3.3.1 Introduction
Morphological causativization is a pervasive phenomenon in Formosan languages. According to Starosta (1974: 283), many Formosan languages have morphological causative verbs marked by a prefix pa-, and five Saisiyat prefixes
related to pa-, including pa-, pae-, pa’-, pak-, and pas- are recognized (Starosta 1974:
328). We do not find the two prefixes pae- and pa’- in our field notes. In Starosta (1974) neither discussion on the function or use of pae- nor any example with pae- is provided. Given that /a/ adjacent to glottal sounds is usually pronounced as /ae/, it is legitimate to postulate that pae’- is an allomorph of pa-. As for pa’-, only the following example is given (with gloss added by the author).
(45) Saisiyat (Starosta 1974: 328)
a. kaheya’ minahaw ila ralom yesterday boil Incpt water ‘Yesterday the water boiled.’
b. koko’ pa’-inaewaeh ka ralom old woman Cau-boil Acc water ‘The old woman made the water boil more.’
Nevertheless, it is possible that the glottal stop in pa’- belongs to the root of the verb
‘boil’ given that verbs beginning with glottal stop take m- as their AF marker. Among the other three causative prefixes, pas- is taken as a lexical prefix in this dissertation because it denote a process in most cases, as in (46a-b), and the meaning of causativization is rendered only when combined with kayzaeh ‘good’, as in (46c).
(46) Saisiyat a. pas-korkoring
make-child
‘give birth to a baby (derogative)’
b. pas-kilkilo’
make-baby beast
‘give birth to a baby (animal)’
c. pas-kayzaeh make-good
‘make something good, prepare’
Consequently, only pa- and pak- are treated as causative morphemes in this
dissertation. As aforementioned, action verbs take the former (47) while stative verbs take the latter (48).
(47) Saisiyat
a. ’oma sopaloy kahoey m-karbon ila Rel big tree AF-fall Incpt ‘The big tree fell.’
b. ba:i’ pa-karbon ila ka kahoey wind Cau-fall Incpt Acc tree ‘The wind blew down th tree.’
(48) Saisiyat
a. baki’ bi’e: ’iniSo’
grandpa angry 2S.Dat ‘Grandpa is angry with you.’
b. hini’ ’owaw pak-bi’e: yakin this matter Cau-angry 1S.Acc ‘This matter made me angry.’
Blust (1999: 349) points out in Pazeh and many other Formosan or non-Formosan languages, there are two forms of causative prefix: pa- and paka-. The former pa- derives causative verbs from action/dynamic verbs, while the latter paka- forms causative verbs from stative verbs. Blust (1999: 354) argues that the two should have been grammatical variants in PAN. However, Zeitoun and Huang (2000: 391) advance that paka- should better be analyzed as bi-morphemic, with pa- indicating causativity and ka- stativity. Although the prefix for forming causative verbs from stative verbs in Saisiyat is mono-syllabic, by comparing such pairs of morphemes as pa-/pak-, ka-/kak-, ’i/’ik that go with action and stative verbs respectively in causative, nominalization, and negative constructions (see Section 2.3.3), we may find that their difference lies in the existence of k, and therefore k should be isolated as a marker for stative verbs in certain constructions.
3.3.2 Function of k in causativization
As mentioned in section 2.3.3, the morpheme k is the functional equivalent of ka-, which is reported to bear a close relationship to stative verbs in many other Formosan languages. As the realization of this morpheme varies with languages, capital KA is used in Yeh (2001a: 17). Regarding the function or status of KA, there are two different views – most studies (Li 1973 on Tanan Ruaki, Starosta 1974, Lin 1992 on Paiwan, Yeh 2001a on Formosan languages)20 take KA as an inchoative marker21 while Zeitoun and Huang (2000) and Zeitoun (2001) regard it as the nonfinite counterpart of the stative marker ma-. The essential difference between the two analysis lies in whether any semantic content of the grammatical marker is assumed. The inchoative analysis (e.g. Li 1973) seems to presuppose that a sense of inchoativity is instilled by
KA whereas the nonfinite marker analysis does not. If we question about why there is an extra marking KA for stative verbs in such constructions as morphological causativization or what motivates the existence of KA in these constructions, it seems that the inchoative marker analysis can tell us more about its function.22
First, let’s examine the difference between actions and states. According to Smith (1997: 17), the main difference lies in their temporal features – actions or events
20 According to Yeh (2001a), it is an inchoative marker in imperative and causative constructions, but has begun to undergo grammaticalization in negative constructions to a ligature. Thus, in some languages the distribution of KA is not restricted to stative verbs any longer.
21 Chang (1992: 34) and Chen (1996: 47-48) analyze ka in Paiwan and ku in Seediq, respectively, as an activizer. Although the term is different, the analysis is in essence the same as the analysis taking it as an inchoative marker.
