110
1998: Annotated Bibliography Adele Goldberg *!*Afarli, Tor. 1987. Non-Subject Pro-Drop in Norwegian. LI 18 2: 339-345. Element can be missing from the second conjunct of a coordinate verb phrase: Jens hugg ved og stablet op Jens chopped firewood and piled (it) up. "She seized them and put (them) in the oven." The construction is acceptable in most places except southeastern Norway. Also exists in Icelandic. The omitted element need not be a DO: "just I take a newspaper and read in (it)" (example 4) may occur within subordinate clauses: "some workers who hauled snow from the streets and tipped (it) into the sea." *!*Ahrens, Kathleen. 1995. The Mental Representation of Verbs. UCSD Dissertation. Discusses Shapiro's (1987) claim that it is hte number of

1998: Annotated Bibliography - Princeton Universityadele/papers/Papers/Anbib98…  · Web view1998: Annotated Bibliography Adele Goldberg *!*Afarli, Tor. 1987. Non-Subject Pro-Drop

  • Upload
    others

  • View
    1

  • Download
    0

Embed Size (px)

Citation preview

Page 1: 1998: Annotated Bibliography - Princeton Universityadele/papers/Papers/Anbib98…  · Web view1998: Annotated Bibliography Adele Goldberg *!*Afarli, Tor. 1987. Non-Subject Pro-Drop

1998: Annotated Bibliography

Adele Goldberg

*!*Afarli, Tor. 1987.  Non-Subject Pro-Drop in Norwegian. LI18 2: 339-345.

Element can be missing from the second conjunct of a coordinate verb phrase:

Jens hugg ved og stablet  opJens chopped firewood and piled (it) up.

"She seized them and put (them) in the oven."

The construction is acceptable in most places except southeastern Norway.Also exists in Icelandic.

The omitted element need not be a DO:

"just I take a newspaper and read in (it)" (example 4)

may occur within subordinate clauses:

"some workers who hauled snow from the streets and tipped (it) into the sea."

*!*Ahrens, Kathleen. 1995.  The Mental Representation of Verbs.UCSD Dissertation.

Discusses Shapiro's (1987) claim that it is hte number of arg strucframes that was the crucial metric in determining the representationalcopmlexity of the verb, not the maximal number of argumentswithin any particular frame.

Schmaumder (1991) failed to replicate.  Shapiro et al. (1991) didreplicate.

KA points out that not all subcat frames and arg structures were

Page 2: 1998: Annotated Bibliography - Princeton Universityadele/papers/Papers/Anbib98…  · Web view1998: Annotated Bibliography Adele Goldberg *!*Afarli, Tor. 1987. Non-Subject Pro-Drop

included for each verb.  return is classified as non-alternatingbut it is listsed as alternating in the Oxford Advanced Larner'sDictionary.  Fix me a sandwich was not included for fix, secured her the tickets was not included for secure.  There wree only 5-6 verbs per class so these types of errors werevery problematic.

KA's chart is on page 69ff.  When length and frequency is controlledfor as best as possible, there is no difference between verbswith two subcat frames and those with 3 subcat frames.

KA shows that the participant role complexity metric doesaccount for the difference (although it boiled down to 12 ms,it was statistically reliable).

KA shows that 2-participant role verbs are faster to integrateinto a sentence than 3-part. role verbs.   Both verb typesintegrate fastaer into the 2-fronted-arg sentence than intothe fronted adjunct+arg sentence.

*!*Allen, Shanley E.M. 1999? MS Learning about Argument Realization in Inuktitut andEnglish: Graduate development in the use of non-ellipsed forms.  Max PlanckInstitute for Psycholinguistics

Children tend to use lexical NPS for arguments they consider most informativeand to ellipse those argumetns they consider redundant (Greenfield \& Smith 1976;Clancy 1980, 1993, 1996; Givon 1983; Chafe 1987; Du Bois 1987; Hirakawa 1993)

Examples of what counts as inforative are cited (including contrasting args,referents not present in physical context, answers to specific queries, newnessin discourse, etc.)

*!*Allen, Shanley E.M. and Heike Schroder.  to appear (2000?) Preferred Argument Structure in Early Inuktitut Speech Data InJ DuBois, L Kumpf and W Ashby, Prefered Argument Structure: Grammar as Architecture for Function.  Amsterdam: Benjamins.

Page 3: 1998: Annotated Bibliography - Princeton Universityadele/papers/Papers/Anbib98…  · Web view1998: Annotated Bibliography Adele Goldberg *!*Afarli, Tor. 1987. Non-Subject Pro-Drop

Full lexical NPs appear relatively frequently in the S and O roles, butrelatively infrequently in the A role.

*!*Anderson, John R. 1991. The Adaptive Nature of Human Categorization. Psychological Review Vol 98 N 3. 409-429.

Origins of categories:

linguistic labels

feature overlap

similar function

Three views need not be in opposition.

A focuses on categorization of living objects: calculationassumes objects are only classified one way.

Coupling parameter: c: fixed probablity c that two objects come from the samecategory (not dependent on the number of objects seen so far).

P(k) = cn$_k$ / (1-c) + cn

n$_k$ is the number of objects assigned to the category k so far

n is the total number of objects seen so far

For large n, this closely approximates n$_k$/n, which means that there is a strongbase rate effect with a bias to put new objects into large categories.

P(0): = probablity that the new object comes from an entirely new category.

P(0) = (1-c) / (1-c) + cn

For large n this closely approximates (1-c)/cn, which means that the prob ofa brandnew category depends on the coupling probability and the number of objects seen so far:

Page 4: 1998: Annotated Bibliography - Princeton Universityadele/papers/Papers/Anbib98…  · Web view1998: Annotated Bibliography Adele Goldberg *!*Afarli, Tor. 1987. Non-Subject Pro-Drop

the greater the coupling probability and the more objects, the less likely it is thatthe new object comes from an entirely new category.[could be used to explain the stabilization of language]

Empirical findings: review Central tendencies

the reliabilty with which an instance is classified decreases as a function of its distance from the central tendency of the category.

Reed (1972): subjects categorized 10 faces into two categories: their categorization varied with distance of the face from the prototype

Rational model: identified two or more categories, depending on the presentation order.The categories chosen in original experiment were preserved as separate categories by the model.

Influence of particular instancesMedin and Schaffer (1978); Nosofsky (1988) : subjects were trained to

classify 12 colors that varied in brightness and saturation.

Showed a sensitivity to the manipulation of the frequency of a stimulus.

[abstracting away from the different mathematical implimentations]Shepard (1987) Anderson (1991): training on one stimuls generalizes to other stimuli as a function ofsimilarity and of the size of the original category.

*!*Barddal Johanna, 1999.  The Dual Nature of Icelandic Psych-Verbs.Working Papers in Scandinavian Syntax 64. 79-101.

*!*Barddol, Johanna.  Ms 2000.  Case Assignment of Nonce Verbs in Icelandic.

Page 5: 1998: Annotated Bibliography - Princeton Universityadele/papers/Papers/Anbib98…  · Web view1998: Annotated Bibliography Adele Goldberg *!*Afarli, Tor. 1987. Non-Subject Pro-Drop

Argues for importance of isolate attraction, citing Goldberg and Sethuraman and Osherson.

She cites Cruse and Croft (in prep book on Cognitive Linguistics: chapter 10, 12) as observing that low-frequencyconstructions often exhibit low degrees of productivity (not none).that this is expected because high type frequency leads to more general or schematicconstructions while high token frequency leads more specific or substantiveconstructions as entrenched.

*!*Barddal Johanna, 1999.  Case in Icelandic--A Construction Grammar Approach. To appear in Tijdschrift Joorykandinavistiek.

Some confusion about my analysis of negative evidence : "..the learner gradually distinguishes the differentmeanings of different constructions and infers that if a given verb is not used in an optimal construction,then such a usage is not warrented." (pg 5)

Narrowly defined subclasses are discussed as a second learningmechanism.

913 verb uses were collected, most of them novel.

39 sentential patterns.

Borrowing from English provides cases of many novel verb uses.

Cluster attraction (Goldberg 1995) and Isolate attraction.

[V+st \'a](e.g., Netast \'a, ``to write to each otheron the internet)" is cited as a productive construction but -st used to have reflexive meaning and \'a may mean to(?).

Boltast um is cited as novel, "to move around heavily"but it too may be compositional ( um = "around"?)

Page 6: 1998: Annotated Bibliography - Princeton Universityadele/papers/Papers/Anbib98…  · Web view1998: Annotated Bibliography Adele Goldberg *!*Afarli, Tor. 1987. Non-Subject Pro-Drop

Isolate attaction:  new verb formed analogically to oneexisting verb with the same or similar meaning.

The forms are given without glosses, page 15, so I can'ttell what exactly the analogy is.

JB suggests that the argument structure of the target languageis sometimes borrowed, citing t\'ekka inn (``check in"), brenna \'ut (burn out); but since these are complex predicatesin English, it could be said that normal lexical borrowingis involved.

Is generous about my lack of informative discussion about casesin Constructions

Most novel verbs take nominative subjects, accusative direct objects.Many take accusative prep objects as well.

A few (7) take dative subjects, many take dative objects and manytake prepositional objects.

There are also 3 accusative subjects, 2 genitive objects in thedatabase.

*!*Benmamoun, Abbas.  Ms 2000.  The Role of the Imperfective Template in Arabic Morphology.

Abbas argues that Vs are not created by combining roots and templates,but rather from word to word mappings (more like English). 

" If the active form has the vowel /u/ initlaly, the last vowel of hte stem isrealized as /a/.  For this analysis to work, however, one must know what thevowel melody of the active verb is, a possibility that is not allowed under the root to template account" pg 7

McCarthy and Prince (1990) and McCarthy (1993) also suggest a partiallyword to word derivation, but their basic form is hte perfective.

Abbas argues the base form is the imperfective (it also has hte widest distribution, he notes).

Page 7: 1998: Annotated Bibliography - Princeton Universityadele/papers/Papers/Anbib98…  · Web view1998: Annotated Bibliography Adele Goldberg *!*Afarli, Tor. 1987. Non-Subject Pro-Drop

*!*Bowerman, Melissa. 1996.  Argument Structure an Learnability: Is a Solution in Sight?BLS 454-468.

Novel causatives appear around age 2, flourished ...between 3-5, and then continue at a lowerlevel until 12, when they end. (C: 225 tokens, 79 types; E 92 tokens, 54 types).  errorsare summarized in appendix.

Errors with come, go, disappear, stay are unlikely to involve incorrect word meaningsas Pinker had proposed.

One shot innovations for lack of better alternative: stay used causatively 43 times,long after she knew and usually used hte more appropriate verbs keep and leave.

Go used causatively 28 times long after she knew verbs take and send.

Sometimes the erroneous causativiszation did not set in until well after the corect form hadalready been established in the child's speech.

