating the following bound pronoun is a clitic.
d’. ini’=saku’=nya’ biq-i pila’ NEG=1S.NOM=3S.GEN give-LF money
‘He didn’t give me money; I wasn’t given money by him’ e’. ini’=mu hpuy-ani (qu’) yaba’
NEG=1S.GEN cook-BF NOM father ‘I didn’t cook for Father’
f’. ini’=mu hpuy-ani (qu’) bagah NEG=1S.GEN cook-IF NOM charcoal ‘I didn’t cook with charcoal’
Below are some of the focus markers in the language which are relavant to the present discussion:
Table 1. Some focus markers of Wulai Atayal
Sentence types Affirmative Negative
Focus types Declarative Imperative Declarative Imperative
Agent focus (AF) m-; -m- ; O O
Patient focus (PF) -un; -n- O -i
Locative focus (LF) -an -i
Instrument focus (IF) Beneficiary focus (BF)
s-; O -ani
In addition to appearing initially in a sentence, a verb may co-occur with others in the same sentence, and thus share a common actor/agent/experiencer participant. Semantically, these verbs express consecutive or simultaneous actions by the named argument, the actor/agent/experiencer. Such a construction is called ‘serial verb construction’ (SVC hereafter) where, as pointed out by Huang (1993:89-92, 1997:65-69), though the first verb may be either an AF verb or a NAF one, may be attached with some tense/aspect/mood markers (e.g. p-, -(i)n-), and may attract bound pronouns, the second verb as well as the following ones (if there are any) must appear with the AF affix m- or its variants (i.e. -m-, O), but no other marking, including tense/aspect/mood markers or bound pronouns. Some SVC examples are given below: (2) a. O-tkusa’ m-’abi (qu’) kneril qasa
AF-pretend AF-sleep NOM woman that ‘That woman is pretending to be sleeping’
b. m-wah=ku’ m-ita’ yaya’=su’ krryax AF-come=1S.NOM AF-see mother=2S.GEN every.day ‘I come to see your mother every day’
c. m<n>_gluw=sami m-aniq hira’ AF<PRF>accompany=1PE.NOM AF-eat yesterday ‘We ate together yesterday’
d. p-gluw=sami m-aniq kira’ IRR-accompany.AF=1PE.NOM AF-eat later ‘We will eat together later’
e. m-wah5=saku’ m-gluw isu’ m-aniq AF-come=1S.NOM AF-accompany 2S.NEU AF-eat ‘I will come to accompany you to eat’
(3) a. suq-un=mu m-aniq kwara’ qulih qani finish-PF=1S.GEN AF-eat all fish this ‘I will finish eating the whole fish’
b. n-wah-an=nha’ q<m>alup suruw=maku’ PRF-come-LF=3P.GEN hunt<AF>hunt after=1S.GEN ‘They came to hunt after I was through’
Note that while the first verbs in (2a-e) are all AF verbs and those in (3a-b) are NAF verbs, the verbs following them are all in AF forms (i.e. attached by m- or -m-).
As for the negative counterparts of sentences (2-3), the negators ini’ and iyat, instead of the verbs, appear sentence initially and attract the bound pronouns (if there are any). The verbs will then follow the negators and appear in their appropriate forms, and the rest of the sentences will follow the syntactic constraints observed in sentences (1a’-f’), (2a-e) and (3a-b), as presented below:
(2) a’. ini’ tkusa’ m-’abi qu’ kneril qasa NEG pretend.AF AF-sleep NOM woman that ‘That woman didn’t pretend to be sleeping’
b’. ini’=ku’ wah m-ita’ yaya’=su’ krryax NEG=1S.NOM come.AF AF-see mother=2S.GEN every.day ‘I do not come to see your mother every day’
5
While most AF verbs in Wulai Atayal use the prefix p- (instead of m- or its variants) to illustrate a future/unrealized event, the ‘come’ and ‘go’ verbs use the same AF forms, i.e. mwah and musa’. Also, note that in Wulai Atayal, the negative counterparts of sentences containing mwah and musa’ would merely have the negator iyat appearing sentence initially, without changing the forms of mwah and
c’. ini’=sami gluw m-aniq hira’ NEG=1PE.NOM accompany.AF AF-eat yesterday ‘We didn’t eat together yesterday’
d’. iyat=sami p-gluw m-aniq kira’ NEG=1PE.NOM IRR-accompany.AF AF-eat later ‘We will not eat together later’
e’. iyat=saku’ m-wah m-gluw isu’ m-aniq NEG=1S.NOM AF-come AF-accompany 2S.NEU AF-eat ‘I will not come to accompany you to eat’
(3) a’. iyat=mu suq-i m-aniq kwara’ qulih qani NEG=1S.GEN finish-PF AF-eat all fish this ‘I will not finish eating the whole fish’
b’. ini’=nha’ wah-i q<m>alup suruw=maku’ NEG=3P.GEN come-LF hunt<AF>hunt follow=1S.GEN ‘They didn’t come to hunt after I was through’
2.2Verbs and auxiliaries: musa’ and mosa’
After the discussion of some basic syntactic structures of Atayal, now consider the following sentences:
(4) a. m-usa’ pqwasan krryax laqi’ qasa AF-go school every.day child that ‘That child goes to school every day’ b. m-usa’=saku’ ngasal=nya’ misuw
AF-go=1S.NOM house=3S.GEN now ‘I am going to his house now’
c. m-usa’=ku’ mngka’ suxan AF-go=1S.NOM Taipei tomorrow ‘I will go to Taipei tomorrow’
Notice that in the above sentences, the word musa’ not only appears initially in the sentence, but, like the verbs ptucing ‘beat (AF)’, bhiyun ‘beat (PF)’, biqan ‘give (LF)’ and sphapuy ‘cook (B/IF)’ in (1b-f), it also attracts bound pronouns (e.g. saku’ in (4b) and ku’ in (4c)), which supports the claim that musa’ is the main verb of each sentence. Consequently, with the lexical meaning ‘go’, the word musa’ is legitimately classified as a motion verb. Below are the negative counterparts of sentences (4a-c), which further supports the claim that musa’ is the main verb of each sentence:
(4) a’. ini’ usa’ pqwasan krryax laqi’ qasa NEG go.AF school every.day child that ‘That child doesn’t go to school every day’ b’. ini’=saku’ usa’ ngasal=nya’ misuw
NEG=1S.NOM AF-go house=3S.GEN now ‘I am not going to his house now’
c’. iyat=ku’ m-usa’ mngka’ suxan NEG=1S.NOM AF-go Taipei tomorrow ‘I will not go to Taipei tomorrow’
However, consider the following sentences, each of which also contains the word musa’, which appears in the sentence initial position:
(5) a. musa’ O-hziq (suxan/*krryax/*misuw) IRR AF-cold tomorrow/every.day/now ‘It will be cold (tomorrow/*every day/*now)’ b. musa’ m-sbehuy
IRR AF-blow.wind ‘It will be windy’
c. musa’ m-qwalax/O-qwalax IRR AF-rain
‘It will rain’
d. musa’ m-nbu’ yaya’=mu IRR AF-ill mother=1S.GEN ‘My mother will be ill’
e. musa’ O-blaq yaya’=su’ IRR AF-good mother=2S.GEN ‘Your mother will be all right’
