*!*Ghomeshi, Jila and Diane Massam. 1994. To Appear, Linguistic Analysis
tải về 302.24 Kb.
|
| transitive tests with these same verbs).
I think that the muddy results that HPG&N get in the intransitive test are the result of children trying to cope with a mismatch of the verb and the argument frame. Of course HPG&N believe that the frame is independent of the verbs with which they appear. This might imply (to them) that the converse is also true: the verb is independent of the frame. And that is wrong. The semantics of frames may be independent of verbs, but the semantics of verbs are not independent of frames. Every verb has a "home frame." Even if turn or bend is deep down intransitive, it is not necessarily intransitive at age 2. And I think this is why they got garbage results for the known verbs in intransitive frames. There are several other experiments that involve nonsense verbs; two of these they claim "replicate the work of Naigles (1990)". In one case it is the 29 month old children who replicate results of Naigles, in the other case it is the 24 mo old children. In the latter experiment, according to HPG&N, "The 29-month-old group's visual fixation means surprisingly did not reach significance." Overall, my reaction is this: Too many of these experiments have muddled or insignificant results. But the one thing that stands out is this: the evidence for abstraction seems to appear around 24 months at the EARLIEST, but it is not consistent at this age. Real evidence for abstraction seems to be there closer to 30 months. This is exactly what I would expect based on the one set of studies that I've seen that show abstraction: Bowerman's evidence for overgeneralization. If I remember correctly, her evidence of overgeneralization of ditransitives and lexical causatives first appears slightly earlier than 30 months. For this to occur, productive syntactic behavior must be in place. So, I expect no evidence for abstraction at 18 months, some dicey evidence at 24 months, and solid evidence for abstraction at 30 months. When HPG&N say that results of abstraction at 29 months is evidence of "early abstraction," I take issue with the word "early." -- Bill
*!*Holland, Gary. ms. Transitivity, causativity, and surface
G looks at verbs which take dative case in Old Norse. He argues against the traditional analysis that these verbs were based on analogical extension with a class of verbs whcih took instrumental case (which through syncretization collapsed with dative), largely on the basis of the fact that the dative objects do not ahve instrumental meaning. G argues instead that V + DAT forms a construction, taking both otherwise transitive and intransitive verbs. It implies that the verb impart some motion to their dative objects, or they control or regulate the motion of the dative objects in some manner. In some cases, a causative meaning is created. This was motivated by the loss of a morphological causative marker. Embedded subjects of both transitive and remarkably, intransitive predicates are in dative case. *!*Hopper, Paul J and Elizabeth Closs Traugott. 1993. Grammaticalization. Cambridge: Cambridge University Press. \noindent ! Chapter 4: Pragmatic Inferencing. \noindent Different motivations for change: acquisition communities and language contact
\noindent ! Pragmatic Motivations (having to do with the relationship between language and context: \noindent ! Hearer's role: actively involved, trying to seek unambiguous interpretation \noindent ! Speaker's role: ease (economy): reduce speech signal (ease) be going to to gonna; repeat well-rehearsed phrases (e.g. you know, like, idioms); using old forms for new means (extending meanings, uses) \noindent Bybee: development of morphology invovles processes in which similar semantic functions are subsumed under colsely related grammatical function (1985:204) (grammaticalizatoin is not a matter of communicative "need" since that would imply a stage at which communicative need went unmet. [but communicative desire is still possible]). \noindent Goal of expressivity counterbalances the above trend. \noindent Questions: Do Hopper and Traugott adopt a teleological view of language change? ANS: yes and no: the language is not teleological, but speakers
\noindent ! View that Change is Syntactically Driven: Lightfoot: syntactic change happened indepenently of semantic and pragmatic factors (this view arises from thinking of change from hearer's point of view, H & T suggest). Lightfoot can't explain why the same semantic meanings (word) tend to get grammaticalized with the same syntactic functions cross linguistically [e.g. go and future marker] \bigskip
\noindent They discuss semantic/pragmatic distinctions without making strong claims about the division (they use the Can you pass the salt example?] \medskip