B argues that C and E routinely violated semantic constraints on the construction, causitivizitingnon-dynamic verbs (like be), when there was no "act" (with be, climb), and when the causation was indirect: I want to watch you this book! Everybody makes me cry...you just cried me.(Most of the latter errors involved dolls.  But the rarity of occurence of these forms with"make someone V" meanings was arguably due to a lack of opportunitites to make such errors.  Thatis,they seldom say "make me laugh" OR "laugh me".

So, we need to explain how children in fact do retreat from a causativizing operation that isoverly general.

Preemption:  `it has been proposed that the child's causative might be preempted by the corresponding preiphrastic causative, e.g. make

Page 8: 1998: Annotated Bibliography - Princeton Universityadele/papers/Papers/Anbib98…  · Web view1998: Annotated Bibliography Adele Goldberg *!*Afarli, Tor. 1987. Non-Subject Pro-Drop

disappear.  Bu the extension of preemption to thesecases seems somewhat dubious (Bowerman 1988): lexical and periphrastic causatives are, as constructions, systematically associated with different meanigs, soa child sould not readily allow oneto be supplanted by the other." pg 461 [but see Brooks and Tomasello: they DO preempt]

For C: forms with and without suppletive counterparts declined largely in parallel

Role of semantic subclasses:  If the induction of semantic categories is important in children'sretreat from causative overgeneralizations, semanticaly distant verbs should decline faster than semanticlaly close verbs.  Buterrors with verbs of volitional action, emotional expression and psychological events (all of whichviolate teh directness constraint) hold their own over time against semantically close verbs.

*!*Brown, Penelope. 1998.  Verb specificity and argument realization in Tzeltal childlanguage. draft 1998

Three issues: whether semantic factors also play a role in argument representation verbs outnember new noun types: effect of verb specificity?[X, Gelman and Shatz] Is there any evidence that children initially prefer 'light verbs'with more general meanings or light meanings for their specific verbs?[discuss role of frequency]

*!*Butt, Miriam and Tracy Holloway King.  ms-2000.  Null Elements inDiscourse Structure.

Cites B. Hoffman 1995 Penn dissertation on variable salience of backgroundedinformation.  The computational analysis of the sntax and interpretationof 'free' word order in Turkish.

Page 9: 1998: Annotated Bibliography - Princeton Universityadele/papers/Papers/Anbib98…  · Web view1998: Annotated Bibliography Adele Goldberg *!*Afarli, Tor. 1987. Non-Subject Pro-Drop

In Hindi:  topics appear intially, foci immediately before the verb, backgtrounded material post-verbally

Only continuing topics and backgrounded info can be dropped.

New topics cannot be dropped:

She will live life according to her own wishes.  *[I] was wrong.

"Most of the instnaces of continuing topic drop" involve the 1st person pronoun, but they note that this may be an artifact of the discourse contextsthey looked at.  *Verb agreement is not necessary*, since verbs in Hindido not agree with non-nominative subjects and yet those subjects canstill be dropped.

More than subjects can be dropped:

"Becuase man dopes not ever make a sacrifce for a woman. nor will (he)topicever make (a sacrifice for a woman)backround"

*!*Bybee, Joan.  1995.  Regular Morphology and hte Lexicon. In Language and Cognitive Processes 10 (5), 425-455.

Source-oriented lexical schemas: relate two forms, e.g., present and past tense forms  ( wait-waited)

product-oriented lexical schemas: relate surface forms ( strung, stung, flung, hung)

high token frequency words: have high lexical strength: easy to access, serve as the bases of morphological relations and exhibit an autonomy that makes them resistant to change and prone to semantic independence.The autonomy is also said to lead to weaker connections to other words: are learned on their own, not in relation to other terms.

" Forms of high token frequency will be more autonomous and more likely to be unanalysed, and less likely to participate in schemas; high token frequency forms will thus not contribute to the productivity of a

Page 10: 1998: Annotated Bibliography - Princeton Universityadele/papers/Papers/Anbib98…  · Web view1998: Annotated Bibliography Adele Goldberg *!*Afarli, Tor. 1987. Non-Subject Pro-Drop

pattern (Bybee 1985; Guillaume 1927/1973; Moder 1992)" (pg 434)

In fact, she argues that high token frequency can actually detract from productivity.

Moder (1992: SUNY Buffalo dissertation): subjects were less likely to produce nonce verb forms with irregular past-tense when primed with a high-frequency prime as compared iwth a low frequency prime. [did all have a family resemblance?]  Again, the idea is that high token frequency leads to storage in unanalysed form.

She speculates that it was the incrase in frequency of went that ledto its ability to separate from wend in order to become the past tense of go

Productivity of a schema depends on: 1) the defining property of the schema (how "open" it is) and 2) its strength (type frequency).Within paradigms, words with higher lexical strength serve as the basis for the formation of new words (Bybee 1985) [what does this mean?]

If patterns have an open schema, they gradually increase in type frequency.

German past participles (clahsen and rothweiler 1992): productive affix is t, despite, they claim the fact that it doesn't have higher type fequency than en.  BUT, bybee points out, they count uses with productive separable prefixes as different types (the equivalent of write down, write off, write on). 

German and Arabic plurals  (Marcus et al. 1993)

German -s plural an Arabic "sound" plural have open schemas (are relatively unrestricted), but have low type frequency. Bybee argues that they are indeed default or "emergency" morphemes, but that they actually have lower productivity than English -s or -ed, since most often another plural form is used.  Moreover their use IS affected by existing lexical itema.

Grm -en is used with Fem nouns, -e is used with masc nouns, and er with neuter nouns.

-en has hte highest type frequency, and children generalize it the

Page 11: 1998: Annotated Bibliography - Princeton Universityadele/papers/Papers/Anbib98…  · Web view1998: Annotated Bibliography Adele Goldberg *!*Afarli, Tor. 1987. Non-Subject Pro-Drop

most freuqently.  Fem loan words also use it.

-s was used productively most often in nounds ending in a full vowel, and environment that almost always takes -s in the German lexicon (janda 1990).

If a novel form counted be treated like an ordinary comon N hen a plural marker other htan -s was preferred; only if it could NOT be regarded as a German common n (it was a loan word or proper name),

was -s preferred.

Arabic "sound" plural is also used with noncanonical stems.

*!*Cacoullos, Rena Torres and Jose' Estaban Herna'ndez. 1999.A Trabajarle: La Construccio'n Intensiva En El Espanol Mexicano.Southwest Journal of Linguistics, Volume 18, Number 2. 79-100.

" the intensive construction places emphasis on the action by de-emphasizing the erstwhile patient, which becomse the general locus of hte agen's action."abstract.

"le" is a dative marker, but now it can be used with some non-motion verbs

Sort of like our "go to it!" "hop to it!" expressions.

It occurs frequently in imperatives, presnt tense and infinitives.

*!*Campos, Hector. 1986.  Indefinite Object Drop.  LinguisticInquiry 17 2. 354-359.

Definite objects in Spanish  must be expressed (by clitics or otherwise).

But indefinite objects can be omitted:

Compraste cafe/?Did you buy coffee?

Si/, compre/.Yes I bought (some).

Objects are not omitted in contexts such as RCs, complex subjects, adjuncts etc:

Page 12: 1998: Annotated Bibliography - Princeton Universityadele/papers/Papers/Anbib98…  · Web view1998: Annotated Bibliography Adele Goldberg *!*Afarli, Tor. 1987. Non-Subject Pro-Drop

placement obeys island constraints.

*!*Choi, Soonja. 1999. Early Development of Verb Structures and caregiver Input in Korean: Two case studies.  ??JOURNAL?  Volume 3. No 2-3,Sept 1999: 241-265.

Two Korean children: 15-20 months: brief period of verb islands,followed by remarkable systematicity and coherence in their use of transitiveand intransitive verbs.  Transitive sentences encoded the child or mother as agent actingupon inanimate entities; intransitive sentences encoded a third partyas agent or experience of an event or state.

Production mirrors input.

Hi Soonja, Thanks for sending your paper on verb acquisition in Korean.  I couldn'ttell what journal it's in--please let me know!

 I agree that children are more systematic earlier than Mike Tomasellohas proposed.  Aliza produced some overgeneralized causatives as earlyas 18 months!  But unless we want to claim the structure is innate, therekind of has to be a period of conservatism, however early or brief, asyou found. 

 I think the systematicity comes from categorizing the instances,so inmy paper I tried to draw on work from categorization to understand how that might work.  The frequency of particular verbs and their semantic/pragmatic properties should play a role in what categories are formed.   In English, light verbs are by far the most frequent in both production and input, and lo and behold their semantics ina particular construction corresponds very closely to the overallmeaning associated with that construction.

 I couldn't tell from your article whether that's true in Korean or not.Certainly there are light verbs among the very earliest verbs, includinggive, do, look, go.    Your note says they weren't particularly privileged,though--do you mean in terms of frequency? 

Page 13: 1998: Annotated Bibliography - Princeton Universityadele/papers/Papers/Anbib98…  · Web view1998: Annotated Bibliography Adele Goldberg *!*Afarli, Tor. 1987. Non-Subject Pro-Drop

It is possible to form a category without having the prototype of the category being hte most frequent token (if in fact light verbs even DO constitute the semantic prototype in Korean), ofcourse, so I wouldn't want to claim that light verbs necessarily have tobe among the most frequenty verbs in all languages.

 I've revised my paper on this topic a lot since I sent you an earlierversion.  It's now submitted for publication and I expect I'll need otmake further changes, so maybe I'll hold off sending it until it'sa final final copy.  but I'd love to talk about these things with you morebecause I'm not clear where we differ.

 All the best, Adele

*!*Cote, Sharon Ann. 1996.  Grammatical and Discourse Properties of NullArguments in English.  U of Penn dissertation.

Morgan (1973): questioned whether grammaticality equals sentencehood, but ultimatelyargued that all non-sentential constructions are derived from sentences.

Argues that null subjects, or null subject/aux are used to indicatediscourse boundaries: either new turns or discourse segments (utterance begins, ends, picks up--last one means that a segment is begun after having been left briefly or beingtemporarily sidetracked): pg 81.  Indications of segment boundaries include "now, next,first, second, finaly, furthermore, so, anyway, okay, fine."

ONly 8 out of 190 null subject utterances in the corpus were notmarked as apearing at some sort of boundary in the discourse (attributed to speakerschanging their minds after uttering null subject).

Aruges that they serve the function of marking boundaries.  ("short" sentences maybe more likely to occur at discourse boundaries than internal to turns and discoursesegments).

Page 14: 1998: Annotated Bibliography - Princeton Universityadele/papers/Papers/Anbib98…  · Web view1998: Annotated Bibliography Adele Goldberg *!*Afarli, Tor. 1987. Non-Subject Pro-Drop

Most common null subject was expletive ( it or there): 37%, then missing I,then he/she. 