Note that in the above sentences, the verbs after the word musa’ (i.e. hziq ‘cold’, (m)sbehuy ‘blow wind’, (m)qwalax ‘rain’, mnbu’ ‘ill’ and blaq ‘good’) refer to natural/weather phenomena or one’s health condition, which thus prevents the initial word musa’ from indicating any physical movement. Consequently, as the English gloss indicates, the verbs following musa’ in (5a-e) designate the main meanings of these sentences, and the word musa’ has lost its concrete verbal meaning of ‘going’; it has acquired a grammatical function and can only be interpreted as an auxiliary which refers to a future/unrealized situation. The claim that the word musa’ in (5a-e) has acquired the grammatical function of indicating futurity is further supported by the possible co-occurrence of the temporal word suxan ‘tomorrow’ in (5a) but not krryax
‘every day’ or misuw ‘now’, unlike the situations found in sentences (4a-b). Moreover, consider (5a’-e’) which are the negative counterparts of (5a-e): (5) a’. musa’ iyat p6-khziq suxan
IRR NEG IRR-cold.AF tomorrow ‘It will not be cold tomorrow’
b’. musa’ iyat p-sbehuy
IRR NEG IRR-blow.wind.AF ‘The wind will not blow’
c’. musa’ iyat p-qwalax IRR NEG IRR-rain.AF ‘It will not rain’
d’. musa’ iyat p-nbu’ yaya’=mu IRR NEG IRR-ill.AF mother=1S.GEN ‘My mother will not be ill’
e’. musa’ iyat p-kblaq yaya’=su’ IRR NEG IRR-good.AF mother=2S.GEN ‘Your mother will not be all right’
Note that in (5a’-e’), the negator iyat can only follow the word musa’, instead of preceding it like in (4c’). The distribution of iyat here further supports the claim that the word musa’ is an auxiliary instead of a verb.
Now, compare the following sentences (6a-c), each of which also contains the word musa’ followed by a verb or verbs:
(6) a. musa’ m-ima’ laqi’=myan kneril=mu
MUSA’ AF-bathe child=1PE.GEN woman=1S.GEN
(i) ‘My wife will go over there to bathe our child (later, not right away)’ (ii) ‘My wife will bathe our child (later, not right away)’
b. musa’=saku’ l<m>om ngasal
MUSA’=1S.NOM burn<AF>burn house
(i) ‘I will go over there to burn the house (later, not right away)’ (ii) ‘I will burn the house (later, not right away)’
6
Atayal speakers in Wulai tend to omit an initial consonant(s) more freely; thus, hziq, sbehuy, qwalax,
c. musa’=saku’ m-gluw hiya’ m-aniq
MUSA’=1S.NOM AF-accompany 3S.NEU AF-eat
(i) ‘I will go over there to accompany him to eat (later, not right away)’ (ii)‘I will accompany him to eat (later, not right away)’
Note that sentences (6a-c) have the second verb (as well as the third one, as in (6c)) appearing in the m- form. As expected, these sentences can be analyzed as SVCs, with the word musa’ being regarded as a full-fleged verb designating a physical action ‘going’. However, for some Atayal speakers, these sentences may also designate future events ‘bathing, burning, accompanying’ without involving any physical movement of ‘going’; in other words, the word musa’ can be regarded as an auxiliary. Consequently, the status of the word musa’ in sentences like (6a-c) may be perceived differently by different speakers, and their negative counterparts can accordingly be structured differently, as shown below:
(6) a’. iyat m-usa’ m-ima’ laqi’=myan kneril=mu NEG AF-go AF-bathe child=1PE.GEN woman=1S.GEN
‘My wife will not go over there to bathe our child (later, not right away)’ a”. musa’ iyat p-ima’ laqi’=myan kneril=mu
IRR NEG IRR-bathe.AF child=1PE.GEN woman=1S.GEN ‘My wife will not bathe our child (later, not right away)’ b’. iyat=saku’ m-usa’ l<m>om ngasal
NEG=1S.NOM AF-go burn<AF>burn house
‘I will not go over there to burn the house (later, not right away)’ b”. musa’=saku’ iyat p-lom ngasal
IRR=1S.NOM NEG IRR-burn.AF house
‘I will not burn the house (later, not right away)’
c’. iyat=saku’ m-usa’ m-gluw hiya’ m-aniq NEG=1S.NOM AF-go AF-accompany 3S.NEU AF-eat
‘I will not go over there to accompany him to eat (later, not right away)’ c”. musa’=saku’ iyat p-gluw hiya’ m-aniq
IRR=1S.NOM NEG IRR-accompany.AF 3S.NEU AF-eat ‘I will not accompany him to eat (later, not right away)’
Notice that in sentences (6a’, 6b’, 6c’), the negator iyat precedes the word musa’, implying that the word musa’ is regarded as a full-fleged verb; whereas in (6a”, 6b”, 6c”), the word musa’ precedes the negator iyat indicating that it is a tense-aspect auxiliary.
(7) a. musa’=maku’ bhy-un suxan laqi’ qani IRR=1S.GEN beat-PF tomorrow child this ‘I will beat this child tomorrow’
b. musa’=maku’ lgu-n bnkis qani IRR=1S.GEN accompany-PF old.man this ‘I will accompany this old man’
c. musa’=maku’ lom-un ngasal qani IRR=1S.GEN burn-PF house this ‘I will burn this house’
Note that sentences (7a-c), unlike (6a-c), have the verb following musa’ marked with-un, the structure of which designates no properties of SVCs as illustrated by (2-3). According to my informants, no physical movement ‘going’ is involved, which can be further indicated by the proximal demonstrative qani ‘this’ marking the involved argument being close to the speaker and thus not requiring the speaker’s physical movement in order to carry out the main event (i.e. beating, accompanying, burning). In other words, (7a-c) are not SVCs, and the word musa’ here should be treated as an auxiliary instead of a regular motion verb; it designates a future/irrealis event only. Consequently, no semantic ambiguity arises like (6a-c), and their negative counterparts are formed as follows:
(7) a’. musa’=maku’ iyat bhy-un suxan laqi qani IRR=1S.GEN NEG beat-PF tomorrow child this ‘I will not beat this child tomorrow’
b’. musa’=maku’ iyat lgu-n bnkis qani IRR=1S.GEN NEG accompany-PF old.man this ‘I will not accompany this old man’
c’. musa’=maku’ iyat lom-un ngasal qani IRR=1S.GEN NEG burn-PF house this ‘I will not burn this house’
In addition to musa’, Squliq Atayal has another word mosa’, which also manifests future/unrealized events. Egerod (1999:343) distinguishes the two forms and claims that musa’ is a preverb for immediate future (‘going to … now’ with time certain) while mosa’ for indefinite future tense (‘will … some time’ with time uncertain). However, such a distinction is not recognized by the informants I have been working with in Wulai. They consider that both musa’ and mosa’ designate future events without showing semantic differences (i.e. they are free variants), and that the word
musa’ seems more often used than mosa’ in Wulai Atayal; the young speakers even do not recognize the existence of the latter.7 When comparing different dialects of Atayal, I propose that the two auxiliaries musa’ and mosa’ in Squliq Atayal must have originated from the same word, have undergone grammaticalization, and are products of different stages respectively, the situation of which will become clear as we discuss below.
