achievement in japanese — theoretical...
TRANSCRIPT
Achievement in Japanese — Theoretical Implications
Toshiyuki Ogihara University of Washington
[email protected] Chronos 11 (17 June, 2014)
This handout and other papers: http://faculty.washington.edu/ogihara/
Basic Data
• -te iru in Japanese seems to be ambiguous. (1) Achievement + teiru: progressive
kanojo-wa utat-te iru ‘She is singing’ she-top sing-te iru
(2) Process + teiru: resultative taore-te iru fall -te iru ‘being on the ground (after having fallen)’
2
Previous analyses
• Ambiguity thesis — an obvious possibility, but not very revealing
• Differences in the lexical seman>cs of verbs in Japanese and English (Ogihara 2004)
• Gist: Japanese achievements refer to states that result from a change; English achievements refer to preparatory stage that lead to a change of state
3
Achievements
• Jp: taore ‘fall’ indicates |-‐-‐-‐-‐-‐-‐-‐-‐-‐-‐-‐-‐-‐-‐-‐-‐-‐-‐-‐-‐-‐| the state of being flat on the ground
• Eng: fall over indicates |-‐-‐-‐-‐-‐-‐-‐-‐-‐-‐-‐-‐-‐-‐-‐-‐-‐-‐-‐-‐-‐| upright on the ground
te iru/be ing indicate being in the middle of a relevant interval. Major difference between Jp and Eng.
4
Activities/Accomplishments
• Jp: aruku ‘walk’ indicates |-‐-‐-‐-‐-‐-‐-‐-‐-‐-‐-‐-‐-‐-‐-‐-‐-‐-‐-‐| the process of walking
• Eng: walk indicates |-‐-‐-‐-‐-‐-‐-‐-‐-‐-‐-‐-‐-‐-‐-‐-‐-‐-‐-‐-‐-‐-‐| the process of walking
te iru/be ing indicate being in the middle of a relevant interval No difference between Jp and Eng here.
5
issues
• One major issue here: intuitively, falling over = taore, and positing a lexical semantic difference is not intuitive.
• Including the preparatory stage or the resultant state as part of the verb’s meaning is not optimal.
• An alternative proposal will be presented in this talk.
6
Thematic roles
• Jp -te iru: sensitive to thematic roles associated with the subject, not the duration (Okuda 1978, Kudo 1995, Shirai 2000)
• on-going process interpretation: the subject is an agent/experiencer
• Resultant state interpretation: the subject is a theme/undergoer
7
Kratzer (1996)
• External argument is introduced by Voice (3) Mittie Agent feed the dog. ⟦feed the dog⟧ λxλe [feed(the dog)(e)] ⟦agent feed the dog⟧ λxλe [Agent(x)(e) & feed(the dog)(e)] ⟦Mittie agent feed the dog⟧ λe [Agent(Mittie)(e) & feed(the dog)(e)]
8
Parsons (1990)
• thematic roles are relations between individuals and events/states; verbs are (simple) event predicates
(4) Brutus killed Caesar. (5) ∃e[Subj(e, b) & Obj(e,c) & killing(e)] • Kratzer’s proposal adopts this re the
subject, but not re the object
9
Japanese data
• Japanese allows almost all nominals to be dropped.
(6) A: Tabe-ta? B: Tabe-ta. eat-PAST eat-PAST
• Perhaps nominals in Japanese are truly optional.
10
• Some “case markers” in Japanese have variable semantics. Kuno (1973) among others
(7) A: Kimi, dare-ga sukina no? you who-NOM like-Q
B: Mary-ga suki-da. Mary-NOM like-PRES Who do you like? I like Mary.
(8) A: Ano otoko-o sukina hito iru-no? that man-ACC like person be-Q B: Mary-ga suki-da (yo). Mary-NOM like-PRES Is there someone who likes that guy? Mary likes him. Intuitively, the meaning of suki-da ‘like’ is invariable in (7) and (8).
11
Hypothesis
• Relevant achievement verbs in English and Japanese have the same semantics.
• All “nominal arguments” are introduced via independent “thematic roles” (or syntactic heads) in Japanese. Verbs denote (simple) predicates of events. This is a neo-Davidsonian proposal faithfully applied to Jp.
12
• Intuition: We must show the difference in what is happening to the subject now.
• Assume that a thematic role is a three-place relation involving a time, an eventuality, and an entity.