Examples: (context is given in diss)\eenumsentence Sounds like you have a little one there. $<$ B Cat breeding, huh? $>$  $<$ A: um. yeah $>$ (You)0 make a few bucks on that. So what about your income tax?  0 Think they're hitting you too hard? I just wish it was a little more convenient-- $<$ B: yeah $>$ to do, you know,0 seems like you're so busy anyway, and then that's just one more thing to have to worry about. 0Just doesn't make sense. Well, that's good.  0 Keeps them active, I'm sure. A: Well, we talked long enough.  B: I think so. A: Ok, well 0 enjoyed it. It was great. 0 love plums.

\subsectionNull Objects: Chapter 3

Salient Object Alternation (SOA) (fillmore's DNC): The VP called;Everyone left)

With Fillmore, Cote argues that these are lexical, although not via twoseparate lexical entries. Indefinite Object Alternation (IOA): I'm ironing; Rich sang. Unspecified NP Deletion: Fraser and Ross 1970; Browne 1971, Dowty 1978.

*John consumed.

(again argues for lexical anaylsis) Arbitrary Object Alternation: clause has generic time reference:

This sign cautions against avelanches

Lexical restricted: warn, caution, advise and  many amuse verbs pg 132.

Rizzi (1986) and Bouchard (1987) have given analyses of these.

Page 15: 1998: Annotated Bibliography - Princeton Universityadele/papers/Papers/Anbib98…  · Web view1998: Annotated Bibliography Adele Goldberg *!*Afarli, Tor. 1987. Non-Subject Pro-Drop

This is more generally allowed with agentive subjects.  Levin (1989)calls this the "characteristic property of agents" construction: That dogbites.

Reflexive Object Alternation: lexically constrained washed, shave"Unlike other alternations, HOA isn't restricted to a certain class of verbs" pg 143.

Habitual Object Alternation: You wash and I'll dry.

like SOAs, must refer to something salient in the discourse:\eenumsentence I wonder waht Bert and Ernie are up to now? Oh, they're in the other room. \# Bert's pushing and Ernie'spulling.

As notd by Resnik (and by Fellbaum and Kegle in their discussion of lexically conditioned objectomissions),these seem to be different than other object omissions.  "The lexicalrepresentation for this construction seems to be more like an abstract lexicaltemplate which may be placed over a verb than a feature associated directlywith a particular verb." pg 145

Not much more is said about these here.

Cote argues against the claim that the omitted object is stereotypical,pointing out that you can get other interpretations in the right contexts.( I can't eat).She attributes common stereotypical interpretations to Gricean maxims:speaker must assume hearer can recover the relevant missing object.

Page 16: 1998: Annotated Bibliography - Princeton Universityadele/papers/Papers/Anbib98…  · Web view1998: Annotated Bibliography Adele Goldberg *!*Afarli, Tor. 1987. Non-Subject Pro-Drop

She agrees though that there may be truth in Rice's and Resnik's (1993)claim that verbs that allow object omission in general constrain thosearguments more than other verbs.

She argues that the antecedent of a SOA (called. left) object tends tobetopical, often the speaker.  At the same time, if a null object is used,``the subsequent discourse will not sound as natural if it is about thatentity."\eenumsentence   I ran into John's mother at he supermarket today. She seemed to be in a good mood. John visited yesterday. ?? Susan also sent her some flowers. pg 157

IOA: null object does not have salient antecedent.

Arbitrary null objects: null object represents a discourse entity.Is most like the IOA in that they cannot have intended discourse antecedent.

Habitual Object Alternation: The baby needs a bath--you wash and I'll dry.HOA: have iterative, or repetitive version of action,

involve inanimate objects, and

contrast two different actions.

\ex0 only has third property, so humormay come from viewing this as reminiscent of an assembly  line like, repetitivetask.

SC argues that contrast is required *now is a good time to spill.

RECIPES

Subject cannot be overt (Sadock 1974; Culy 1987; Massam 1989): 

Cote: are not aimed at particular actor:  \#OK, sue, now shape into little balls.

Page 17: 1998: Annotated Bibliography - Princeton Universityadele/papers/Papers/Anbib98…  · Web view1998: Annotated Bibliography Adele Goldberg *!*Afarli, Tor. 1987. Non-Subject Pro-Drop

??You bake 0 1 hour.

\subsectionCentering TheoryNice intro to DRT and other discourse theories of pronominal disambiguation

Hajic/vova/' and Vrbova/' (1982): hiearchy of salience is most important factorin determining reference: entity expressed by a weak pronoun or deleted form: retains the same level of saliency represented as an NP in focus: maximum salience definite NP in topic: next highest level of salience increased saliencey on one entity then all inferrables get two levels less than that entity As for a, concerning a: give a one less than maximum salience entity is not expressed or associated with any entities represented in utterance then it goes down two levels of salience.

Joshi and Kuhn (1979), Joshi and Weinstein (1981).  Grosz, Joshi and Weinstein (1986); formal implementation in Brennan, Friedman and Pollard (1987).

"center" avoids ``murky conflict" over `topic, focus, theme'

center: the most salient/necessary/important thread of continuity in a discourse segment.

Cb: backward-looking center: tied to previous utterance

Ce: established center: center of the first utterance

Hierarchy of ranked entities as possible future centers: Forward looking centers (Cf-)list.

A:Speakers will be most likely to have the same Cb assoiated with two consecutive utterances

B:Will also construct an utterance to make the Cb the most hghly ranked entity on the Cf list (i.e., Cp the preferred center)

Page 18: 1998: Annotated Bibliography - Princeton Universityadele/papers/Papers/Anbib98…  · Web view1998: Annotated Bibliography Adele Goldberg *!*Afarli, Tor. 1987. Non-Subject Pro-Drop

CONTINUATIONS: both A and B

RETAINS: A but not B

SMOOTH TRANSITIONS:  Cb has chagned, but the Cb is hte Cp.

ROUGH SHIFTS: neither A nor B holds.

Brennen et al. 1987 used Cf template based only on grammatical relations:

Subj $>$ Obj $>$ OBJ2 $>$ OTHERS

But Cote points out this doesn't deal with null arguments or inferrables.

She suggests using Jackendoff's LCS to at least deal with null arguments(and also capture the grammatical relations hierarchy, too)

*!*Culicover, Peter W. 1999.  Syntactic Nuts: hard Cases, Syntactic Theory and Language Acquisition reviewed by Jhn R Taylor in Cognitive Linguistics 10-3. 1999. 251-261.

He rejects LAD and proposes the Conservative Attentive Learner (CAL).

CAL has minimal initial structure and learns by generalizing over large structure.   (similar to Tomasello 1995).

Taylor notes that there are affinities between his approach and mine (so doesCulicover).

Doesn't spell out alternative theory too specifically.

He proposes a continuum of markedness between syntactic form and conceptual structure, with so-called "core" grammar being unmarked.

Words and their properties are represented as points in a multidimensional space, and generalizations arise through the clustering of points in the space.

*!*Danziger, Eve. 1998 ms.  Semantics and Form Class in Mopan

Page 19: 1998: Annotated Bibliography - Princeton Universityadele/papers/Papers/Anbib98…  · Web view1998: Annotated Bibliography Adele Goldberg *!*Afarli, Tor. 1987. Non-Subject Pro-Drop

Maya: Implications for bootstrapping.  Dept of Anthropology, U of Virginia.

"Although Mopan children would have no trouble isolating the sematnic class of 'action concepts' from other classes on the basis of its syntactic privileges ofoccurence, they could not necessarily identify the members of htis class asVerbs." (pg 4).

Thus "children will have to do their syntactic bootstrapping language by language--that is, with reference to non-universal semantics-syntax links"*!*De Villiers, Jill G. 1985.  Learning how to use verbs: lexicalcoding and hte influence of hte input.  Journal of Child Language 12. 587-595.

The corpus  of Adam and Eve was analyzed to examine the role of maternal inputin children's use of verbs in particular morpho-syntactic contexts ("constructions").She looked for predictors of children's range of usages for particular verbs.Factoring out the child's own frequency of a given verb (since more frequentverbs had more chances to appear in more frames), she found that the maternalvariety of use was a highly significant predictor of children's use of the sameverbs.  Overall maternal frequency did not predict children's variability.

It is argued that the chidlren were monitoring the grammatical patterns of useof individual verb in the input they received.  The extent of novel verbs bythe child cannot be assessed form these data..." pg 587 from abstract.

 I am probably missing something here, but a paper by De Villiers (1985), JCL,is often cited as demonstrating that a child's own mother's speech has moreinfluence than a random mother's speech.  That is, she tried to show that

Page 20: 1998: Annotated Bibliography - Princeton Universityadele/papers/Papers/Anbib98…  · Web view1998: Annotated Bibliography Adele Goldberg *!*Afarli, Tor. 1987. Non-Subject Pro-Drop

Adam's own mother's variety of use of verbs was a better predictor than someone else'smother's variety of use (Eve's).  The way she talks about how to decided this is the following:

``A step-wise regression analysis was then performed with Adam's variety of useas the dependent variable.  In such an analysis, the indenpendent variablesare added one at a time in a prespecified order as long as they continue to add predictivepower to the preceding variables...Adam's frequency and Adam's mother's variety were chosen for the regression, but Eve's mother's variety and frequency measures failed to meetthe criteria for inclusion.  Evidently, Adam's own mother is a significantly betterpredictor of his own pattern of verb-use than an unacquainted mother.."

My question is this: if variables are added one at a time and Adam's frequency and *Adam's mother's variety* were added FIRST, then Eve's mother's variety would only show an effect if it somehow were a BETTER predictor than Adam's mother's variety, or if it somehow addedsomething that Adam's own mother's speech didn't.

Wouldn't the better thing to do be: add Eve's mother's variety directly toAdam's frequency (which has to be controlled for) and compare that correlationwith the one in which Adam's own mother's frequency is added?

*!*Du Bois, John W.  1987.  The discourse basis for ergativity. Language 63:805-855

Preferred Argument Structure:  Avoid more than one new argument (or one lexical NP) per clause.  Avoid new  (lexical) Actors in transitive sentences.

Full lexical NPs appear relatively frequently in the S and O roles, but relativelyinfrequently in the A role.

Page 21: 1998: Annotated Bibliography - Princeton Universityadele/papers/Papers/Anbib98…  · Web view1998: Annotated Bibliography Adele Goldberg *!*Afarli, Tor. 1987. Non-Subject Pro-Drop

Allen and Schroder, to appear. in J. DuBois, L. Kumpf and W. Ashby, Preferred Argument Structure: Grammaras Architecture for function: Amsterdam: Benjamins. discuss INuktitut acquisition.  Children display the above tendencies.  (have on file)

*!*Chung, Ting Ting Rachel and Peter Gordon. 1998.  The Acquisitionof the Chinese Dative Construction.  A. Greenhill et al. (eds) BUCLD 22Proceedings. 109-120.

They tested 29 Chinese sepaking children from Taiwan on a forced-choicegrammaticality judgment test.  Results showed evidence for NarrowRange Rules that define semantic subclasses a la Pinker 1989.

They note that donate does not ditransitivize in Chinese( juan1) or in Arabic (Al-Osaili 1993): likely to besemantic explanation therefore.