Working on Mayrinax (a C’uli’ dialect), Huang (1995:148) points out that the named variant utilizes the additional presence of the vowel a in AF constructions to designate future/unrealized happenings. That is, the irrealis form of the verb musa’ ‘go’ is mausa’. Consider:
(8) Mayrinax Atayal
m-a-usa’ ’i’ yumin AF-RED-go NOM Yumin ‘Yumin will go’
It is also noted that such a sequence of a and u is often pronounced as o in Squliq Atayal, as pointed out by Li (1980:373):
(9) Squliq Atayal
a. shngon < *shnga + un ‘catch up’ b. bkon < *bka + un ‘split’
c. kton < *kita + un ‘see’
Li (1980:373) further remarks that such contraction in Squliq ‘is evident when compared with another dialect, C’uli’. Consider (Li 1980:373):
(10) Squliq C’uli’ gloss rom raum ‘needle’ smom samaum ‘wipe’
gong gaung ‘creek; stream’
It is then not surprising to find that the verb mausa’ in Mayrinax Atayal has its corresponding form mosa’ in Squliq Atayal. Furthermore, as widely observed, Wulai Squliq has a tendency of utilizing the same (and perhaps simplier) form instead of various forms. For example, the 1st person singular Nominative pronoun in Squliq
7
In his first edition of the Atayal-English dictionary, Egerod (1980:767) states that the form mosa’ is more commonly pronounced in Tao-yuan Prefecture where he collected his data.
Atayal is saku’ or ku’, and the 1st person singular Genitive pronoun is maku’ or mu. However, as Huang (1993:12) points out, ‘Some speakers of the dialect of Wulai … permit the presence of the bound Nominative pronoun ku/ when the Genitive maku’/mu is expected.’ In other words, Atayal speakers in Wulai tend to use the same and simplier pronominal form ku’ in different situations. The word musa’ may also be one such case. That is, musa’ is used in Wulai Atayal instead of mosa’, even when the latter is expected. The result is that the motion verb musa’ seems to play dual functions, manifesting both non-future (e.g. (11a-b)) and future events (e.g. (11c)): (11) a. m-usa’ ngasal=su’ tali’ krryax
AF-go house=2S.GEN Tali’ every.day ‘Tali’ goes to your house every day’ b. m-usa’ ngasal=su’ tali’ hira’
AF-go house=2S.GEN Tali’ yesterday ‘Tali’ went to your house yesterday’ c. m-usa’ ngasal=su’ tali’ suxan
AF-go house=1S.GEN Tali’ tomorrow ‘Tali’ will go to your house tomorrow’
The above neutralization in Squliq Atayal seems to confirm what Hopper and Traugott (1993:1) observe in English; that is, ‘… the auxiliary which expresses immediate futurity derives historically from the motion verb go in a highly specific context, and that the two coexistent forms are polysemous. Such meaning-form correlations are found in a wide number of languages around the world.’
To sum up the present discussion on the word musa’, I notice that the Atayal words mausa’, mosa’ and musa’ are rather similar to, though not exactly identical with, the English phrase be going to/be gonna (cf. Hopper and Traugott 1993:3).8 That is,
(i) mausa’, the original form indicating the future/irrealis form of the verb ‘go’, has become mosa’ through the phonological contraction of a and u;
(ii) mosa’ is then substituted by the original verbal form musa’;
(iii) musa’ is further undergoing the loss of its original relatively concrete meaning of ‘going’, specifically motion and directionality, and is in the process of grammaticalization into an auxiliary functioning as a tense/aspect/mood marker.
8
Hopper and Traugott (1993:3), while examining the named expression, remark that ‘… be going to can undergo changes typical of auxiliaries, such as phonological reduction’. That is, the English phrase be going to is phonologically reduced into be gonna.
The above three stages can be diagrammed as follows:
Stage: I II III
•
Concrete Abstract
Meaning/function: ‘will go’ ‘will go’ Irrealis marker
Part of speech: verb verb auxiliary
Form: m-a-usa’ mosa’ musa’
Process: vowel reduplication vowels contraction grammaticalization
Figure 1. Path of grammaticalization of musa’ in Squliq Atayal
The following diagram may also explain the neutralization between the motion verb musa’ and the auxiliary musa’ in Wulai Atayal:
(12) a. m-a-usa’ (verb + reduplicated a) mosa’ (contraction) musa’ (auxiliary) b. musa’ (verb)
If the above analysis of the evolvement of the auxiliary musa’ is plausible, there is one more thing that deserves our attention before we move on to the next section. That is, note that mosa’ is phonologically reduced (or contracted) from the irrealis/future verb mausa’ and so its signaling futurity is more apparent. On the other hand, since the grammaticalized auxiliary musa’ exactly resembles the non-future motion verb musa’, there seems less futurity signaled when it is compared with the auxiliary mosa’; ambiguity sometimes thus arises. This may explain why Egerod (1999:343) treats mosa’ as a preverb for indefinite future tense and musa’ for immediate future.
2.3Verbs and auxiliaries: wal and wayal
The verb wal or wayal in Wulai Atayal may function as the main verb of a sentence, and designate a perfective/actualized event ‘go’, whether appearing as the only verb (e.g. (13a-f)) or co-occurring with another one (e.g. (13g)):
(13) a. wal=saku’ mngka’ go.AF.PRF=1S.NOM Taipei ‘I went to Taipei’
b. wal ngasal=nya’ watan go.AF.PRF house=3S.GEN Watan ‘Watan went to his house’
c. wal=su’ inu’ la ? go.AF.PRF=2S.NOM where PART ‘Where did you go?’
d. wal inu’ qu’ bnkis qasa la ? go.AF.PRF where NOM old.man that PART ‘Where did that old man go?’
e. wayal inu’ laqi’=su’ la go.AF.PRF where child=2S.GEN PART ‘Where did your child go?’
f. wayal te pzi’an mali go.AF.PRF LOC playground ball ‘He went to the ball ground’
g. wal O-kangi qnxan=nha’ te’ i tranan rwa (Egerod 1999:315) go.AF.PRF AF-seek place=3P.GEN LOC Tranan PART
‘They went to seek a place to live in the direction of Tranan’
Like musa’ and mosa’, wal and wayal may also function as auxiliaries9 (to be exact, tense/aspect/mood markers), except that they indicate perfectivity instead of futurity designated by musa’/mosa’. The new function of wal and wayal, i.e. indicating perfectivity, must have derived from their original lexical semantics of indicating an actualized physical motion ‘went’. In other words, similar to musa’ and mosa’, wal and wayal must also have undergone a process of grammaticalization and may now serve as both auxiliaries and verbs, though as we have observed, they function as perfective markers more often than serving as plain verbs.
