CHAPTER 3 YAMI VOICE AFFIXES
3.2.2 N ON - ACTOR VOICE AFFIXES
3.2.2.4 The affix ka- -an
國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
that the beneficiary participant could be queried. The causative pa- and pi- are usually observed when the instrument or beneficiary participant is mentioned. This has implied that instrument and beneficiary participants are not core arguments as the actor and undergoer participants, and their presence requires affixation of a causative affix to increase the valency of the predicate.
(73)
a. sino o ya na i-pi-ano-anood ni Mapapo
who NOM AUX 3.S.GEN IV-PI-RED-sing GEN PN
‘Who is the one that Mapapo is singing for? (For whom is Mapapo singing?)’
(Ho 1990: 84) b. sino o i-pa-narang
(i-pan-sarang)
na so ayob ni mapapo
who NOM IV-CAUS-buy 3.S.GEN OBL dress GEN PN
‘Who is the one that Mapapo will buy a dress for? (For whom will Mapapo buys a dress?)’ (Ho 1990: 84)
3.2.2.4 The affix ka- -an
The affix ka- -an is found to attach to roots/stems of cognition, such as know, recognize, etc. Two participants are involved in ka- -an clauses, as illustrated in (74). The sentient participant is genitive case-marked, and the other participant that gives rise to the sentience of sentient participant is nominative case-marked.
(74)
a. ko ka-teneng-an imo 1.S.GEN PV-know 2.S.NOM
‘I know you.’ (Shih 2013: 106)
b. ko ka-tovil-an si mapay
1.S.GEN PV-recognize NOM PN
‘I recognized Mapay.’ (Shih 2013: 106) c. ko ka-tops-an imo
1.S.GEN PV-know 2.S.NOM
‘I know you.’
‧
The ka- -an affixed predicate, its meaning, number of participants involved, semantic property of the involved participants, and its case marking are summarized in Table 36.
Table 36 Case and semantic property of involved participant(s) of ka- -an clauses
The subject of ka- -an clauses is the entity that has been mentioned or recognized and is considered to have the property of being in a state, as shown in Figure 12. The case
marking of the involved participants of ka- -an clauses is the same as the -an and -en clauses.
The nominative case-marked subject of the ka- -an clause has exhibited the property of being in a state that is on the undergoer end of the modified version of the revised macro-role hierarchy while the other participant exhibits the property (sentience) towards the actor end of the hierarchy. This is the reason that the ka- -an clauses are considered as a sub-type of PV clauses in the present study. Furthermore, the roots/stems that can be affixed with ka- -an are not compatible with the affixes -en and -an. This has again suggested that the property or meaning of roots/stems has influenced the choice of verbal affixes.
Figure 12 Semantic property of ka- -an clause subject Actor: volitional performer
causing) an event or change of state sentience
In this section, clauses containing the affixes -en, ni-, -an, i-, and ka- -an are discussed. The semantic property of subjects of these clauses is summarized in Figure 13.
Root/
teneng ka-teneng-an know 2 [NOM], [GEN] stationary, sentience tovil ka-tovil-an recognize 2 [NOM], [GEN] stationary, sentience topos ka-topos-an know 2 [NOM], [GEN] stationary, sentience
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
Figure 13 Semantic property of -en, ni-, -an, i-, and ka- -an clause subjects Actor: volitional performer
causing) an event or change of state sentience
movement A U
stationary
causally affected affected -an, affected i- Undergoer: undergoing a change in state -en/ni-, ka- -an
or being in a state
The -en and ni- clauses that differ only in one feature—perfective, have an undergoer-like subject. There are two types of -an clauses—the affected -an clause whose subject is the entity that is causally affected by the action, and the locative -an clause whose subject is the location where the action takes place. For i- clauses, three different types have been
identified. These include the affected i- clause whose subject is causally affected by the action, the instrument i- clause whose subject is the instrument that is used for carrying out the action, and the beneficiary i- clause whose subject is the participant that benefits from the action. The affix ka- -an that only attaches to cognitive roots/stems involves two
participants—a sentient participant and the entity being in a state that gives rise to the sentience.
Both the affixes ni- and -en serve the same function to signal the undergoer subject.
While the affix ni- signals a perfective event, the affix -en signals an imperfective event. The affix -an serves dual voice functions in the language—signaling an undergoer or location subject, whereas the affix i- serves at least three functions—signaling the undergoer, beneficiary, or instrument subject. Types of different -an and i- clauses are summarized in Table 37. In this section, it is confirmed that one of the functions of Yami verbal affixes -en, ni-, -an, i-, and ka- -an is to signal the subject role of clauses containing them. In additional to that, they serve other functions, such as signaling the perfectiveness of the event.
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
Table 37 Property of clause subject and number of participant involved in i- and -an clauses Clause Type Semantic Property of Subject Number of Participant(s)
Involved affected -an clause causally affected 2
locative -an clause location 3
affected i- clause causally affected 2
instrument i- clause instrument 3
beneficiary i- clause beneficiary 3
Note that location, instrument, and beneficiary are not included in the modified version of the revised macro-role hierarchy. This hierarchy is originally designed to account for events that involve two participants, but the locative -an clause, instrument i- clause, and beneficiary i- clause involve more than two participants. Hence, the subject role of these three types of clauses is not included in Figure 13.