They note, with Kathleen Ahrens that the DO can be a source/loserwhich appears to violate the ``universal" linking rule that formsthe basis of Pinker's broad range rule.  This is actually thedefault (but not most frequent interpretation): if a sentence isambiguous, the DO is construed as source not recipient (e.g.``steal" could be used to mean steal from or steal for., Stealfrom is chosen).

They checked ot see if this class of verbs presented a problemfor the children.

The children were instructed to reward the puppet that said a "good"(grammatical) sentence by putting a sticker on teh color box.

Children were in age groups averge: 4;8, 5;6 and 8;1.

"If children use the semantic subclasses to license use of a particular verb int hedouble object construction, then we should find high correlations in theirjudgments between pairs of verbs within the same semantic subclasses.

We woudl not predict that the semantic subclasses of verbs that fail to

Page 22: 1998: Annotated Bibliography - Princeton Universityadele/papers/Papers/Anbib98…  · Web view1998: Annotated Bibliography Adele Goldberg *!*Afarli, Tor. 1987. Non-Subject Pro-Drop

dativize would showsignificantly high correlations within those subclasses: a confirmation ofthis rules out the possibiity that the intra-class correlationswere simply a reflection of semantic similarity ratherthan a reflection of the role of Narrow range rules. pg 116

Both predictions were confirmed [but I don't really understandhow the latter was arrived at]For dativisable verbs, within-subclass correlation coefficients weresignificantly higher than the between-subclass correlationcoefficients (p<.0005).

Preliminary data does not show that the deprivational sense was particularlydiffficult.

*!*Farrar, William T.  1998.  Investigating Single-Word SntacticPrimes in Naming Tasks: A recurrent network approach.  Journalof Experimetnal Psychoology. Human Perception and Performance. Vol 24. No. 2.648-663.

*!*Francis, Elaine. 1999.  Variation within Leixcal Categories.U of Chicago DIssertation

Francis argues that predicate nominals are best accounted forby positing (two) constructions: the identification and XX.The construction contributes the form and predicational structure.

Very nice.   A lot of functional work is cited.

SHe argues for an autonomous syntax module by noting that these construtionsare too dissimilar to the transitive cosntruction.  ALl they share isthe syntax: no semantics. 

*!*Givo'n, Talmy.  andUte Language Program, Southern Ute Tribe, Ignacio,Colorado. 1980.   The Binding Hierarchy and the Typology of Complements.  Studies in Language 4.3. 333-377.

Another early construction paper: motivation

Page 23: 1998: Annotated Bibliography - Princeton Universityadele/papers/Papers/Anbib98…  · Web view1998: Annotated Bibliography Adele Goldberg *!*Afarli, Tor. 1987. Non-Subject Pro-Drop

Implicative: cause, make, have, force Non-implicative: tell, order, ask, suggest, allow forbid Neg implicative: prevent, stop from

Binding: the stronger the influence exerted over the agent ofthe compleent cluase by the gaent of the main clause verb, by whatever means, the higheris hte main verb on the binding scale. Independence: The higher a verb is on the binding scale the less is the agentof its complement clause sapable of acting independently. Success: The less independence possesssed by the embeded clause agent, and the higherthe main clause verb on the binding scale, the more is the intended manipulation likelyto succeed.

``the `binding strength' of a verb roughly correlates to the degree to whichits complement appears syntactically like an independent/main cluase:

The higher a verb is on the binding scale, the less would its complement tend tobe syntactically coded as an independent/main clause. pg 337.

*!*Goschke, Thomas.  Chapter 8 of book: Implicit Learning and UnconsciousKnwledg: Mental Representation, Computationl Mechanisms, and Brain Structures.

read with Giulia 1999.

*!*Mary Hare,Bowling Green State UniversityH. Wind CowlesUniversity of California, San DiegoMatthew WalenskiUniversity of California, San DiegoRobert KluenderUniversity of California, San [email protected]

Page 24: 1998: Annotated Bibliography - Princeton Universityadele/papers/Papers/Anbib98…  · Web view1998: Annotated Bibliography Adele Goldberg *!*Afarli, Tor. 1987. Non-Subject Pro-Drop

1999Plausibility Effects Following Clausally Biased Verbs:  An ERP Study

Recent research shows that verb-specific structural biases have rapideffects in on-line sentence comprehension, but it remains controversialwhether verb bias affects the initial parse.  This question is oftenaddressed by manipulating the semantic plausibility of potential directobjects after verbs with differing biases.  For example, in an ERP study,Garnsey et al. (1998) found that implausible DOs following NP-biased verbselicited an N400 effect (associated with semantic anomalies), while thosefollowing clause-biased verbs did not.  This suggests that implausible DOsfollowing the two classes of verbs are processed differentially on line.

In an independent ERP study, we addressed this same question usingonly clause-biased verbs.  Unlike earlier studies, our design interposeda prepositional phrase with ambiguous attachment options between thepostverbal NP and the disambiguation point (the embedded auxilliary),to isolate effects at these two points of interest:

  The priest believed the DOCTRINE/CAR despite its problems

                WAS still fundamentally sound.

Plausible ("DOCTRINE") and implausible ("CAR") versions of each sentencepair were assigned to separate lists, augmented by filler sentencesexhibitingvarious complementation options.

The results were as follows:

1)  Implausible DOs ("CAR") elicited a pronounced symmetric slow negative    potential over anterior regions of scalp relative to plausible DOs    ("DOCTRINE"). In addition, implausible DOs showed an N400 effect    relative to plausible DOs, as is the case with implausible DOs    following NP-biased verbs (Garnsey et al. 1998).

Page 25: 1998: Annotated Bibliography - Princeton Universityadele/papers/Papers/Anbib98…  · Web view1998: Annotated Bibliography Adele Goldberg *!*Afarli, Tor. 1987. Non-Subject Pro-Drop

    This suggests that subjects entertained parallel parses of theimplausible    NP:  while they appeared to project this NP as the subject of a newclause,    as indexed by the anterior negativity, they also appeared to hold out the    possibility that it was the DO of the verb, as indexed by the N400.

2)  At the disambiguating auxiliary of the verb phrase, both conditions    elicited a late positivity that continued through the end of the    sentence.  However, this effect was much larger in the plausiblecondition    ("DOCTRINE...WAS").

    The late positivity to both conditions suggests that at this point    in the sentence subjects were forced to reanalyze a possible verb + DO    interpretation.  However, reanalysis was more difficult following    plausible DOs, as indexed by the greater amplitude of this positivity.    This indicates a greater initial commitment to the verb + DOinterpretation    in the case of plausible DOs.

Overall, our results argue that, consistent with claims of lexicalist andconstraint satisfaction theories, verb complement bias does influencethe initial parses that readers assign to postverbal NPs.

Garnsey, S., Atchley, R., Wilson, M., Kennison, S., and Pearlmutter, N.   (1998). `An Event-Related Brain Potential Study of Verb Bias andPlausibility   in the Comprehension of Temporarily Ambiguous Sentences', talk given at   the 11th Annual CUNY Conference on Human Sentence Processing, Rutgers   University, New Brunswick, New Jersey.

Pickering, M. and Traxler, M. (1998). `Plausibility and Recovery from Garden   Paths', Journal of Experimental Psychology: Learning, Memory, andCognition   24 (4), 940-961.

*!*Hock, Hans Henrich. 1984.  CLS.  A Zide D. Magier and E Schiller

Page 26: 1998: Annotated Bibliography - Princeton Universityadele/papers/Papers/Anbib98…  · Web view1998: Annotated Bibliography Adele Goldberg *!*Afarli, Tor. 1987. Non-Subject Pro-Drop

(eds). Proposes a transitivity hierarchy based on South Asian:

intransitives verbs of motion verbs of speaking, oral communication verbs of cognition verbs of consuming other 'affected-agent' verbs (clothe, fight, etc) other transitives

*!*Israel, Michael, Christopher Johnson and Patricia J. Brooks. ms 2000.>From States to Events: the Acquisition of English Passive Participles.

The child is scared clear stative

The child will get scared if you do that (OVERLAP)

The child got scared by the loud noise clear eventive

They argue that children's earliest uses of participles are stative (cf. alsoBorer and Wexler's claim that grammar of passives matures over time),and that they produce overlap (equivocal) uses before clear eventives.

"children build on their knowledge of adjectival, stative participles to learn the properties ofeventive, verbal passives." pg 19.  Earliest uses are functionaly and distributionally indistinguishable from adjectives.

Also, they argue that children should master the morphosyntactic relation between participles and verbs before they extend the participle category to truly eventive, verbal uses.

consistent with "usage-based model" of langacker 1991.

The reason statives are preferred, they suggest is that they are "symbolically

Page 27: 1998: Annotated Bibliography - Princeton Universityadele/papers/Papers/Anbib98…  · Web view1998: Annotated Bibliography Adele Goldberg *!*Afarli, Tor. 1987. Non-Subject Pro-Drop

simpler".  Because states are stable, they are more likely to hold the intsantthey are referred to (cite Gentner for parallel as to why Ns are easier tolearn than Vs). pg 9.

Collected data from 7 corpora in CHILDES (including Abe, Adam, Eve, Sarah, Bloom's Sarah), gathered by hand.

Hymans 1986 and Crain 1992 are cited as nativists.

*!*Iwata, Seizi. 2000.  Locative Alternation and Two Levels ofVerb Meaning. Ms. Gifu University, Japan.

load/spray alternation is discussed

Two levels of verb meaning:

L-meaning: Lexical head level meaning

P-meaning: Phrase level meaning

Certain parts of L-meaing may be profiled: if these are compatiblewith a thematic core (with the rest of the L-meaning being backgrounded),a lexical verb appears in the corresponding syntactic frame. pg 10.

the L-meaning of spray (see below) has 3 stages:  1) a substance-emissionevent (The hydrant sprayed water) 2) motion of a liquid onto a surface(she sprayed water onto the truck or Water sprayed onto the truck) and 3) completing activity (She sprayedteh wall with water).

single L-meaning gives rise to two P-meanings

Notes the large overlap with my constructional approach.

Agrees that Semantic Coherence is necessary.

Disputes Correspondence Principle on following grounds:

Page 28: 1998: Annotated Bibliography - Princeton Universityadele/papers/Papers/Anbib98…  · Web view1998: Annotated Bibliography Adele Goldberg *!*Afarli, Tor. 1987. Non-Subject Pro-Drop

I had:

 load <*loader, container [loaded-theme]>

when in fact it should be:

 load <*loader* [container] loaded-theme>:

Sam loaded the hay [onto the truck]

Sam loaded the truck (with whatever)

Also,

pile < *piler* location, *piled-goods*>

stuff <*stuffer*, *container*, stuffed-theme>

Argh! she's right about this.

Correspondence Principle assumes both non-subject roles are profiled, sinceone or the other gets mapped to obj.