Below are some examples with wal and wayal serving as auxiliaries, without involving any physical movement at all:
(14) a. wal m-qwalax hira’ PRF AF-rain yesterday ‘It rained yesterday’
9
According to Egerod (1999:349), only wal (but not wayal) can serve as a preverb (i.e. auxiliary in our term) indicating past tense, which is different from my observation during the fieldwork though it is true that wal tends to be used more often than wayal.
b. wal m-nbu’ yaba’=mu PRF AF-ill father=1S.GEN ‘My father was ill’
c. wal m-hoqil yaba’=nya’ PRF AF-die father=3S.GEN ‘His father died’
d. wayal m-qas balay qu’ mlikuy qasa PRF AF-happy very NOM man that ‘That man was very happy’
Functioning as auxiliaries, the words wal and wayal thus allow either an AF or NAF verb to follow, as shown in (15a-d) and (16a-g), respectively, the property of which resembles that of musa’ as discussed in Section 2.2 (cf. sentences (6a-c) and (7a-c)): (15) a. wal=saku’ m-lukus lukus talah
PRF=1S.NOM AF-put.on clothes red ‘I have put on red clothes’
b. wal m-aziy mngka’ ngasal yaba’=mu la PRF AF-buy Taipei house father=1S.GEN PART ‘My father has bought a house in Taipei (already)’ c. wal=su’ m-aras cimu’ ru’ puy-un=ta’
PRF=2S.NOM AF-bring salt and cook-PF=1PI.GEN ayang qulih kira’ qani ga’
soup fish later this Q
‘Have you brought the salt which we will use in cooking this fish soup later?’ d. wayal muci k<m>al qu’ mrhuw mlikuy qasa
PRF thus say<AF>say NOM chief male that ‘That chief has thus said’
(16) a. wal=nya’ gal-un turuy=su’ la’ PRF=3S.GEN take-PF car=2S.GEN PART ‘He has taken your car (already)’
b. wal=nya’ gngi-an ke’=maku’ PRF=3S.GEN forget-PF word=1S.GEN ‘He has forgotten my words (what I said)’
c. pira’ pila’ wal=su’ bir-an bzyoq qasa how.much money PRF=2S.GEN buy-PF pig that ‘How much is the pig you bought?’
d. wal=nya’ s-kut temu’ lalaw qani PRF=3S.GEN IF-kill Temu’ aboriginal.knife this ‘He used this knife to kill Temu’ ’
e. wayal niq-un nqu’ bnkis qu’ ngahi PRF eat-PF GEN old.man NOM yam ‘The yam has been eaten by the old man’ f. wayal htag-un qu’ biru’ qasa mha
PRF lose-PF NOM book that EVI ‘It is said that the books have been lost’
g. wayal=nya’ swal-an qu’ laqi’ na’ kneril PRF=3S.GEN promise-PF NOM child LIN woman ‘He promised to the girl (that he would marry her)’
There is another piece of evidence supporting the claim that wal and wayal may function as tense-aspect auxiliaries. Consider the following negative constructions:10
(17) a. wal ini’ qwalax hira’ PRF NEG rain.AF yesterday ‘It didn’t rain yesterday’
b. wal=saku’ ini’ pkucu’ iqas kucu’ PRF=1S.NOM NEG put.on.shoes.AF new shoe ‘I still haven’t worn new shoes’
c. wal=ku’ ini’ qaniq ciwal ryax PRF=1S.NOM NEG eat.AF three day ‘I haven’t eaten for three days’
Note here that the Negator ini’ follows the word wal, instead of preceding it, which indicates that the word wal does not function as a regular verb.
To sum up the present discussion, I observed that the two words wal and wayal are polysemous morphemes, with their functioning as perfective tense/aspect markers more prominent than their serving as plain verbs. In other words, the process of grammaticalization of wal and wayal seems to be more completed than that of musa’/mosa’. Finally, the grammaticalization of wal/wayal can be illustrated in Figure 2:
10
Taoshan Atayal seems to have a similar construction, as reported by Mei-chuan Su (2004:81): wayal ini’ swayal (qu’) tali’ shira’ lrwa, laxi qhl-iy la ! PRF NEG willing.AF NOM Tali’ yesterday PART NEG force-LF PART ‘Tali’ was not willing yesterday already. Don’t force him!’
Stage: I II
•
Concrete Abstract
Meaning/function: ‘went’ Perfective marker
Part of speech: verb auxiliary
Form: wal/wayal wal/wayal
Process: grammaticalization
Figure 2. Path of grammaticalization of wal/wayal in Squliq Atayal
2.4Verbs and auxiliaries: nyux and cyux
Similar to the above-mentioned polysemous words musa’/mosa’ and wal/wayal, nyux and cyux may serve both as regular verbs and as auxiliaries. First, consider the following examples in which nyux and cyux function as main verbs and designate what English speakers may interpret as ‘existence’ (e.g. (18a-a’)), ‘possession’ (e.g (18b-b’)) and ‘location’ (e.g. (18c-c’)):11
(18) a. nyux qutux qhuniq lingay ngasal=mu exist.PROX one tree side house=1S.GEN ‘There is a tree by my house’
a’.cyux biru’ squ’ nniqan exist.DIST book LOC table
‘There is a book on the table (away from here)’ b. nyux pila’=mu
exist.PROX money=1S.GEN ‘I have money’
b’.cyux pila’=nha’ exist.DIST money=3P.GEN ‘They have money’
c. nyux ngasal yaba’=mu exist.PROX house father=1S.GEN ‘My father is home’
c’.cyux ngasal yaba’=su’ exist.DIST house father=2S.GEN
‘Is your father home?’ (with rising intonation)
As the English gloss indicates above, the difference between the two verbs nyux and
11
The three constructions mentioned here are related in almost all the Formosan languages. Please refer to Zeitoun et al. (1999) for a more detailed discussion.
cyux is that the former indicates the ‘existence, possession, location’ of the involved argument is closer to the speaker, spatially and/or temporally, whereas the latter away from the speaker. Also, while serving as main verbs, nyux and cyux are among the very few verbs in the language that have no corresponding NAF forms.
Like musa’/mosa’ and wal/wayal, nyux and cyux may serve as tense/aspect auxiliaries as well. One may expect such a development is because of the inherited semantics of designating ‘existence’ and ‘location’ of nyux and cyux, i.e. ‘exist, be at’ > ‘be ... Ving’, which seems to be a rather common practice in the languages of the world (e.g. za$i in Mandarin Chinese as discussed in Huang 1988:249-310).