! Semantics: ``meanings that are relatively stable out of context, typically arbitary and analyzable in terms o fthe logical conditions under which they could be true." \noindent ! Pragmatics: ``beliefs and inferences about the nature of the assumptions made by participatns and the purposes for hwich utterances are used in the context of communicative lng use." pg 69 \noindent Discussion of Polysemy vs Homonymy \noindent What is the Haiman/Croft criterion of polysemy as oppsed to homonymy? ANS: cross-linguistic parallels indicate polysemy \noindent What is the difference between conversational and conventional implicatures? How are they related? ANS: conversational implicatures often give rise to conventional implicatures. e.g. since (from temporal to causal) They also briefly mention pragmatic polysemies [e.g. Horn's example of metalinguistic negation] \noindent ! Implication: (post hoc ergo propter hoc) pg 75 A: If it rained last night the ground will be wet. B:The ground is wet. \noindent Ask, Is the inference that it rained last night an entailemnt or an implicature? \noindent ! METAPHOR \eenumsentence{ be going to (future) int the years ahead (future) drink up/down (completive) drink on (continuaticve) come to believe (ingressive) \noindent Do H &T discus any constraints on the grammaticalization of metaphorical processes? ANS: Yes, Talmy (1983, 1988): certain types of spatial concepts become grammaticalized: topological notions. \noindent They argue that certain processes that look metaphorical can be understood as metonymic: arise out of contiguity instead of or in addition to out of analogy. \noindent We need to think of: Lexical item in discourse $>$ grammatical item \noindent
Lexical item $>$ grammatical item \noindent E.g. it is not go which was grammaticalized, but be going to. The contiguity of purposive to was important to change to future meaning. \noindent (they note that "metonymy" is usually used to refer to contiguity in the non-linguistic domain. They use it,quoting Stern, `` ` a word [being] used in a phrase where a notion i nsome way connected with its meaning is liable to form an element of the context." pg 81) use it here). In Rama "go" is grammaticalized as both a purposive and a temporal marker of imminence ("I am going in order to look at the baby"; "I am going to look at the baby" examples pg 84: not entirely clear to me) ``the capacity to create metaphors of time from space may well provide a cognitive framework that supports the changes." pg 84. They discuss their use of the term "pragmatic strengthening" with respect to the common observation that grammatical change involves semantic "bleaching." They want to say that the pragmatic strengthening refers to the initial state of the process: that over time, meanings do weaken, but initiall there is a shift, not a loss of meaning [interesting point]. They suggest that the development of empty "do" might be proposed as a counterexample to the idea that semantic loss is always gradual. It may have arisen from causative do in 1132: ``did him give up" to mean "make him give up." But recent analyses have suggested that the grammaticalizatoin did not arise directly form loss of causative meaning, but via a perfect meaning: discusion pg 89-90. Argument against the immediate loss of meaning: traces of original lexical meaning tend to stickto grammaticalized element (Hopper's persistence): e.g. G\~a (West African lang.): k\'e is an accusative marker derived from the verb "take:" She acc. book lay-down (from She took a book [and] laid [it] down). To this day, k\'e only applies to DOs which are affected,
*!*Hornstein, Norbert and Amy Weinberg. 1981. Case Theory and Preposition Stranding. Linguistic Inquiry 21 No. 1. 55-91. They propose that the marked nature of preposition stranding can be located in a language specific rule of reanalysis. ``the Reanalysis rule is a marked rule of grammar." pg 63. ``
VP] can form a complex V" pg 60. { John [vp [v talked to Harry about Fred]] {John [vp [v talked to Harry about] Fred] ``The only way to allow movement of a wh-element directly governed by a preposition is to prevent its trace from being marked oblique. However the only way to do this is to mark it either objective or nominative. Since nominative Case is limited to subjects and tensed clauses, it follws that we must marked the trace objective. This, however can only be done if the trace of the wh is governed by the verb, which in turn can be achieved only by Reanalysis of the type we have proposed. " pg 63 { *[np e$_{oblique$]: universal filter (Bresnan and Grimshaw (1978): *[PP(temp) P [NP e]] )
On the face of it Reanalysis would seem to predict much more general application of Passive. However, passive is claimed to involve only strings which are possible semantic words. taken advantage of; talked about; cared for; kept tabs on Meanings of strings of words which are sematnic words are