I posited:

spray < sprayer, *target* *liquid*>

Spray and splash also allow:

The broken fire hydrant sprayed water all afternoon.  (Croft 1998 in Buttvolume; 43)

The mudpots spattered mud just as we arrived.

with target unexpressed.  I think my DOC construction makes sense ofthis case now. 

Croft (1998): "other verbs in the spray/load class do not occur inthe simple transitive construction but that is due to the sematnic

Page 29: 1998: Annotated Bibliography - Princeton Universityadele/papers/Papers/Anbib98…  · Web view1998: Annotated Bibliography Adele Goldberg *!*Afarli, Tor. 1987. Non-Subject Pro-Drop

unnaturalness of their construal as a substance emission event or a throwingevent" (pg 43).

``I do not dare say that the Correspondence Principle can be entirelydispensed with.  The last section has simply shown tha the profiledstatus cannot be determined o the basis of hte obligatory expression ofa role; there may be some other means to satisfactorily define profiling." pg 29.

"simply specifying the theme and goal roles as being profiled doesnotallow us to account for the fact that the alternation is not alwayspossible:

 *He spread the sleeping child with a blanket.

Claims animacy is not critical, arguing that the following is ok:

 She spread the baby with baby oil.  ([**!])

..."thus it is the semantic compatibility between the eventualityand the relevant thematic core that ultimately decides whetherthe with variant is possible or not, rather htan the profiledstatus of the role...[spreading] can also be construed as a coveringevent:...one ends up exerting force on that surface as well, albeit indirectly.  The sense of force being applied onto a surface amkes it possible to construe the act in question as affecting the surface.  Hence, spreadfits the thematic core "Affect object by adding substance" and appearsin the syntactic frame V NP with NP." pg 33

She goes on to claim that the fact that pack has only one profiled role(the packer) is a problem since it can appear in the locative alternation.But constructions can impose profiled status (see pg 53 of Constructionsdiscussion of mail in the ditransitive).

Nice discussion of morphological derivation in Japanese and Hungarian:

Japanese: pg 37 ff.

Page 30: 1998: Annotated Bibliography - Princeton Universityadele/papers/Papers/Anbib98…  · Web view1998: Annotated Bibliography Adele Goldberg *!*Afarli, Tor. 1987. Non-Subject Pro-Drop

"smear"

NP1 o NP2 ni V"v NP1 onto NP2"

NP2 o NP1 de V"V Np2 with NP1"

most verbs only occur in one variant but many more can when V is compbinedwith "Tsukusu (to exhaust)":

"sprinkle"

*NP1 o NP2 de V"v NP1 onto NP2"

NP2 o NP1 de V-tsukusu"V NP2 with NP1"

NP1 o NP2 ni V-tsukusu"V NP1 on the NP2"

Note V-tsukusu is a lexical verb.  I. claims that the L-meaningof "maku" (sprinke) is only compatible with the thematic core of"move substance" while the more complex verb "maku-tsukusu" iscompatible with either.

Hungarian: Ackerman (1992)

bare "load" only appears in the onto variant

"meg-load" (with perfective marking) appears only in the with variant.

Argues that "the distinctive character of English should lie in itsexical encoding: that hte L-meaning of a verb can be wider andmore general than in other languages." pg 43

****Dear Professor Iwata,

 Thank you for sending me your papers.  I just finished, Locative Alternation and Two Levels of Verb Meaning, and I foundit really interesting.  I felt there were a couple of misunderstandingsthough, so I wanted to send you some comments.

Page 31: 1998: Annotated Bibliography - Princeton Universityadele/papers/Papers/Anbib98…  · Web view1998: Annotated Bibliography Adele Goldberg *!*Afarli, Tor. 1987. Non-Subject Pro-Drop

 I really appreciate that you have given the Correspondence Principle (andmy analysis of the locative alternation) so much thought.  Yours are the first challenging cases I've been made aware of.

 The general idea behind the Correspondence Principle is to assure that participants that are prominent in the lexical semantics appear in positions that give themsufficient prominence in the discourse: in general lexical semanticsand pragmatics are aligned. 

 You suggest that the fact that pack has only one profiled role(the packer) is a problem since it can appear in the locative alternation.But constructions can impose profiled status on unprofiled participantroles (and sometimes add profiled roles altogether) (see pg 53 of Constructions discussion of mail in the ditransitivefor example of the former, and discussion of kick in the ditransitivefor the latter).  

So there is not problem with my positing for pack:

 pack $<$ *pack*, goods, container $>$

The caused motion construction would effectively profile the goods (puttinggoods in DO position); the causative construction would profile thecontainer.  The above representation seems to be just right when weconsider that the following is also ok:

 c. I already packed.

with indefinite goods and container roles unexpressed.

The correspondence principle only claims that IF a role is profiled,then it should appear in a prominent spot (subj, obj, obj2). 

I do believe you are right that I have an error in my representation for load when I wrote:

 load <*loader*, *container* *[loaded-theme]*>

given the following judgements:

 d. Sam loaded the hay [onto the truck]

Page 32: 1998: Annotated Bibliography - Princeton Universityadele/papers/Papers/Anbib98…  · Web view1998: Annotated Bibliography Adele Goldberg *!*Afarli, Tor. 1987. Non-Subject Pro-Drop

 e. Sam loaded the truck (with whatever)

(there's not really cause to ascribe unethical motives to the error, though, is there?)

The representation you suggest:

 load <*loader* *[container]* loaded-theme>:

accounts for (e) in a straightforward way.  I think (d) can also beaccounted for once the Semantic Coherence principle is taken into account.If the participant roles were to be unanalyzed diachritics, then we'dexpect the container role to ALWAYS map to DO position, but theyare not.  They are supposed to be a shorthand way of capturing therich frame semantic constraints on the participant roles.  SemanticCoherence would prevent the container role from being fused with the"theme" role of the caused motion construction, since the motionof containers is not at issue. So the Coherence and Correspondence principles arein conflict, and as you note, Coherence is more important.

In a paper I just finished (or finished a draft of), Argument Realization,I discuss how Semantic Coherence always holds, but how the Correspondenceprinciple is a default principle that can be overridden by specificconstructions.  I suggest a few particular constructions (beyond passivesand middles which are mentiond in the book) that serve to override theprinciple. I'll send you a copy of that ms.

 In your comparison of your approach with mine, you argue that Ido not take lexical semantics seriously enough.  But the fact thatI use some skeletal representation of verb meaning in terms ofparticipant roles shouldn't be confused with a claim that that's allthat is syntactically relevant.  I have five pages of discussion beforeI introduce participant roles that argues that rich lexical meaningis necessary (27-31).  And as I mention above, the participant rolesthemselves are only intended as a shorthand way to capture the specificlexical semantics associated with the roles.

I don't see a particular problem with spread. Since I associate the syntactic frame involved as a causative one with

Page 33: 1998: Annotated Bibliography - Princeton Universityadele/papers/Papers/Anbib98…  · Web view1998: Annotated Bibliography Adele Goldberg *!*Afarli, Tor. 1987. Non-Subject Pro-Drop

a "with" adjunct (closely related and sometimes identical to theinstrumental with adjunct), I do claim (in a parallel way to what you argue) the event must be construed causatively in order for the "with" variantto be acceptable.  If a construal of applied force is necessary to getthe causative interpretation for a spreading event (i.e. in order toconstrue the surface as a patient to satisfy Semantic Coherence), then it would explain why He spread the sleeping child with a blanket isout.  That is, "The baby" must be construed to be both the surface participant of "spread" AND a patient (affected) argument of the construction.

But actually, I find (b) just as bad as (a).

 a. *He spread the sleeping child with a blanket. b. (*) She spread the baby with baby oil. 

So I think the real problem in this particular case (at least in mydialect) is that it's hard to construe a spreading event as causinga change of state of the surface.  I get:

 I spread the bread with peanut butter.

(marginally), and in that case the p.b. changes the whole form, color,smell and taste (squished, brown, peanutty).

As you note the following example presents an issue for my representation ofspray below:

The broken fire hydrant sprayed water all afternoon.  (Croft 1998 in Buttvolume; 43)

spray < sprayer, *target* *liquid*>

Presumably the sprayer can be construed as a source, so the subject argumentis ok.  What's unexpected then is that the profiled target role can beomitted.  I acually think that a construction I posit in the ArgumentRealization paper mentioned above, and also in a new paper to appear,entitled, Patient Arguments of Causative Verbs can be omitted, can

Page 34: 1998: Annotated Bibliography - Princeton Universityadele/papers/Papers/Anbib98…  · Web view1998: Annotated Bibliography Adele Goldberg *!*Afarli, Tor. 1987. Non-Subject Pro-Drop

accountfor that case.  I describe a discourse context which serves to backgrounda normally prominent (profiled) role.   The combination of discourse contextand omission can be captured by a construction, and therefore can serveto override the Coherence principle, since the principle is only a default.

Finally, while I agree that our accounts agree more than they disagree,I agree that there are certain differences between your account and mine.The most central one in my opinion is that you associate more specific

meanings with the various syntactic patterns than I did.  You associate an "onto" frame with "move substance" when I assumed the framecould be generalized to a Subj V Obj Obl(direction) construction that meant "cause to move".  There may be reasons to posit more specificconstructional meanings--I'd be interested in that question.  AlsoI assume the "with" variant is a general causative with a "with" adjunct,and you ascribe a more specific semantics of "affect object by addingsubstance". 

For me, the following are instances of only two construction:

caused-motion:She put the book on the table.She loaded the hay onto the wagon.She threw the ball over the fence.She sneezed the napkin off the table.She gave the book to the man.

causative + with:She loaded the wagon with hay.She adorned the tree with lights....She broke the ball with a hammer.

Also, I think that constructions can add (and rarely subtract) meaningassociated with verbs, not only highlight aspects of meaning that arealready there.

Page 35: 1998: Annotated Bibliography - Princeton Universityadele/papers/Papers/Anbib98…  · Web view1998: Annotated Bibliography Adele Goldberg *!*Afarli, Tor. 1987. Non-Subject Pro-Drop

In any case, I do hope we can keep in touch--I welcome your comments andlook forward to your future work!

All the best,Adele

*!*Jackendoff, Ray. Chapter 7 draft.  The lexicon: Storage vs onlineconstruction

There is no principled dividing line between happy birthday, signed, sealed and delivered; the bigger they come the harder they fall, and the Illiad.

Lexical item: any unit that is stored, be it bigger or smaller than a word.

Argues that semiproductive and idiosyncratic morphology are both stored, item by item.

Generalization within semiproductive morphology can be stated by lexical (redundancy) rules, but these are really epiphenomenal based on an associative memory.

Cites evidence that high frequency regular combinations are stored too (Baayen references given).

Fully productive morphology is done by rule: free combination.  So "dogs" is a product of free combination, "men" is found in the lexicon.

Notes that this is inhomogeneous solution [but is not really so inhomogeneous, given later discussion?  "[N-s]" in the lexicon as a word-level construction?]