Now examine the following sentences where nyux and cyux designate on-going actions:
(19) a. nyux=sami m-lukus IMPRF.PROX=1PE.NOM AF-put.on
‘We are putting on clothes (here and now)’ b. nyux m-aniq tbaku tali’
IMPRF.PROG AF-eat cigarette Tali’ ‘Tali’ is smoking (here)’
(20) a. cyux m-aniq tbaku tali’ IMPRF.DIST AF-eat cigarette Tali’ ‘Tali’ is smoking (there)’
b. cyux m-hnaw turi=nya’ IMPRF.DIST AF-rest car=3S.GEN ‘His car is resting (there)’
Notice that, as the English gloss indicates above, the difference between the two auxiliaries nyux and cyux is that the former indicates an action taking place close to the speaker, temporally and/or spatially, and hence is treated as the proximal imperfective marker (e.g. (19a-b)), and the latter away from the speaker and thus the distal imperfective marker (e.g. (20a-b)). Such difference must have derived from their original semantic difference as mentioned above.
Another piece of evidence that supports the two elements nyux and cyux in similar contexts to be auxiliaries instead of existential/possessive/locative verbs is that, like musa’/mosa’ and wal/wayal discussed in Section 2.2 and Section 2.3, the two words nyux and cyux can be followed by an -un marked verb, which indicates they are not ordinary verbs but auxiliaries:
(19) a’.nyux=myan lukus-un lukus=nha’ IMPRF.PROX=1PE.GEN put.on-PF dress=3P.GEN
‘We are putting on their clothes; their clothes are being put on by us’ b’.nyux niq-un tali’ tbaku
IMPRF.PROX eat-PF Tali’ cigarette
‘The cigarette is being smoked by Tali’ (here)’ (20) a’.cyux niq-un tali’ tbaku
IMPRF.DIST eat-PF Tali’ cigarette
‘The cigarette is being smoked by Tali’ (there)’
c. cyux=nya’ yut-un puniq misuw qani IMPRF.DIST=3S.GEN extinguish-PF fire right.now ‘He is putting out the fire now’
Furthermore, while the negative counterparts of existential/possessive/locative constructions utilize the verb ungat (as pointed out in Huang 1993:74), those of sentences with nyux and cyux serving as tense-aspect markers pattern differently, as exemplied below:12
(21) a. nyux=ku’ ini’ qaniq na (Huang 1993:83) IMPRF.PROX=1S.NOM NEG eat.AF still
‘I am still not eating yet’
b. nyux ini’ qaniq tbaku tali’ IMPRF.PROX NEG eat.AF cigarette Tali’ ‘Tali’ is not smoking (here)’
c. nyux=maku’ ini’ baq-i kwara’ qaya’ na cinun=nha’ IMPRF.PROX=1S.GEN NEG know-LF all thing LIN weave=3P.GEN ga’, … (Egerod 1991:44)
TOP
‘I have no idea about all the things used in the process of weaving’ d. cyux ini’ huqil bnkis qasa
IMPRF.DIST NEG die.AF old.man that ‘That old man is not dead yet’
12
Similar structures also appear in Taoshan Atayal, as reported by Mei-chuan Su (2004:81): (a) nyux ini’ swayal (qu’) payas lruw, swa’=su’ qhl-un lpi ?
PROG NEG willing.AF NOM Payas PART why=2S.GEN force-PF PART ‘Payas is not willing now; why should you force him?’
(b) cyux ini’ tuliq payas na’ PROG NEG get.up.AF Payas still ‘Payas hasn’t got up yet’
e. cyux ini’ plukus sayun na IMPRF.DIST NEG put.on.AF Sayun still ‘Sayun hasn’t put on clothes yet’
To conclude, like musa’/mosa’ and wal/wayal, nyux and cyux are also polysemous. Sometimes they serve as verbs indicating ‘existence’, ‘possession’ and ‘location’; and sometimes they are auxiliaries functioning as imperfective/progressive aspectual markers. Concerning the grammaticalization of nyux and cyux, i.e. from plain verbs to auxiliaries, the following figure illustrates the above evolement:
Stage: I II
•
Concrete Abstract
Meaning/function: ‘exist/possess/locate’ Imperfective marker
Part of speech: verb auxiliary
Form: nyux/cyux nyux/cyux
Process: grammaticalization
Figure 3. Path of grammaticalization of nyux/cyux in Squliq Atayal
3. Verb and negator: laxi
This section will examine the word laxi, a variant of the Atayal verb malax ‘give up; take off’ that has been grammaticalized and may function as a negator. First, consider the following examples with the word malax ‘give up; take off’ occurring in the sentence initial position and serving as the main verb:
(22) a. m-alax=ku’ tbaku ru’ qwaw la AF-give.up=1S.NOM cigarette and wine PART ‘I have given up cigarette and wine already’
b. m-alax lukus krryax hiya’ AF-give.up clothes every.day 3S.NEU ‘He throws away clothes often’
c. m<n>alax kneril=nya’ likuy qasa AF<PRF>give.up woman=3S.GEN man that ‘That man has abandoned his wife’
d. m-alax m-aniq tbaku hiya’ AF-give.up AF-eat cigarette 3S.NEU ‘He has given up smoking’
Note here that the verb malax may precede a Patient argument (e.g. (22a-c)), or precede another verb and form a SVC (e.g. (22d)).
In addition to the AF form malax, the verb has some other forms like alax, laxun, laxan, s’alax and laxi, as given in the following examples:
(23) a. alax lukus=su’ take.off.AF clothes=2S.GEN ‘Take off your clothes!’
b. ini’ alax lukus krryax hiya’ NEG give.up.AF clothes every.day 3S.NEU ‘He doesn’t throw away clothes often’
c. lax-un=su’ inu’ suxan huzil qasa give.up-PF=2S.GEN where tomorrow dog that ‘Where will you abandon that dog tomorrow?’
d. wal=nya’ lax-an turi hupaw=nya’ PRF=3S.GEN leave-LF car purse=3S.GEN ‘He left his purse in the car (on purpose)’ e. wal=nya’ s-’alax turi hupaw=nya’
PRF=3S.GEN IF-leave car purse=3S.GEN ‘He left his purse in the car (by accident)’
f. ini’=nha’ lax-i qu’ qwaw hiya’ NEG=3P.GEN give.up-LF NOM wine EMP ‘They cannot refuse the wine’
g. lax-i lukus qani give.up-LF clothes this ‘Abandon the clothes!’
Among the above-named verbal forms, laxi is an LF verb used in imperatives, meaning ‘to give up something’. This verb seems to have undergone the process of grammaticalization, and now can be used as a negator in imperative sentences. Consider the following sentences:
(24) a. laxi qaniq tbaku sqani NEG eat.AF cigarette here ‘Don’t smoke here!’
a’.laxi niq-i mami qani NEG eat-LF rice this ‘Don’t eat this rice!’
b. laxi kita’ tali’ NEG look.AF Tali’ ‘Don’t look at Tali’!’ b’.laxi kta-y tali’
NEG look-LF Tali’ ‘Don’t look at Tali’!’ c. laxi bahuq lukus
NEG wash.AF clothes ‘Don’t wash clothes!’
c’.laxi hbaq-i lukus=mu hiya’. hbaq-un=mu nanaq NEG wash-LF clothes=1S.GEN EMP wash-PF=1S.GEN self ‘Don’t wash my clothes! I’ll wash (them) myself.’