They cite Baltin (1978) as arguing for the need for the notion of "natural predicates": he considers deverbal adjectives such as a talked about reporter; a referred to solution; a taken advantage of saint; *a referred to in 1964 solution, *a talked to Harry about reporter. Interesting facts: \eenumsentence{ Who did Pugsley give a book to yesterday? *Who did Pugsley give a book yesterday to? pg 70 (Pugsley gave a book, yesterday, to Paul) Interesting facts: PPs attached at S do not allow stranding: \eenumsentence{ *what time did John arrive at? * What inning did the Yankees lose the ball game in? (same is true in Dutch) pg 56 What did JOhn decide on? (not ambigious: must have argument reading) pg 58 *WHo did you speak to Harry yesterday about? pg 59 (Who did you speak to Harry about yesterday?) Dutch:
where have you on counted \\ What did you count on? pg 56 [ Notice P is not sentence final.] (but in Dutch all PPs are arguably within the VP since they appear between the Aux and V. H & W suggest that maybe they are sometimes base generated elsewhere, admitting they have no evidence for this, or, maybe the rule should be stated that the PP must be an argument of the verb) They note that the Reanalysis rule cannot really be stated so simply: ``cannot break up P-PP constructions:" \eenumsentence{ *Under what did you jump out from? *What did John jump from behind? *Behind what did John jump from? (footnote 9 page 60) Pesetsky (p.c.): Sentential PPs can't strand at all when they are fronted: \eenumsentence{ In the garden the men talked politics. *what garden in did the men talk politics? pg62 note 12 \eenumsentence{ John talked about Sam to Harry yeterday. ??Who did John talk about to Harry yesterday? pg 71 ??Who did John talk about Harry to yesterday? [bad?] pg 71 \eenumsentence{ JOhn arrived immediately at a solution. Which soln did John immediate arrive at? ??Which soln did J arrive immediately at? pg 71 Problems for reanalysis: \eenumsentence{ WHich problems has Harry been talked to about? Who do you like to be sung to by? (they answer that these are not fully acceptable) \eenumsentence{ What day did J leave on? Which act did J leave the theater before? WHich act did J leave teh theater after? (from Pesetsky pc) pg 79 (THey note taht complicating the structure reduces acceptability: \eenumsentence{ ??What day didn't J take the car on? *It was SUnday that J left for Europe from La Guardia on. *It was the third act of Othello that J left the theater before. *The third act of Othello is tough to leave the theater after. (``We can explain these facts if the procesing of simple sentences like 1a-c can bypass the grammar or at least use it less robustly than thin the processing of the sentnces in ii pg 79") *!*Hovav, Malka Rappaport and Beth Levin. On the Nature of Complex Events. Bar Ilan and Northwestern Universities. 1996. New version published as: Two Structures for Compositionally Derived Events. In Proceedings of SALT 9, 1999. [went over this in class and found out there's something to it. BUT, it should be stated as intrasitives: |---------| transitives: |-----|---------| where dimension is time. Question: how must events be related in order to be conflated in a single
Is the complexity of the event conflation reflected in the syntax? Claim: Two types of event conflation should be distinguished:
Lexical rule account relies on the idea that there are statements of the form, ``A verb in semantic class x is also a member of semantic class y' (1995:29). Problems: it is unlikely that semantic clases are lexical primitives and that
Wechsler (to appear): The semantic relation between the verb and the result XP in the intransitive pattern is tighter (``control": XP is canonical result) than in the two transitive patterns (``raising"). Problems: Does not account for the obligatory appearance of reflexive with unergatives ``canonical result" is difficult to apply: not clear why hoarse is less of a canonical result of yelling than dying is of bleeding. H&L: Claim: The temporal relation between the verb and the result state in the two transitive patterns is different from that in the intransitive pattern. Intransitive pattern: \eenumsentence{ The trapdoor banged shut. The gate creaked open Tracy ran to the station Tracy slowly ran to the station. The temporal progress of the event denoted by the verb is necessarily dependent on the temporal progress towards achievement of result state. the two events share a participant the event denoted by the verb and the event of the achievement of the result state are `coidentified" and analyzed as a simple event with only one subevent. [How are the following analyzed? \eenumsentence{
Tracy ran quickly along the bridge. No result state here: are these analyzed
The temporal relation here is different from that above: the running and the being "towards the station " or "along the bridge" are coextensive, whereas above the running precedes being "at the station" ALso, how would the following be analyzed? \eenumsentence{ \% THe bird chirped out of the tree. \% The man scratched out of the room. I don't find this example acceptable myself (there's
Also,
She leapt to his aide. She leapt over the rail. The milk spilled all over the floor. He tripped into the ravine. *She washed clean. even though there is a homomorphic relationship.