The only difference is that in productive rules, there is a VARIABLE [which heclaims connectionist models are incapable of representing].

Suggests that the child tries to analyze new morphological patterns both ways at once, with one mechanism eventually extinguishing the other. (pg 10).

Page 36: 1998: Annotated Bibliography - Princeton Universityadele/papers/Papers/Anbib98…  · Web view1998: Annotated Bibliography Adele Goldberg *!*Afarli, Tor. 1987. Non-Subject Pro-Drop

What the child has to learn to learn a rule is to learn variables from instances pg 34.  Implicit patterns "go productive" (pg 34) when variable is introduced.o

But then some handwaving: could be a matter of degree [?]; "factors like ease and speed of acquisition relative to stored forms undoubtedly play a role in how freely productive the rule is in performance" Cites Tomasello 1999 approvingly (pg 35)

In P\&P and Optimality Theory, "learning a grammar is conceived of rather like customizing a software package: everything is there, and hte learner has only to choose the options that suit the environment" pg 35.  BUT, UG must be extremely elaborate and very finely tuned: it is hard to imagine all of this structure emerging in the brain prior to experience or being encoded genetically (Elman et al 1996; Tomasello; Deacon 1997).  ALSO, few of the actual parameters have been successfully worked out, despite 20 years of work in the P\&P tradition (Culicover 1999; Newmeyer 1998; Ackerman and Webelhuth 1999; Jackendoff, to appear).

Extracting general variables from previously stored items is guided by UG.Following Ackerman and Webelhuth, UG consists of a collection of skeletal fragments of l-rules built into lexical memory:  "grammatical archetypes."  In traditional terms, the learner is faced with competing generalizations.  If one of thesegeneralizations inherits structure from existing fragments of UG, that one will win the competition. (pg 36--this seems really weak).

UG:

Wdi = X0 (a phonological word correspondes to an X0)

NP = [physical object]

VP = [action]

a syllable consists of a consonant followed by a vowel.

XP-> X0 ...("a phrasal syntacitc category dominates a head of hte corresponding lexical cateogry--Xbar theory")

Each of thee can be violted in the "marked" case based on primary linguistic input.

Page 37: 1998: Annotated Bibliography - Princeton Universityadele/papers/Papers/Anbib98…  · Web view1998: Annotated Bibliography Adele Goldberg *!*Afarli, Tor. 1987. Non-Subject Pro-Drop

[The critique (pg 8) of the past tense model as unable to distinguish between ringV and ringN or hang (thing) from hang (person) is not really fair since it is simply solved by adding at least one node for semantics, which is independently necessary--then the two WILL be distinguished]

On page 7, you note: "the upshot is that the outputs of semiproductive rules must be listed (at least in part) in longterm memory"

But then on page 15: "What would it mean to remove this redundant information from stored regulars? It would amount to deriving them by rules alone--which is exactly what is denied by claiming they are stored!"

I don't understand why you consider telicity to be a definitional part of

resultatives (note 6 page 16).  In your (wonderful) NLLT paper, you argued that aspect does not normally determine argument structure.  So why assume it does in this case?

V pro's way PPV NP[time period] awayV pro's head off

Strict subcategorization is also done constructionally:

[express NP]

movement of the NP is done with a trace in object position..."probably not the only realizatio of these details"

Open question: how constructional idioms compose with each other:

E.g., middle resultative:  that pot doesn't cook black so easily.

"Would you like wine, or would you like a DRINK-drink" (Ghomeshi, Jackendoff, Rosen and Russell 2000).

I don't really see the allure of a "REPEAT" command to account for reduplication.Can't it just be accounted for with a repeated variable? [ a ]-[ a ]

Page 38: 1998: Annotated Bibliography - Princeton Universityadele/papers/Papers/Anbib98…  · Web view1998: Annotated Bibliography Adele Goldberg *!*Afarli, Tor. 1987. Non-Subject Pro-Drop

Default principles that add indpendent biases of "constructional meaning":Preferably:  "subject corresponds to agent, object to patient"Preferably: "subject corresponds to Figure; Object to Ground" pg 24

*!*Jackendoff, Ray. Chapter 12. Draft Phrasal semantics.

"operator" that means "person associated with"  for ham sandwich)cases.

"the sense of repetition must be "coerced" into the sentence: it is constructed online in working memory, with no overt phonological or syntactic evidence" pg 12.

The condition that a *sentence* must have all open variables satisfied ifit is to be well-formed might need some further clarification.  Pro-droplanguages (mentioned in note 4) are the tip of teh iceberg.  There are ofcourse languages that drop any recoverable argument and there are contextseven in English that allow you to drop arguments even when they are notrecoverable (e.g., Scarface killed [] again, or Tigers only kill at night.)I just wrote a paper on this type of omission, that I can send you formore examples and an explanation couched in terms of information structure.

I like this new info tier! 

As Laura may have already chewed your ear about, focus doesn't necessarilycorrespond to just what is stressed (EX?) or what is new to the conversation (*I* think so!).

"Ordinary focus": (contrastive focus)

"Restrictive focus": It was BOB who went to the party.  (individual's role in the situation that is new).

I don't know if it's really true that every sentence has a focus, although I know that that's been widely claimed.  In the paper I'll send we argue that the real constraint is only Gricean: the speaker should have some reason to utter a sentence.  So the cases you mention in note 25

Page 39: 1998: Annotated Bibliography - Princeton Universityadele/papers/Papers/Anbib98…  · Web view1998: Annotated Bibliography Adele Goldberg *!*Afarli, Tor. 1987. Non-Subject Pro-Drop

(You LIED to me!) doesn't convey any new information to the speaker,

although I agree that the purpose is to enlarge the common ground and thisis what satisfies the Gricean imperative.

I also like your take on "syntactocentrism"--nicely put!

You mention the facts about reflexive binding to argue that linguists needto be responsible for accounting for metonymies.  Have you thought about howto account for certain differences among metonymies in whether anaphoric reference is allowed?  Fauconnier 1985 talked about the following asymmetries:

 Plato is on the top shelf.  He was a great philospher.  (he = the man, not the book) The ham sandwich wants his check.  *It was runny. (it = the sandwich, not the man).

I admire your willingness to try to formalize the conceptual structures you'retalking about.  It would get a little scary to try to formalize what'sneeded to expand the sentence fragments, and I notice you don't try to dothem.  I wondered what status the formalism has--is it just a toolfor expository purposes, or do you think there should be some sort of isomorphic mapping to something psychologically real?

You seem to reject treating "she" in "She's coming!" as a topic, since it's notspecially marked, but then you go on to include "my friend BILL" in  (56)as a topic and note that many subjects are topics (pg 30-31).   

 (56) "My friend BILL bought a new JACKET"

But isn't the intonation on "my friend BILL" determined by the nature of the NP, not the discourse property of the NP?  That is, isn'tthe stress determined by the internal, not external syntax of the NP?

 I gave my friend BILL a  BOOK. ?I gave my friend bill a BOOK.

But if that that phrase isn't specially marked in (56), why treat it as

Page 40: 1998: Annotated Bibliography - Princeton Universityadele/papers/Papers/Anbib98…  · Web view1998: Annotated Bibliography Adele Goldberg *!*Afarli, Tor. 1987. Non-Subject Pro-Drop

moreof a topic than "she" in "She's COMING!"?

I like your inclusion of the discussion of conversational implicatures, along with more "grammatical" implicatures.

*!*Jackendoff, Ray. 1999 ms. What's in the Lexicon?

I think HPSG uses the synsem feature instead of splitting up syntaxand semantics. (pg 5)

The Gopnik interpretation of SLI is controversial, to put it politely.Do you know of the Varghakhadem et al work?  They argue that SLI patientshave articulatory, intellectual and orofacial function deficits--the regular past tense trouble could result from trouble co-articulatingmore than one consonant.  (Proceedings of the Nat'l Academy of Sciences 1995)I just looked up references to Gopnik's original paper, and of the 95 papersthat cite it, probably two thirds have a real or implied question intheir titles, like ..a deficit in grammar or processing?

But more generally,I'm also not sure you've given the other side of the Pinker-Rumelhart debateenough credit.  The Hare et al article you cite shows that higher typefrequency is not necessary for the network model to work.Also, papers by Ellis and Schmidt (Lang Cog Proc 1998) and Marchman, Plunkettand Goodman (JCL 1997) address some earlier challenges made my theMarkus camp.

It seems imaginable to me that variables arise out of simply generalizing overspecifics: Do you know of the Osherson et al. 1990 paper?  I discuss it briefly at the end of the Learning paper I sent.  They show that both similarity and "coverage" to existing instances playa role in determining the extent of generalizations.  I think it could provide the key to explaining degrees of productivity,including relatively general productivity. It's not simple frequency that's

Page 41: 1998: Annotated Bibliography - Princeton Universityadele/papers/Papers/Anbib98…  · Web view1998: Annotated Bibliography Adele Goldberg *!*Afarli, Tor. 1987. Non-Subject Pro-Drop

important, but the degree to which attested cases "cover" the relevantfield--very roughly how distinct they are.

Ok, I know, my California roots are showing.

Anyway, I do agree that we don't simply memorize instances.  We're alwaysstriving to make generalizations over the instances. (I'd be curious toknow what you think of the story I'm suggesting in that Learning paperabout how people go from item-specific verb-centered categories to moregeneral linking generalizations.)

I couldn't help notice that in the first part of the paperyou stress the idea that we should be ready to embrace a heterogenoussystem, while in the latter part you champion the idea of a unifiedlexicon for both words, affixes, idioms and constructions, at leastin part because it would be a unified system.  (I actually agree with both points, since there are times we are forced to heterogeneity, butwe shouldn't proliferate mechanisms just for the fun of it. youmight consider adding a sentence explaining the superficially different preferences the paper seems to endorse).

The ditransitive is arguably a constructional idiom, too, even thoughit doesn't have the V NP PP form.

I'm not sure it's important to claim that the unified account stems fromtreating all as having open slots in a VP (pg 21), since as you go on to say,constructions can be (possibly much) larger than a VP.

These are all niggly points--in general I heartily endorse the perspective!!!I also really appreciate your calling a spade a spade when you say, e.g.,"virtually no scientist outside linguistics finds [the language organ]plausible" and your footnote "Regretfully...may linguistsdo not recognize their unnaturalness anymore." 

Oh, gee, and don't let me forget to mention the deep blush I experiencedwhen I read your terribly kind acknowledgment!

Page 42: 1998: Annotated Bibliography - Princeton Universityadele/papers/Papers/Anbib98…  · Web view1998: Annotated Bibliography Adele Goldberg *!*Afarli, Tor. 1987. Non-Subject Pro-Drop

*!*Jackendoff, Ray. ms. 1998.   The Ecology of English N-N Compounds.

Mapping between syntax and semantics:

Cognitive Grammar/(my) construction Grammar:  each piece of syntactic form has a meaning

Transformational approach (Lees 1960, ...Levi 1978): transformation from paraphrase.Generative Semantics.  Lees and Levin proposed a colection of predicatestht contribute their meaning to compounds.  Each compound is massively ambiguous..