3.3 Conclusion
In this chapter, the clauses containing Yami verbal affixes (Ø, m-, mi-, <om>/om-, ma-, man-/mang-, -en, -an, i-, and ka- -an) are investigated. Studies on Austronesian
languages tend to treat /m/ related affixes (e.g. m-, mi-, <om>/om-, ma-, man-/mang-, etc.) as actor voice affixes and others as non-actor voice affixes (e.g. -en, -an, i-, ka- -an, etc.). The semantic property of the subject of clauses that contain these verbal affixes in Yami is examined via the modified version of the revised macro-role hierarchy and is summarized in Figure 14. The subject of mi-, activity ma-, and man-/mang- clauses exhibits properties towards the actor end of the hierarchy, and the subject of Ø, stative ma-, resultative ma-, potentive ma-, -en/ni-, affected -an, affected i-, and beneficiary i- clauses exhibits properties towards the undergoer end of the hierarchy. From Figure 14, the /m/ related verbal affixes in Yami do not always co-occur with an actor subject. The clauses containing the affixes Ø,
<om>/om-, and ma- have subjects that exhibit property towards the undergoer ends of the hierarchy. The subject of both -en and -an clauses exhibits properties towards the undergoer end of the hierarchy with some subtle meaning differences. The -en clause subject might involve a change of state, and -an clause subject does not.
‧
Figure 14 Semantic property of clause subject
Actor: volitional performer m-, mi-, <om>,
activity ma-, man-/mang- causing an event or change of state activity ma-, man-/mang- sentience
movement m- A U
stationary m-
causally affected affected -an, affected i-, beneficiary i-
The subject of m- clauses exhibits properties including volitional performer,
movement, or stationary. These are properties towards the actor end of the hierarchy, except that stationary is in middle of the hierarchy and could be either considered as actor or undergoer. As for the subject of <om>/om- clauses, it is scattered along both ends of the hierarchy—a volitional performer is actor-like and being in a state is undergoer-like. Based on the available data, <om>/om- clauses are classified into three subtypes—the
one-participant clause with an actor subject, the one-one-participant clause with an undergoer subject, and the two-participant clause with an actor subject. Moreover, the affix <om>/om- is observed in different types of syntactic constructions as an independent or dependent clause in declarative or interrogative clauses. The factors that might have influenced the affixation of <om>/om- involve syntactic (dependent vs. independent clauses) and semantic (stativity) features. The affix <om>om- is observed in interrogative and dependent clauses and is observed in stative-like predicates, such as t-om-ava ‘fat’ and activity-like predicate, such as k-om-an ‘to eat’. This affix serves more functions than one has expected, and its distribution and functions remain unclear and will be left for future studies.
The ma- clauses are subdivided into four different sub-types—stative ma-, resultative ma-, potentive ma-, and activity ma- based on semantic properties of their subject and the meaning of the affixed predicate. These four types of ma- clauses will be further discussed and compared with other clauses in chapter four to further justify our classifications.
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
For clauses that involve more than two participants, the choice of clause subject extends to location, instrument, and beneficiary via verbal affixation of -an, i-, and i-,
respectively. The affix -an that has been considered as a locative or locative applicative affix in Austronesian literature is observed to participate in clauses with an undergoer-like subject in Yami. Hence, it is proposed that the affix -an in Yami serves more functions than
expected, and in addition to functioning as locative affix, one of its functions in Yami is to construct undergoer subject clauses. This will be further discussed and justified in the following chapter.
The verbal affixes seem to solely serve the function of signaling the thematic role of the clause subject or valency changing according to voice and transitivity analyses,
respectively. Under these analyses, the thematic role of clause subject could sometimes be divided into actor and non-actor for the sake of simplicity. However, this subdivision has simplified the problem and neglected the subtle meaning difference among these affixes and their interaction with different sets of roots/stems. The re-examination on the semantic property of clause subjects via the modified version of the revised macro-role hierarchy has shown that these affixes serve various functions in the language and might not be compatible with all the roots/stems. In addition to signaling the thematic role of the clause subject and valency of the predicate, some signal modality (e.g. potentive ma-), perfectiveness (e.g.
undergoer ni-), imperfectiveness (e.g. undergoer -en), or interrogation (e.g. <om>), etc. and the function or functions of these verbal affixes under investigation in this chapter are summarized in Table 38.
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
Table 38 Functions of Yami verbal affixes
Voice Tense Aspect Mood
Actor Undergoer Perfect Imperfect Telic Atelic Potentive Interrogative Dependent
Ø ∨
m- ∨
mi ∨
<om>/om- ∨ ∨ ∨ ∨
stative
ma- ∨ ∨
activity
ma- ∨ ∨
potentive
ma- ∨ ∨
resultative
ma- ∨ ∨
man-/mang- ∨
-en ∨ ∨
ni- ∨ ∨
-an ∨
ka- -an ∨
affected i- ∨
Furthermore, some affixes can only be attached to a specific set of roots/stems and some cannot, and this might be due to the lexical property of roots/stems that remains unexplored. Fortunately, we are able to uncover the subtle difference between -an and -en undergoer clauses, and observe that these two affixes attach to different sets of roots/stems to construct clauses with undergoer or undergoer-like subjects. This has suggested that the structure of a clause (number of participants, thematic role of clause subject, etc.) and its meaning depend on the choice of verbal affix and the set of roots/stems it attaches to.