Recall the discussion of SVCs in Section 2.1; that is, if two or more verbs cooccur, the 2nd (and also the following ones, if any) verb(s) need to be in the AF form. However, notice in the above that following laxi, the verb can be either in its AF form (e.g. qaniq, kita’ and bahuq, but note that they are not affixed with affirmative declarative AF m-/-m-/O) or may contain the LF marker -i (e.g. niq-i, kta-i and hbaq-i). Thus, laxi here cannot be an ordinary verb; it is a negator.
The reason to claim that the verbal form laxi has been grammaticalized and may serve as a negator will be further supported when the following Mayrinax sentences are examined (Huang 1995:168-169):
(25) Mayrinax Atayal
a. m-alax=ci’ cu’ quwaw AF-give.up=1S.NOM ACC.RF wine ‘I don’t want wine’
b. m-alax=ci’ cu’ xuil AF-give.up=1S.NOM ACC.RF dog ‘I don’t want a dog’
c. lax-i ku’ quwaw give.up-LF NOM.RF wine
‘Give up the wine!’ (Don’t want the wine!) d. lax-i ku’ xuil
give.up-LF NOM.RF dog
‘Give up the dog!’ (Don’t want the dog!)
As the above examples illustrate, Maryinax Atayal has the AF verb malax ‘give up’ in affirmative declarative sentences, and its LF form laxi in imperative constructions.
Note that (i) the noun expressing the Patient argument, when following malax, is marked by the Accusative case marker cu’ (e.g. (25a-b)), and (ii) the noun manifesting the same argument, when following laxi, is preceded with the Nominative case marker ku’ (e.g. (25c-d)). Furthermore, with its negative meaning, laxi has gradually developed into one of the two imperative negators in Mayrinax, as exemplified in (26-27):
(26) Mayrinax Atayal (Huang 1995:169)
a. laxi ku’ m-nubuwag cu’ quwaw NEG NOM.RF AF-drink ACC.NRF wine ‘Don’t drink wine!’
b. laxi ku’ m-usa’ i’ uray NEG NOM.RF AF-go PREP Wulai ‘Don’t go to Wulai!’
c. laxi ku’ ma-tuting NEG NOM.RF REC-beat
‘Don’t fight against each other!’ (27) Mayrinax Atayal (Huang 1995:167)
a. kaa nubuwag cu’ quwaw NEG drink.AF ACC.NRF wine ‘Don’t drink wine!’
b. kaa usa’ i’ uray NEG go.AF PREP Wulai ‘Don’t go to Wulai!’
c. kaa tuting-i ’i’ ba’ay NEG beat-LF NOM Ba’ay ‘Don’t beat Ba’ay!’
While Mayrinax Atayal has two imperative negators kaa and laxi, they seem to merge into one in Wulai Atayal; that is, utilizing the form of laxi but following the syntactic behaviors of kaa. In other words, while only Nominative nouns or noun phrases may follow laxi in Mayrinax, either AF root verbs or LF -i suffixed verbs may follow laxi in Wulai Atayal, which resembles Mayrinax kaa, as illustrated in (24a-c) and (27a-c).
To sum up the present discussion, I conclude that laxi in Wulai Atayal plays dual role and is polysemous: (i) serving as an LF verb in negative declarative construction (e.g. (23e)) meaning ‘give up; refuse’, and (ii) functioning as an imperative negator with less concrete meaning after undergoing the process of grammaticalization (e.g. (24a-c)). The following figure illustrates the evolvement of laxi from a full-fledged LF verb to an imperative negator:
Stage: I II
•
Concrete Abstract
Meaning/function: ‘give up; refuse’ Imperative negator
Part of speech: verb auxiliary
Form: lax-i laxi
Process: grammaticalization
Figure 4. Path of grammaticalization of laxi in Squliq Atayal
4. Verb and complementizer/evidential marker: mha
In quite a few languages around the world, e.g. Chinese, Russian, Japanese (cf. Hwang, Lyovin and Baika 1998), it is often found that a saying verb may be grammaticalized and serve as a complementizer or even as a particle (e.g. shuo# in Mandarin Chinese; kong in Southern Min; molvil in Russian; yuu in Japanese). In Squliq Atayal, the verb mha meaning ‘say’ seems to work in a similar way. First, consider the following sentences with the named word appearing in the sentence/clause initial position and functioning as the main verb (e.g. (28a-e)), just like some other saying verbs (e.g. kmal ‘say’, maqut ‘ask’, and smyuk ‘answer’) as in (28f) and (29a-c):13
(28) a. mha ke’=nya’ (Egerod 1999:156) say.AF word=3S.GEN
‘… so the story goes’
b. mha nanu’ knita’=su’ zyaw qani ? say.AF what viewpoint=2S.GEN thing this ‘What’s your viewpoint of this thing?’
Lit. ‘What does your view of this thing say?’ c. mha nanu’ pinkyalan=mamu la’ ?
say.AF what discussion=2P.GEN PART ‘What is the result of your discussion?’
Lit. ‘What does the result of your discussion say?’
13
In my fieldnotes, there are not many examples with mha functioning as the main verb. Here I would like to thank Mr. Tali’ Hayung for his providing me with, or pointing out to me, some of the following examples containing the word mha, both as a plain verb and as a complementizer.
d. mha nanu’ qu’ t’ringan na matas hiya’ ? (Yuqih & Yupas 1991: 87) say.AF what NOM beginning GEN tattoo EMP
‘What is the original meaning of matas?’
e. m-qas balay sayun ru’ mha {m-huway isu’ balay, temu’} AF-happy very Sayun and say.AF AF-thank 2S.NEU very Temu’ ‘Sayun was very happy and said, “Thank you very much, Temu’!”’
f. pqut-an nqu’ mqu qu’ kneril mha {wal=simu s-htuy balay ni ask-LF GEN snake NOM woman mha PRF=2P.NOM IF-deter truly GEN utux kayal mha {laxi qaniq bway na ana nanu qhuniq!}
God say.AF NEG.IMP eat.AF fruit GEN any tree (The Bible Society in Taiwan 2003: Inringan 03:2-3)
‘The snake asked the girl, “Has the God forbidden you and said, “Don’t eat any fruits!”? ’
(29) a. k<m>al=sami ke’=myan nanak na’ say<AF>say=1PE.NOM language=1PE.GEN self still ‘We still speak our own language’
b. m-aqut kuzing krryax tali’ ru’ syuk-un=mu kwara’ AF-ask 1S.NEU every.day Tali’ and answer-PF=1S.GEN all ‘Tali’ often asks me (questions) and I always respond them all’
c. s’ang-un=saku’=nya’ ru’ s<m>yuk=saku’ uzi scold-PF=1S.NOM=3S.GEN and answer<AF>answer=1S.NOM too ‘He scolds me and I respond too’
Notice that the verb mha, like the existential/possessive/locative verbs nyux and cyux discussed in Section 2.4, does not seem to have any corresponding NAF forms, whereas other saying verbs do; that is, kmal ‘say.AF’ vs. kyalun ‘say.PF’, maqut ‘ask.AF’ vs. pqutun ‘ask.PF’, and smyuk ‘answer.AF’ vs. syukun ‘answer.PF’.