Jackendoff:  1) the morphosyntax provides NN has a combinatorial expression;2) canonical principles of syntax to semantics mapping specify the Head Principle.

To account for certain idiosyncracies, you need a theory of thelexicon that permits a degree of redundancy and predictability, since compoundsare also idiosyncratic in limited ways.

On the other hand, NN are fully productive.

Same (of course!) is true of VPs.

Ryder (1994), within Cognitive Grammar, provides a related approach.Lexically specific compounds inherit from more general, productiveconstruction.

subjects in experiment by Ryder (also Gleitman and Gleitman 1970):

 willow forest "a willow that grows in the forest" giraffe land "a giraffe on land"

[One wonders whether experimental fatigue played a role]

Downing 1977: range of meaning relations is completely unsystematic.

Page 43: 1998: Annotated Bibliography - Princeton Universityadele/papers/Papers/Anbib98…  · Web view1998: Annotated Bibliography Adele Goldberg *!*Afarli, Tor. 1987. Non-Subject Pro-Drop

Possible meanings is not entirely unconstrained: "Owl-house" is unlikelyto be interpreted as a house that owls fall on" (Gleitman and Gleitman).But Downing gives counterexamples to some putative impossible relationships(such as negation: earthquake schools: schools immune to earthquake.)

"In standard syntax, the meaning of a novel phrase is pretty welldetermined by the meaning of the words and the principlesfor mapping syntactic structure into combinatorialrelations among constituent meanings. In the case of a novel compound,though, the general principles yield only a broad range of possible approximatemeanings.  The language user must home in on the intended meaning of a novel compound by making use of (a) the sematnic details of the constituent words and (b) thediscourse and extralinguistic context."" [really??? "cat is on the mat,"]

J argues that NN interpretation is fragile and difficult.

J notes that the pragmatic knowledge of N and N constrain possible readings.The meanings of the two words are used to construct a meaning relationbetween them ("co-composition").

*!*Jackendoff, Ray.  1996.  The Proper Treatment of Measuring Out,Telicity, and Perhaps Even Quantification in English.  NaturalLanguage and Linguistic Theory. 14: 305-354

J argues that the notion of `incremental theme' does not play a rolein argument realization since subjects and path PPs in addition toobjects can be incremental themes.  (Dowty also pointed out that incrementality and DO are not necessary--see examble b below)He suggests that what led people to associate incremental themehoodwith DOs is that individuals (often subjects) are not incremental themes.

Page 44: 1998: Annotated Bibliography - Princeton Universityadele/papers/Papers/Anbib98…  · Web view1998: Annotated Bibliography Adele Goldberg *!*Afarli, Tor. 1987. Non-Subject Pro-Drop

\eenumsentence The parade passed the house. (Declerck 1979) John entered the icy water (very slowly) (Dowty 1991) The crowd exited the auditorium (in 21 minutes) (Dowty 1991)(these changes have no effect on the argument structure)

Assumes, apparently with Tenny (1987), Vendler (1957), Verkuyl (1972) Dowty (1979)that telic = accomplishments.

In/*for indicates telic sentences

If DO is bare mass noun or plural, the event is "atelic (activity, process)"

for/*in indicates atlic sentences

in/for indicates ambiguous telicity

The choice of PP affects telicity with motion verbs (Jackendoff 1991)

Some entities ARE affected but do not measure it out: chew, knead, jiggle, spin pg 311:test for affectedness (Lakoff 1970, Jackendoff 1990): What John did to the bread was chew/knead/jiggle/spin it.

But the event is NOT telic: John chewed/kneaded/jiggled/spun the loaf of bread for/*in an hour (pg 312)

One can eat a handful of peanuts in one gulp: ``Thus the measuring out propeorty of eat is evidently connected in part to the nature of the objects in question and htepragmatics of how the action can be carried out on them. pg 312  But whether or not theDO "measures out" the event or makes it telic has no effect on the argument structure.

``In short, Tenny is correct in perceiving an important link beween affectedness (or Patienthood)and the direct argument position (Jackendoff, 1990, chpater 11).  But a range of relativelystraightforward data shows that affectedness and argument structure are not directly connectedwith measuring out and telicity, as Tenny claims." pg 313.

Page 45: 1998: Annotated Bibliography - Princeton Universityadele/papers/Papers/Anbib98…  · Web view1998: Annotated Bibliography Adele Goldberg *!*Afarli, Tor. 1987. Non-Subject Pro-Drop

\bigskip

J also argues that motion should be viewed as a continuous process--he arguesagainst the 'snapshot' analysis of Tenny, early work of his own, Langacker etc:``Except under strobe lighting and in certain cases of pathology, we donot experience the world like an old-time movie, flickering fromone state to another." pg 317.

Also argues for reducing "GO" to "BE"....

On pragmatic nature: requires interpretation of how parade works (\ex0a).Also, eating a handful of peanuts may occur in one shot; a computer candraw a circle by spraying the entire array all at once...

Conceptual Semantics: no distinction in representation between semanticsand pragmatics (jackendoff 1983 chapter 6)

He argues that lexical semanticsof the verb interacts with pragmatic knowledge about the argumentsto determine what is an incremental theme.

*!*Michael P Kaschak and Arthur M Glenberg.  ms.1999.Constructing Meaning: The ROle of Affordances and GrammaticalConstructions in Sentence Comprehension. U of Wisconsin atMadison.

They use denominal verbs, e.g., crutch in ditrnasitive andtransitive constructions and show that the constructions playa role in the interpretatoin. (not a big surprize, but nice thatsomeone did it).  They also show that the arguments play a roleand that non-congruent arguments delay processing time.  Thewhole hting is couched within COnstructoin Grammar and theidea of affordances (Gibson 1979) and meshing (Glenberg 1997).

*!*Kay, Paul and Charles J. Fillmore. 1999.  Grammaticalconstructions and linguistic generalizations: The What's X doing Y?construction.  1-34. Language 75:1.

Page 46: 1998: Annotated Bibliography - Princeton Universityadele/papers/Papers/Anbib98…  · Web view1998: Annotated Bibliography Adele Goldberg *!*Afarli, Tor. 1987. Non-Subject Pro-Drop

\eenumsentence What are they doing resuscitating constructions? (pg 1) What's it doing raining? (pg 3) What's he doing knowing hte answer? (pg 3) I wonder what the salesman will say this house is doing without a ktichen

Semantics:  inconcruity What are your children doing playing in my garden? Howadorable?

Possible origin:  A is up to no good and B asks what A is doing:conversational implicature, but now a conventional implicature.

Must contain the verb do in progressive form:\eenumsentence What was she engaged in under the bed? *what does this scratch do on the table?Does not permit negation: *What aren't my brushes doing soaking in water?

Head + complement construction:

[ [role head; lex + ]  [ role filler; loc +]+  ]

(loc = local complement)

VP construction inherits from this, adds only [cat v] and thatfiller is non-subj.

Head feature principle:The values of the synsem | intrinsic | syn | head path of a head daughter and its mother are shared.

(head attributes of verbs include: inflection, voice, aux, etc)

Subset principle: set values of a head daughter are subsets of correspondingvalues of its mother.

Subset principle is discussed in terms of adjuncts: can be present in

Page 47: 1998: Annotated Bibliography - Princeton Universityadele/papers/Papers/Anbib98…  · Web view1998: Annotated Bibliography Adele Goldberg *!*Afarli, Tor. 1987. Non-Subject Pro-Drop

mother node, but not daughter node. (pg 9)

Valence principle:  A local filler daughter's synsem value is sharedwith the synsem value of a valence element of the mother.

(All constituents appearing in local positions are in fact complements)

"Arguments are valence elements of the minimal lexical verb." pg 11

"Adjuncts are contributed by particular constructions that unifywith a verbal structure" (e.g. setting construction) pg 11. 

Linking constructions: e.g., passive, receipient: unify with minimal lexical constructions:  assign grammatical functionsand often syntactic form to the theta roles of a lexical item.

Failure to occur in certain constructions is accounted for by fullyspecifying a particular entry: contribute, rumored, resembledare given as examples [Problems with this: pull may not occurin the ditrans, but it occurs in several other constructions:cannot stipulate that the pulled thing must be a DO: She pulledon the rope.

LI (left isolation) construction: long distance dependency relatesa constituent with a valence element (not a constituent with a gap).Constituent directly satisfies a complement or adjunct function, insteadof being coindexed with a trace which in turn satisfies the valencerequirement.

Nonsubject Wh-questions inherit both LI and Inversion (SAI pg 18) constructions.

[inherit LI

[wh +] [inherit SAI]  ]

Construction looks like a minimal entry for be, but it is not marked [lex +].``Treating our construction as a lexical entry for be ...wouldlend itself to misleading and unnecessary locutions about a 'sense' of

Page 48: 1998: Annotated Bibliography - Princeton Universityadele/papers/Papers/Anbib98…  · Web view1998: Annotated Bibliography Adele Goldberg *!*Afarli, Tor. 1987. Non-Subject Pro-Drop

be having theinconguity meaning and 'selecting' everything else that defines the construction."pg 19.

Construction represented on page 20.

*!*Kersten, Alan W.  1998. A division oflabor Between Ns and Vs in therepresentation of motion.  Journal of Experimetnal Psychology.

suggests N vs V is grounded in "what" vs "where" neuroanatomical distinction

*!*Kersten, Alan W. and Dorrit Billman. 1997. Event Category Learning. Journal of Experimental Psychology: Learning, Memory and Cognition Vol 23 No 3.638-658.

Contrasts view that objects and events should be learned differently from viewthat learners should use the same strategies on both.

\medskip

Rosch, Mervis, Gray , Johnson and Boyes-Braem (1976): ``good" categories tend toform around rich correlational structure: i.e., some sets of properties arefoud together often, while others rarely or never coocucur.  Given oneco-occurring property, one can predict that the others will also be present.

Learning mechanisms must be capable of detecting rich correlational structurewhen it is present in the environment.

\medskip

Billman and Knutson (1996): people were more likely to discover a correlationbetween the size of a head and tail of a novelanimal when those attributes were related

Page 49: 1998: Annotated Bibliography - Princeton Universityadele/papers/Papers/Anbib98…  · Web view1998: Annotated Bibliography Adele Goldberg *!*Afarli, Tor. 1987. Non-Subject Pro-Drop

to other attributes such as body textureand the time of day the animal appeared.

\medskip

\noindent! Instrumental Verbs

Huttenlocher, Smiley and Charney (1983): demonstrated better comprehension inchildren for ``verbs that involve highly associated objects" (pg 82), likeinstrumental verbs saw, hammer.

Behrend (1990): first verbs used by both children and adults to describe anevent were instrumental verbs than verbs that desribed either the action orthe result of an event.

Gentner (1981) observes that objects have richer correlational structure thando events in general.