22 The following pair of sentences from Tanan Rukai (Li 1973: 159) shows that ka- indicates inchoative.
(i) Tanan Rukai (Li 1973: 159)
a. ma-lisilisi nakua kay maruDang is-being-angry me this old man ‘This old man is being angry at me.’
b. ay-ka-lisi-lisi musua kay maruDang Fut-be-getting-angry you this old man ‘This old man will be getting angry at you.’
possess “stage property”, i.e., they occur in successive stages located at different moments while states consist of an undifferentiated period without any internal structure. Now let’s turn to look at the meaning of imperative and causative constructions. The prototypical use of imperatives, getting someone to do something (Cruse 2000: 339), involves a change. Therefore, imperatives require non-stative verbs and as noted in Smith (1997: 59), counterexamples imply inchoativity: change into a state. According to Croft (1994: 94), causation also involves a change (“a transmission of force”), while states “do not normally effect a transmission of force”.
Given this, it is not surprising for stative verbs to undergo some morphological process to occur in a causative construction. This motivates the analysis of KA as an inchoative marker as in Li (1973) and Starosta (1974).
3.3.3 Causatives and ditransitives
Like most Formosan languages, pa- is added to a verb root in AF constructions.
In other words, pa- does not co-occur with the overt AF marker -om-, m- or ma-.
However, as shown below, causative verbs can have PF forms, and the suffix PF -en marks Causee as the subject.
(49) Saisiyat
a. ’oya’ pa-∅-pa’rem ka korkoring mother Cau-AF-sleep Acc child ‘Mother make the child sleep.’
b. korkoring pa-pa’rem-en ni ’oya’
child Cau-sleep-PF Gen mother ‘The child was made sleep by Mother.’
Besides PF forms, causative verbs can also surface as I/BF forms to mark the Patient argument of the verb root as the subject.
(50) Saisiyat
a. yako s-om-i’ael ka ’alaw 1S.Nom eat-AF- Acc fish ‘I ate the fish.’
b. ’oya’ pa-si’ael ka korkoring ka pazay mother Cau-eat Acc child Acc rice ‘Mother fed the child with rice.’
c. korkoring pa-si’ael-en ka ’awpil ni ’oya’
child Cau-eat-PF Acc sweet.potato Gen mother ‘The child was fed with sweet potato by Mother.’
d. pazay si-pa-si’ael ka korkoring rice I/BF-Cau-eat Acc child ‘The rice is to feed the child.’
Note that both the case realization and the focus marking of causative verbs in Saisiyat pattern with those in ditransitive verbs. Compare (50) with the following examples:
(51) Saisiyat
a. yako ’am mobay ka rayhil hisia 1S.Nom Fut AF-give Acc money 3S.Acc ‘I will give him money.’
b. ’ataw be:ay-en ni ’iban ’aehae’ korkoring ’ataw give-PF Gen ’iban one child ‘’ataw was given a child by ’iban.’
b. nisia tatpo’ si-be:ay ila ka korkoring 3S.Gen hat I/BF-give Incpt Acc child ‘The hat has been given by him to the child.’
For ease of comparison, we present the case realization and focus marking of the two kinds of verbs below:
DTRANSITIVE : AGENT RECIPIENT TRANSPORTED PATIENT
-om- Nom Acc Acc -en Gen Nom Acc si- Gen Acc Nom
Figure 3.5 Focus and Case Marking of Ditransitive Verbs
CAUSATIVE: CAUSER CAUSEE PATIENT
pa- Nom Acc Acc -en Gen Nom Acc si- Gen Acc Nom
Figure 3.6 Focus and Case Marking of Causative Verbs
Such relationship between causatives and ditransitives is by no means idiosyncratic of Saisiyat. Croft (1991c: 54) states that “ditransitive verbs are essentially causative versions of the transitive stative verbs (causing a locative, possessive, or the other stative relation to come about, as in put, give and show, respectively) and so can be subsumed under the causative type”. Goldberg (1995: 144-145) observes that the English verbs bring, buy, get, give are polysomous – besides their basic sense involving a transfer from an Agent to a Recipient, they also imply a sense of causation, as shown below.
(52) English (Goldberg 1995: 144) a. The medicine brought him relief.
b. The rain bought us some time.
c. She got me a ticket by distracting me while I was driving.
d. She gave me the flu.
Goldberg (1995: 145) argues that these senses are instances of the metaphor “causal events as transfers”. In other words, transfer in the domain of giving and receiving is metaphorically extended to the domain of causation. Therefore, causing someone to doing something is perceived as causing someone to receive something.