Moreover, the word mha may follow any of the above-mentioned saying verbs, with the content being said appearing as a direct quote. Examples follow (the direct quote appears inside brackets):
(30) a. k<m>al yaya’=mu kuzing mha {hlhul lukus=su’ qani?} say<AF>say mother=1S.GEN 1S.NEU MHA warm.AF clothes=2S.GEN this ‘My mother said to me, “Are your clothes warm?” ’
b. yungay csuru m-ita’ squ’ yaya’=nya’ k<m>ayal mha monkey turn.head.AF AF-see LOC mother=3S.GEN say<AF>say MHA
{kuzing ga’ m-qelang balay na squliq. laxi=saku’ lngis-i} 1S.NEU TOP AF-lazy very GEN person NEG=1S.NOM cry-LF.IMP (Wang 2007:23)
‘The monkey looked back at his mother and said, “I am a very lazy person. Don’t feel sympathetic for me!” ’
c. m-aqut kuzing mha {m-usa’=su’ ngasal tali’ kira’} AF-ask 1S.NEU MHA AF-go=2S.NOM house Tali’ later ‘He asked me (saying), “Are you going to Tali’s place later?” ’ d. s<m>yuk laqi’=mu mha {yat=ku’ qaniq}
answer<AF>answer child=1S.GEN MHA NEG=1S.NOM eat.AF ‘My child answered (saying), “I will not eat!” ’
(31) a. kyal-un=mu mha {h<m>swa’=su’ la’ ? swa’ nyux say-PF=1S.GEN MHA why<AF>why=2S.NOM PART why IMPRF.PROX m-huziq lukus=su’ ru’ m-’yabux kwara’ qu’ rqyas=su’ ?} AF-wet clothes=2S.GEN and AF-sweat all NOM face=2S.GEN ‘I told him (saying), “What has happened to you? Why are your clothes wet? And why is your face sweating?” ’
b. kyal-i sayun mha {nyux=ku’ m-ahuq lukus, tell-LF.IMP Sayun MHA IMPRF.PROX=1S.NOM AF-wash clothes nanu’ yasa’ yat=ku’ kita’ hiya’}
therefore NEG=1S.NOM see.AF 3S.NEU
‘Tell Sayun (saying), “I am washing clothes, so I can’t see her!” ’ c. pqut-an=mu hiya’ mha {m<n>aniq=su’ inu’ hira’}
ask-LF=1S.GEN 3S.NEU MHA AF<PRF>eat=2S.NOM where yesterday ‘I asked him, “Where did you eat yesterday?” ’
As illustrated above, the saying verbs in the sentence initial position (the saying verb in (30b) does not occur in the sentence intial position as calimed by the author) can be in either AF form (e.g. kmal ‘say.AF’ in (30a), kmayal ‘say.AF’ in (30b), maqut ‘ask.AF’ in (30c), smyuk ‘answer’ in (30d)) or NAF one (e.g. kyalun/kyali ‘say.NAF’ in (31a-b), pqutan ‘ask.NAF’ in (31c)), with the word mha following the matrix verb (i.e. saying verbs) and the shared Agent argument, the structure of which resembles SVCs as discussed in Section 2.1; in other words, mha serves as the second verb. However, since the matrix verb is already a saying verb that presents the sense of ‘saying; asking; answering’, the presence of mha here is not semantically significant but somewhat redundant, and thus can be left out without producing any semantic differences, as remarked by my informant:
(30) a’. k<m>al yaya’=mu kuzing O {hlhul lukus=su’ qani?} say<AF>say mother=1S.GEN 1S.NEU warm.AF clothes=2S.GEN this ‘My mother said to me, “Are your clothes warm?” ’
b’. yungay csuru m-ita’ squ’ yaya’=nya’ k<m>ayal O monkey turn.head.AF AF-see LOC mother=3S.GEN say<AF>say {kuzing ga’ m-qelang balay na squliq. laxi=saku’ lngis-i}
1S.NEU TOP AF-lazy very GEN person NEG=1S.NOM cry-LF.IMP ‘The monkey looked back at his mother and said, “I am a very lazy person. Don’t feel sympathetic for me!” ’
c’. m-aqut kuzing O {m-usa’=su’ ngasal tali’ kira’} AF-ask 1S.NEU AF-go=2S.NOM house Tali’ later ‘He asked me (saying), “Are you going to Tali’s place later?” ’ d’. s<m>yuk laqi’=mu O {yat=ku’ qaniq}
answer<AF>answer child=1S.GEN NEG=1S.NOM eat.AF ‘My child answered (saying), “I will not eat!” ’
(31) a’. kyal-un=mu O {h<m>swa’=su’ la’ ? swa’ nyux say-PF=1S.GEN why<AF>why=2S.NOM PART why IMPRF.PROX m-huziq lukus=su’ ru’ m-’yabux kwara’ qu’ rqyas=su’ ?} AF-wet clothes=2S.GEN and AF-sweat all NOM face=2S.GEN ‘I told him (saying), “What has happened to you? Why are your clothes wet? And why is your face sweating?” ’
b’. kyal-i sayun O {nyux=ku’ m-ahuq lukus, tell-LF.IMP Sayun IMPRF.PROX=1S.NOM AF-wash clothes nanu’ yasa’ yat=ku’ kita’ hiya’}
therefore NEG=1S.NOM see.AF 3S.NEU
‘Tell Sayun (saying), “I am washing clothes, so I can’t see her!” ’ c’. pqut-an=mu hiya’ O {m<n>aniq=su’ inu’ hira’}
ask-LF=1S.GEN 3S.NEU AF<PRF>eat=2S.NOM where yesterday ‘I asked him, “Where did you eat yesterday?” ’
Consequently, it also seems plausible to postulate that the word mha in such a structure has lost its verbal meaning and begun to function more like a quotative marker, introducing a direct-quote that serves as the object to the matrix verb. To sum up the present discussion, the word mha when co-appearing with saying verbs can be either treated as the second verb meaning ‘say’, or as a reportative marker or a linker which is semantically less concrete.
Similarly, verbs like ‘sing, teach, write’, though not exactly saying/utterance verbs, may co-occur with the word mha in a sentence, where mha again seems to serve more like a quotative marker, introducing the content being ‘sung, taught, written’, as
illustrated below:
(32) a. cyux m-qwas mha {t<m>inun! t<m>inun!