Transitives (Non subcategorized and reflexive pattern): The temporal progress towards the achievement of the result is not necessarily dependent on the temporal progress of the process. \eenumsentence{ The runners slowly ran the pavement thin (running need not be slow) He sneezed the napkin off the table (it is likely that the sneeze ends before the napkin is off the table) [Not construed that way, though] I've sung myself hoarse (upon realizing she is hoarse from yesterday's singing) Tom rocked himself to sleep quickly. (rocking can be slow) [But, He choked to death can mean he choked on something and died sometime after being in a coma] ``Due to the nature of the relation between the process denoted by the verb and the result, the temporal progress towards the achievement of the result is not ncessarily dependent on the temporal progress of hte process." pg 4 [My intuition about these is a little different. I actually argued that the event designated by the verb must extend until the result state. He ate himself sick cannot be used to mean he ate a meal that later made him sick. It has to
But I agree with your judgments above, you can use those sentences in the nonimmediate contexts you describe. Incremental theme: events have a temporal trace and certain verbs specify a semantic constituent which stands in a homomorphic relation to that trace (dowty 1991, Krifka 1992, Tenny 1994). There is only one per event (Goldberg 1995, Tenny 1994). The result XPs are related in varous ways to the incremental theme of the event. Intransitive Pattern: case I:
case II:
an additional implicit event, involving a scale or path. case III: nonsubcategorized NP and reflexive patterns: verb typically doesn't specify an incremental theme dance, sneeze, yell. The result XP names an endpoint. The nonsubcategorized "NP + result" introduce a new event--the acheivement of a rsult. There is no necessary relation between the temporal progress of the event denoted by the verb and that of the result. The entire event is analyzed as complex. Like other accomplishments: activity CAUSE BECOME state. {The widow poisoned the old man to death by putting arsenic in his coffee (The act of putting arsenic in the coffee doesn't extend to th epoint of death) {The policeman shot the escaped convict to death.
Intransitive resultatives are not true causatives (contra Croft 1991 and van Valin 1990): \eenumsentence{ pat jogged himelf stiff. \\ Pat caused himself to become stiff by jogging all day Pat ran to the station ($<>$ Pat caused herself to get to the station by running [no?] In the transitive case, the causing event must precede the caused event and both subevents ahve to be temporally contiguous. In the intransitive case, the temporal relation is much tighter. In the intransitive case, cause and effect do not map onto fixed positions in the sentence (``suggesting that the causal structure is not relevant here''): \eenumsentence{ The car rumbled into the driveway \\ the movement causes the rumblin Tracy ran into the driveway \\ the running causes the motion into the driveway Well formedness condition on syntactic realization (RH&L 1988): There must be at least one argument XP in the syntax per subevent in the event structure template (Grimshaw and Vikner 1993; Van Hout 1996). Accomplishments have bipartite event structure, tehy must have two arguments: no unspecified objects. [They spit into the sink???] Activities are associated with simple event structure, so one argument can be unexpressed: \eenumsentence{ Leslie swept/scrubbed (the floor) Killy broke *(the dishes). As derived accomplishments, reflexive and nonsubvategorized Np resultatives are associated with a bipartite event structure and must take an object to meet wel-formedness condition.o Many verbs specify endstate and not activity (break, melt), but there are no verbs which specify the activity and not the endstate, meaning ``Tracy sang and caused an unspecified result state in the audience" [But see Talmy for Chinese verbs that seem to do just this: they are unspecified for causal effect] Explanation: If an event has only one incremental theme and therefore can only be bounded once, and if the causing subevent precedes the result subevent, then teh result subevent must include the incremental theme and bound the event. Therefore the result subevent cannot be left unspecified. [say what??] Conclusion: If a process and result state share an argument and if their temporal progress
tải về 302.24 Kb. Chia sẻ với bạn bè của bạn:
|