It should be noted that English makes uses of the ditransitive verbs to express causing someone to receive something or into some state. Thus, the meaning implied appears to be more stative. However, in Saisiyat, causative verbs make use of the focus and case marking of ditransitive verbs because of cognitive contiguity. As shown in (53), what is implied is an action or event rather than a state – cause someone to do something.23
(53) Saisiyat
a. ’iban pa-hiwa: yakin ka baboy ’iban Cau-cut 1S.Nom Acc pig ‘’iban made me kill a pig.’
b. yako pa-hiwa:-en ni ’iban ka baboy 1S.Nom Cau-cut-PF Gen ’iban Acc pig ‘I was made to kill a pig by ’iban.’
c. baboy ni ’iban si-pa-hiwa: hi ’obay pig Gen ’iban I/BF-Cau-cut Acc ’obay ‘’iban’s pig was killed by ’obay.’
‘The pig was killed by ’obay under the force of ’iban.’
It is postulated that the difference arises because in Saisiyat, the received Patient is marked by si-, the focus affix for marking instrument.24 In other words, it is perceived as an instrument causing an event and therefore the meaning can be action.
23 Therefore, stative verbs have to be inchaotivized first by adding k- so as to appear in a causative construction.
24 It is noteworthy that in English, recipient can be understood as beneficiary (Goldberg 1995: 146).
Given that si- can also mark beneficiary, we may understand the relation between ditransitive and causative and constructions from the standing point of beneficiary. However, in view of focus marking, we find that recipient is interpreted as patient or a kind of goal because it is marked by PF or LF in Formosan languages.
3.4 Reciprocality
In Section 3.2, we mentioned that Ca- reduplication in Saisiyat indicates reciprocality, as shown below.
(54) Saisiyat
a. korkoring ki ’ahoe’ sa-so:aw child Com dog Ca-chase ‘The child and the dog chased each other.’
b. lasia pa-panae’
3P.Nom Ca-shoot
‘They are shooting each other.’
As just discussed, pa- is the causative morpheme for action verbs. However, it is also found that a morpheme of the same shape as pa- can also express reciprocality.
Compare the following sentences.
(55) Saisiyat (Teaching child)
a. So ’oya’ ki ’yaba’ m-pa-’oe’oe’,
when mother and father AF-Recp-shout ’izi’ pa-k-bazae’ ka korkoring
Neg Cau-Inch-hear Acc child
‘When mother and father have a quarrel, do not let children hear it.’
b. ’izi’ pa-’oe’oe’ kir ma’iyaeh Neg Recp-shout with people ‘Do not quarrel with people.’
c. lasia ’-om-oe’oe’
3Pl.Nom shout-AF- ‘They are crying out loud.’
(56) Saisiyat (Stone and feather)
a. bato’ ki patonay So si-pa-tono’
stone and steel when I/BF-Recp-bump risa m-wa:i’ ka hapoy
then AF-come Acc fire
‘If stone and steel are brought to collide, fire comes about.’
b. ’izi’ tono’
Neg stike ‘Don’t strike.’
As shown in the above sentences, pa- should be the root form of the reciprocal prefix, and the corresponding AF form is ma-. This reciprocal pa- and the causative pa- should be taken as different morphemes that are homophonous for at least two reasons.
First, the meanings they express do not appear to be related. Second, they show different paradigm in terms of focus marking – while causative pa- do not co-occur with any overt AF affix, the reciprocal pa- takes m- to form its AF form. Historically, they are of different sources: the PAN causative prefix is *pa-, whereas the reciprocal prefix should be *paR- inferring from the remark in Ross (1995: 772) that the Formosan reflexes of PAN *maR- function only as reciprocal marker. According to Blust (1999: 345), ma- also marks reciprocality in Pazeh; however, he does not mention if the root form of reciprocal ma- is pa-. According to Li and Tsuchida (2001:
172), the reciprocal prefix paa- contains long vowel and therefore is morphologically distinct from the causative prefix pa-, and the AF form of the reciprocal prefix is maa-.
Listed below are examples from Li and Tsuchida (2001).
(57) Pazeh (Li and Tsuchida 2001: 173)
a. maa-pana ‘throw at or shoot at each other’
b. maa-patus ‘shoot each other with guns’
Ishbukun Bunun also marks reciprocality by pa-, which can co-occur with the AF prefix ma-, as shown below.25
25 More examples include ma-pa-ludah ‘beat each other’, ma-pa-sida ‘get married’ and so on.
(58) Ishbukun Bunun (Yeh 2002: 8)
a. ma-i-pa-sadu kata takana AF-Past-Recp-see 1P.Nom yesterday ‘We met each other yesterday.'
b. ma-pa-kalat a asu AF-Recp-bite Nom dog ‘The dogs are biting each other.’