IMPRF.DIST AF-sing MHA weave<AF>weave weave<AF>weave t<m>inun lukus. ...}
weave<AF>weave clothes ..
‘(She) was singing, “Weave, weave, weave clothes …!” ’
b. s-pqwas=saku’=nha’ mha {laqi! ana=su’ m-usa’ inu’, BF-sing=1S.NOM=3P.GEN MHA child no.matter=2S.NOM AF-go where Laxi zng-i qu’ ke=ta’ ita’ tayal}
NEG forget-LF NOM word=1PI.GEN 1PI.NEU Atayal
‘They will sing for me, ‘Kids! Wherever you go, don’t forget our Atayal language.”
c. p<in>qzyw-an=saku’ ni mama=maku’ mha {qinxan=ta’ teach<PRF>teach-LF=1S.NOM GEN uncle=1S.GEN MHA life=1PI.GEN sraral ga’, ungat balay turuy glg-an=ta’. …}
before TOP not.exist really car ride-LF=1PI.GEN …
‘My uncle taught me, “In our past life, there was no car for us to ride. …”’ d. psbzinah m-usa’ m-ita’ squ’ p<in>qbaq-an=ta’ te
turn.back.AF AF-go AF-see LOC learn<PRF>-learnLF=1PI.GEN LOC lpgan ru’, blq-i s<m>yuk qu’ nyux s-paqut text and do.well-LF.IMP answer<AF>answer NOM IMPRF.PROX IF-ask zik hugal qani. kia qu’ ini’ ptkari ga’, ptas-i mha “O”; bottom below this if NOM NEG get.lost.AFTOP write-LF.IMP MHA “O” kia qu’ m-tkari ga’, ptas-i mha “X”. (Yuqih and Yupas 1991:75) if NOM AF-get.lost TOP write-LF.IMP MHA “X”
‘Recall and review what has been learned in the textbook, and answer properly the following questions. If they are correct, write O; if they are wrong, write X.’
Note that these semi-utterance verbs may also precede direct quotes, and that the elements inside brackets can either be full clauses/discourse (e.g. (32a-c)) or lexical words (e.g. (32d)). Syntactically, these clauses or words serve as the object to the matrix verbs.
Besides serving as the main predicative verb (e.g. (28a-e)) or following an utterance/semi-utterance verb (e.g. (30-32)), the word mha may co-occur with a cognition verb (e.g. know, think in English). Consider:
(33) a. baq-un tali’ mha {nyux=saku’ m-nbu’} know-PF Tali’ MHA IMPRF.PROX=1S.NOM AF-ill ‘Tali’ knows that he is ill; Lit. Tali’ knows, “I am ill” ’
b. m-nbu’ qu’ yaki nqu’ silan hayung. m-nglung qu’ AF-ill NOM grandmother GEN Silan Hayung AF-think NOM silan hayung mha {aki=ta’ s-’usa’ pnep qulih qu’ Silan Hayung mha should=1PI.GEN BF-go fish.AF fish NOM yaki ha.} (Yuqih and Yupas 1991:47)
grandmother PART
‘Silan Hayung’s grandma was ill. Silan Hayung was thinking, “We should go fishing for Grandma.”
As observed above, sentences (33a-b) containing the cognition verbs (i.e. baqun ‘know.PF’ and mnglung ‘think.AF’) and the word mha resemble sentences (30-31), though mha does not seem to present any sense of ‘saying’ but functions to link the matrix verb and its object which surfaces as a direct quote in (33a-b).
However, such an embedded object does not have to be a direct quote construction, as shown below:
(34) a. m-aki’ squ’ kska na hlahuy qu’ losing watan qasa ga’, AF-live LOC middle GEN forest NOM Losing Watan that TOP ini’ kngungu’ ana cikuy. m-uci m-’abi nanak. baq-un=nya’ NEG fear.AF even little AF-such AF-sleep alone know-PF=3S.GEN mha {musa’ h<m>swa’ m-qyanux m-aki’ qsahuy na hlahuy.}
MHA IRR how<AF>how AF-live AF-live inner GEN forest (Yuqih and Yupas 1991:53)
‘When Losing Watan was in the forest, he was not afraid at all. He was sleeping alone. He knew how he would live inside the forest.’
Note that in order to be a direct quote, the elements inside the brackets need to have the pronoun ku’ ‘I; 1S.NOM’ attached to the initial auxiliary musa’ (i.e. musa’=ku’ hmswa’ mqyanux maki’ qsahuy na hlahuy ‘How will I live inside the forest?’); without the presence of the pronoun, the named elements do not form a direct quote. Thus, the word mha in (34a) seems to be better analyzed as a complementizer. Furthermore, like the English complementizer that, this complementizer as well as the quotative marker discussed above may be left out from the named constructions without producing any semantic differences, as remarked by my informant:
(33) a’.baq-un tali’ O {nyux=saku’ m-nbu’} know-PF Tali’ IMPRF.PROX=1S.NOM AF-ill ‘Tali’ knows that he is ill; Lit. Tali’ knows, “I am ill” ’
(34) a’.m-aki’ squ’ kska na hlahuy qu’ Losing Watan qasa ga’, AF-live LOC middle GEN forest NOM Losing Watan that TOP ini’ kngungu’ ana cikuy. m-uci m-’abi nanak. baq-un=nya’ NEG fear.AF even little AF-such AF-sleep alone know-PF=3S.GEN
O {musa’ h<m>swa’ m-qyanux m-aki’ qsahuy na hlahuy.}
IRR how<AF>how AF-live AF-live inner GEN forest
‘When Losing Watan was in the forest, he was not afraid at all. He was sleeping alone. He knew how he would live inside the forest.’
In addition to preceding a direct quote structure, the word mha, when serving as the matrix verb, may follow the named clause as well. Consider:
(35) a. {hbaq-i nanaq} mha yaya’=mu wash-LF.IMP self say.AF mother=1S.GEN ‘“Wash (them) yourself,” said my mother.’ b. {biq-i=saku’ pila’} mha tali’
give-LF.IMP=1S.NOM money say.AF Tali’ ‘“Give me the money,” said Tali’.’
c. {musa’=su’=nya’ s-laquh} mha tali’ IRR=2S.NOM=3S.GEN BF-win say.AF Tali’ ‘“He will beat (win over) you,” said Tali’.’
d. {hilaw=saku’ m-wah} mha hiya’ ru’, ana ga’ fast.AF=1S.NOM AF-come say.AF 3S.NEU and but ini’ khilaw m-wah
NEG fast.AF AF-come
‘“I’ll come back soon,’ he said, but he didn’t come back soon’
Similarly, when co-occurring with an utterance verb or a cognition verb, the word mha may also follow the direct quote clause. For example:
(30) a’. k<m>al yaya’=mu kuzing {O-hlhul lukus=su’ qani?} mha say<AF>say mother=1S.GEN 1S.NEU AF-warm clothes=2S.GEN this MHA