• -te iru interacts with the subject thematic role
(9) Ki-ga taore-te iru tree-NOM fall-te iru. ∃t1∃e[now⊆ t1 & result(t1, e, the_tree) &
falling-over(e)] At the utterance time, the tree has a
resultant state of falling over. (resultant state interpretation)
13
• Simple past/present tense sentences refer to events directly.
(10) Ki-ga taore-ta. ‘The tree fell over.’ ∃t1∃e[TIME(e) < now & result(t1, e,
the_tree) & fall_over (e)] The tree has a resultant state of e at the utterance time.
(11) Ki-ga taore-ru. ‘The tree falls/will fall over.’
∃t1∃e[TIME (e) > now & result(t1, e, the_tree) & fall_over (e)]
Note: TIME is a function from eventualities to times.
14
(12) Tom-ga hasit-te iru. Tom-NOM run-te iru
∃t1∃e[now⊆ t1 & agent(t1, e, Tom) & running(t2, e)] Tom is the agent of e at the utterance time.
(13) Tom-ga hasit-ta. Tom-NOM run-PAST
∃t1∃e[Time(e) < now & agent(t1, e, Tom) & running(t2, e)]
15
The net effect
• The above proposal allows us to distinguish between activities/accomplishments and achievements in relation to -te iru without positing a semantic difference between fall over in English and taore in Japanese.
• The proposal also allows us to see that simple past sentences in Jp and Eng receive the same interpretation.
16
Compositional Semantics
• -te iru is interpreted as in 14: (14) λf<i,<e,t>> . λt . λe [now⊆t & f(t)(e)] • Past tense is interpreted as in 15: (15) λf<i,<e,t>> . λt . λe
[TIME(e) < now & f(t)(e)] • (14) or (15) combines with a sentence
meaning of type <i,<e,t>>. • Types: i = times, e = entities
17
Syntax
My proposal is like Kratzer’s only so far as severing the external argument from its verb. I “extend” Kratzer’s proposal in that even the internal argument is separated from its verb in the semantics.
As for the syntactic analysis, I merely propose that each thematic role introduces a nominal expression.
18
Syntax
(16) [[[Ki-‐ga] [taore]] te iru] tree-‐nom fall teiru
(16) ⟦taore⟧ λe [falling_over (e)] ⟦ki-ga taore⟧ λt. λe [state (t, e, the_tree) & falling_over (e)]
⟦ki-ga taore teiru⟧ λt. λe [now⊆t & state (t, e, the_tree) & falling_over (e)]
19
English
• The role that be -‐ing is similar to that of -‐te iru. However, some nominals bear different thema>c roles than Japanese.
(16) The tree is falling over. (17) ∃t1∃e[now⊆ t1 & theme(t1, e, the_tree) & falling-over(e)] The tree is now the theme of an event of falling over. (on-‐going process interpreta>on)
20
Further mo>va>on
• Thema>c roles can be slightly different across languages. Ikegami (1981), etc.
(18) # Mary burned the firewood. But it did not burn. (19) Al-‐wa takigi-‐o moyasi-‐ta ga, Al-‐top firewood-‐acc burn-‐past but, moenakat-‐ta. burn-‐neg-‐past ‘Al tried to burn the firewood. But it did not catch fire.’ We can say that there is a subtle but important difference in thema>c roles associated with the object, not necessarily that burn and moyasu have different seman>cs.
21
References
• Dowty, David R. 1979. Word meaning and montague grammar. Dordrecht: Kluwer.
• Kratzer, Angelika. 1996. Severing the external argument from its verb. In Phrase structure and the lexicon. Ed. J. Rooryck and L. Zaring. 109-‐137. Dordrecht: Kluwer.
• Kudo, M. 1995. Asupekuto, Tensu Taikei to Tekusuto: Gendai Nihongo no Jikanno Hyougen. Hituzi Syobo. Tokyo.
22
References (2)
• Ogihara, T. 1998. ‘The ambiguity of the -‐te iru form in Japanese’. Journal of East Asian LinguisEcs 7:87–120.
• Okuda, Y. 1978. ‘Asupekuto no kenkyuu o megune (1), (2) (on the study of aspect).’ Kyoiku Kokugo 33–44: 14–27.
• Shirai, Y. 2000. ‘The seman>cs of the Japanese imperfec>ve -‐teiru: An integra>ve approach’. Journal of Pragma>cs 32:327–361.
23