Besides Ca- reduplication and pa-, Yeh (2000a: 60) and Zeitoun (2001: 131) both list makak- as a reciprocal marker. The following sentence pairs from Zeitoun (2001: 131) reveal that the root form is pakak-, and makak- is the AF form.
(59) Saisiyat (Zeitoun 2001: 131)
a. yami ’ina makak-Siae’
1PE.Nom Exp Recp-happy ‘We played (together).’
b. yami ’oka’ ’i pakak-Siae’
1PE.Nom Neg Lig Recp-happy ‘We did not play (together).’
(60) Saisiyat (Zeitoun 2001: 131) a. yami makak-sekela’
1PE.Nom Recp-know ‘We know each other.’
b. ’ita’ ta-pakak-skela’
1PI.Nom Mod-Recp-know ‘Let’s (get to) know each other.’
Zeitoun (2001: 131) points out that action verbs undergo Ca- reduplication to form reciprocal verbs while stative verbs take the prefix makak- to form reciprocal verbs.
Given that pa- or ma- alone can denote reciprocality, it is legitimate to assume that pa- is the morpheme for reciprocality. Then, what is the function of kak-? In Section 2.3.2, a distinction between ka- and kak- for action and stative verbs in nominalization constructions is noticed. Consequently, k should be the same as the k in causative
morpheme. If kak is composed of the nominalization ka- and the inchoative marker k, then we should be able to find maka- marking reciprocality of action verbs. However, instead of maka-, we find ma- co-occurring with Ca- reduplication to express reciprocality. Consider:
In other words, reciprocality is double-marked – by ma- and Ca- reduplication simultaneously. Given that ka- is the functional equivalent of Ca- (see Section 1.2.7.2), it is possible that makak- is composed of ma-, ka- and k, with the first two morphemes denoting reciprocality and the third one indicating inchoativity, though why there exists double marking in the reciprocal marker ma-Ca- remains to be accounted for.27
Table 3.4 below summarizes the reciprocal markers in Saisiyat.
26 Though why sometimes in negative constructions, there is no Ca- reduplication remains to be explained. Compare:
27 In Pazeh, the reciprocal prefix pa- can also be combined with (CVCV-) reduplication or ka-.
Consider:
(i)Pazeh (Li and Tsuchida 2001: 172)
a. maa-tebe-teber ‘to hit each other with fists’
b. maa-teze-tezek ‘to kick each other’
c. maa-bake-baket ‘to hit each other’
(ii) Pazeh (Li and Tsuchida 2001: 173) a. maaka-kawas ‘to talk with each other’
b. maaka-hape-hapet ‘to love each other’
The functions of CVCV- reduplication and of ka- are not clear.
Table 3.4 Reciprocal Markers in Saisiyat
Form Example Meaning
Ca- ka-koring ‘fight with each other’
ma- (pa-) ma-’oe’oe’ ‘quarrel’
ma-ka-k- (pa-ka-k-) makak-Si:ae’ ‘play with each other’
ma-Ca- (pa-Ca-) ma-Sa-Sebet ‘beat each other’
The two markers ma-ka-k and ma-Ca- can be differentiated by action and state of the verbs, whereas the distinction between Ca- reduplication and ma- and why the two reciprocal markers pa- and Ca- reduplication are combined to express reciprocality remained unexplained. Further investigation is necessary.
3.5 Summary
In this chapter, the mismatches between form and meaning manifested by reduplication, morphological causativization and reciprocality are discussed. For reduplication, it is found that different patterns designate different semantic contents and may show different morphosyntactic behaviors (e.g. they take different verb stems). Under the working hypothesis that similarity in form reflects contiguity in meaning, we try to find the semantic relationship between Ca- verb forms and Ca- instrumental nouns, which is considered to be unrelated in Blust (1998). Besides, various meanings signaled by CVC- reduplication, including the seemingly contradictory ones like plurality and diminution are found to result from the difference in projecting the iconic meaning ‘increased quantity’ conveyed by CVC- reduplication. However, not all the cases get accounted for. The semantic relationship between reciprocality and I/BF future or instrument nouns remains unclear and needs further investigation.
For morphological causativization, it is found that the two causative morphemes
pa- and pak- take different verb roots – the former takes action verbs, while the latter goes with stative verbs. The causative prefix pak- is argued to be composed of two morphemes – the causative marker pa- and the inchoative marker k. Besides, the similarity in focus and case marking between causative and ditransitive constructions can be ascribed to the cognitive contiguity of the two constructions.
Regarding reciprocality, through the comparison with other languages like Pazeh, it is found that the two markers – reciprocal pa- and causative pa- should be taken as two different morphemes although they are in the same form. The identity or similarity in form results from historical sound change where the final consonant of PAN *paR- gets lost.