\medskip\noindent! Experiments:

Tested whether event categories with rich correlational structure are learnedmore easily than less structured categories.  Tested for knowledge of eventcategories without labels:unsupervised learning task.\medskip

More specifically, tested whether correlations between event attributes are easier tolearn when forming part of a system of correlations than when isolated fromother correlations.\medskip

high value systematicity: an attribute that predicts the value of one otherattribute also predicts the values of several other attributes.

Page 50: 1998: Annotated Bibliography - Princeton Universityadele/papers/Papers/Anbib98…  · Web view1998: Annotated Bibliography Adele Goldberg *!*Afarli, Tor. 1987. Non-Subject Pro-Drop

``human categorization is geared toward learning categories with high value systematicitybecause such categories allow many inferences and are thus very useful."\medskip

Most attributes in the other (MATRIX) condition varied independently and allowedfew predictions.They suggest that this condition is more like the organization of the Englishrelational terms, since, e.g., verbs and prepositions can vary independentlyfrom each other.\medskip

Participants rated test events as to how well they matched learning events.

\medskip

Events were things like: circle approaching a square, exploding and moving away.

In correlated condition: event instances in learning were all the same (differed only in thatsome aspect was obscured in some way, by shading or whatever).But ONE correlation was chosen to be each participant's target rule:either agent path -- environment, manner of motion---environment or agent path--manner of motion.

In matrix condition: event instances correlated on three independent pairs of attributes.One of these was chosen as the target.

120 learning events.

54 test events (18 test events tested for knowledge of the target rule)

quotethere are two kinds of creatures on another planet, one of which always moved to the other and changes its appearance.You should learn about the kinds of events that happen on this plant, and yo uwill be tested

Page 51: 1998: Annotated Bibliography - Princeton Universityadele/papers/Papers/Anbib98…  · Web view1998: Annotated Bibliography Adele Goldberg *!*Afarli, Tor. 1987. Non-Subject Pro-Drop

later on how well you can differentiate events that can take place on the planetfrom those that can't.

After the 120 learning events, the 6 example test events were presented.  Then the particpantwas asked to rate each of the 54 test events as to ``how well it fits in with" the learningevents.  5 point scale.quote

Results: structured categories were learned easier.Could have to do with competition in Matrix condition from other correlations.

\medskip

\noindent! Experiment 2: 

Structured condition: all four attributes were correlated for each participant.

\medskip

Other condition: Only the two target rule attributes were correlated; all otherattributes varied randomly: there were no other independent correlationsto potentially draw attention away from target rule attributes.[would seem to still have to learn to IGNORE all the random attributes]

\medskip

structured condition still did better.

\bigskip[what was the learning phase like?  was feedback given, and both instancesand non-instances present?  Or does ``unsupervised" implyall 120 instances were assumed to be instances, with varying correlationsdepending on condition?

Page 52: 1998: Annotated Bibliography - Princeton Universityadele/papers/Papers/Anbib98…  · Web view1998: Annotated Bibliography Adele Goldberg *!*Afarli, Tor. 1987. Non-Subject Pro-Drop

What was testing like?  was feedback given?  must have been or how couldsubjects pick out the ``target correlation"?]

*!*Kim, Soowon and Joan Maling.  Resultatives: The View fromAbroad. 1999--submitted to Language

They argue that Resultative Formation in English is a lexical rulethat radically restructures the theta-grid of the matrix verb, turning itinto a monotransitive verb that subcategorizes for a single object.

They note that resultatives in Korean, tamil Warlpiri and Finnishare much less syntactically restricted: do not involve ResultativeFormation.

"fake" objects should probably be attributed to Simpson 1983.This has been used to argue that unergative verbs are accusative case-assigners (are underlyingly transitive).

[In Korean, the "fake" object appears in the NOM, not ACC case;In Korean, all verb types can occur with resultatives (but not verbs ofinherently directed motion), and a resultativephrase can be predicated of any argument NP.]

You argue against the idea that the resultative is an adjunct subordinate clause, but isn't it still possible that it's a different kindof adjunct, more parallel to the English depictive?  The focus domain that you describe in Korean holds of the English depictive as well:

(1) She didn't serve him the soup hot. (it was cold.)

I felt your discussion of the constraint I proposed was taken out ofcontext a little bit.  The fact that resultatives cannot appear with (goal) ditransitives was just one descriptive generalization among several. The real constraint was that only one (real or metaphorical) path could be be predicated per clause.  If you're right that examples like (2)

(2) The chef cooked me my steak well-done.

involve benefactives and not goals, then the one path constraint wouldn'tbe violated.  I hadn't thought about examples like those, but the

Page 53: 1998: Annotated Bibliography - Princeton Universityadele/papers/Papers/Anbib98…  · Web view1998: Annotated Bibliography Adele Goldberg *!*Afarli, Tor. 1987. Non-Subject Pro-Drop

account would predict that they should be ok. (the steak "goes from" being raw to being cooked. No other path is entailed).

The unique path constraint also explains why your examples in (34) areout:

(34) *they deprived him poor of his inheritance.

since the inheritance is moving away from him AND he is "moving to"the state of being poor.  Stating the constraint this way allows us toalso account for the fact that resultatives don't occur with directedmotion verbs used literally, and also for the fact that they DO occur with directed motion verbs as long as a single path is involved: 

(3) She drove him crazy.

Anyway, this is a relatively minor point.  I enjoyed your paper a lot.I think people have often too quickly assumed constructions must bethe "same" across different languages whenever they havea similar function.  But clearly  the same functioncan be expressed by different syntactic constructions cross-linguistically.  Have you thought at all about Chinese?  They also allow subject-controlresulatives.  I wonder if they pattern like Korean more generally.

All the best,Adele

*!*Labendz, Jacob. 1998.  Using standard American English manner ofSpeaking and Sound emission verbs as speech verbs.  Brandeis UniversitySenior Essay.

distinguishes these two classes that are often conflated or confused.

*!*Lambrecht, Knud.  2000.  When subjects behave like objectsAn analysis of hte merging of S and O in Sentence-Focus Constructionsacross languages.  In Studies in Language.

*!*Lassaline, Mary E. and Gregory L. Murphy.  INduction and categorycoherence.  Psychonomic Bulletin and REview 1996, 3(1), 95-99.

Page 54: 1998: Annotated Bibliography - Princeton Universityadele/papers/Papers/Anbib98…  · Web view1998: Annotated Bibliography Adele Goldberg *!*Afarli, Tor. 1987. Non-Subject Pro-Drop

Medin et al. (1987) found that subjects tend to sort on a single dimension.But single-dimension concepts have been widely acknowledged as being uninformative.

Two exceptions to unidimensional sorting:1) when there is conceptual knowledge that makes interproperty relationshipssalient or when 2) causal relationships are explicitly given.

Present experiment:  when subjects must actually make inductive inferences, they aremore likely to form family resemblences categories:

 Given that this animal has a short tail, what kidn of teeth would you expect it to have?  

These subjects were three times more likely to produce family resemblance sorts.

*!*Levin, Beth. 1999. Objecthood: An event structure perspective. In CSL 35 Vol 1: the Main Session.

Event templates $\approx$ argument structure constructions: "structural" aspects ofverb meaning.

\noindentCore meaning: what is idiosyncratic to particular verbs

\medskip

Core transitive verbs (! CTV): have transitive event template: cross-linguistically always transitive. (patient/theme objects): cut, destroy, kill, break, open.

\medskip

! NCTV: second argument is contributed only by constant: event structure templateis intransitive. ( praise, sweep, touch, own, imagine, study, praise, greet, fight, meet, visit, follow):

Page 55: 1998: Annotated Bibliography - Princeton Universityadele/papers/Papers/Anbib98…  · Web view1998: Annotated Bibliography Adele Goldberg *!*Afarli, Tor. 1987. Non-Subject Pro-Drop

Argument status varies cross linguistically and within a language: DO, oblique, dative. (linking rules for constant participants exist but are not universal).

Catford (1975:44) on N.W. Caucasian: 2nd argument of beat, bite, expect, harm, help, kiss, push, stab, read, wait for, look at are in nominative case. Bloom (1988): hungarian verbs with dative complements: answer, congratulate, greet

Nichols (1975:346-347):Russian: 2nd complement is marked as instrument: rule, govern, commad, manage, master, conduct

[Why should this be?] % answer: nominative case used for objects that are prototypically% human; Hungarian dative indicates conduit metaphor]

\medskip

Some good discussion of why event structure templates should not be defined aspectually. 

\medskip

! NCTV: simple event structure: one structural participant.Levin cites also Van Valin 1990; 1993 who actually argues that ACTIVITY VERBS' second argument isnot a macrorole [this is different than what Levin is describing]

\smallskip

! CTV: complex event structure: two subevents, two structural participants

Levin: evidence that 1 and 2 argument activity and

Page 56: 1998: Annotated Bibliography - Princeton Universityadele/papers/Papers/Anbib98…  · Web view1998: Annotated Bibliography Adele Goldberg *!*Afarli, Tor. 1987. Non-Subject Pro-Drop

semelfactive verbs ( beep, blink, tap, cough, sneeze, etc) (=  NCTV?)  should be treated alike:

the actor of both is subject both allow fake  reflexives and "way" construction both allow out- prefixation (Haj Ross, p. c.) ``The distribution of data supports an event structure-based generalization: theseverbs are found in the reflexive construction because they share a simple eventstructure, independent of whether they select one argument or two."

\medskip

RH\&L(1998): basic claim is that argument realization reflects event complexity:

quote!The Structure participant condition: there must be an argument XP in the syntax for each structure participant in the event structure.quoteCite again claim that verbs of surface contact and motion show more argumentexpression options than verbs of change of state (citing my paper in footnote 12):

Verbs of surface contact allow unspecified objects without generic or repetitivecontexts, change of state verbs do not:

BUT:  *Kelly caressed/touched/hit this morning.

Two argument ! statives are also descrived as simple event structure NCTVs, althoughthey are not claimed to have the same event structure template as activitiesand semelfactives.

``This study suggests that the notion of semantic role label is well-defined forstructure arguments only." [similar point in Constructions]

Page 57: 1998: Annotated Bibliography - Princeton Universityadele/papers/Papers/Anbib98…  · Web view1998: Annotated Bibliography Adele Goldberg *!*Afarli, Tor. 1987. Non-Subject Pro-Drop

*!*Liversedge, Simon P., Holy P. Branigan, Martin J. Pickering andRoger P.G. van Gompel.  Processing Arguments and Adjuncts in Isolationand in context: the Case of By-Phrase Ambiguitities in Passives.1998.  Journal of Experimental Psychology: Learning Memory and Cognition.Vol 24  2: 461-475.

Exp 1: The shrubs were planted by the apprentice/greenhouse that morning

Without context, subjects showed a preference to interpret "by" phraseagentively.

\medskip

Exp2: The head gardener decided who should/where to plant the shrubs.expect either agent or location.  Argued that verb and who, where activate thematicroles, and that phrases that express activated roles are easy to process.

*!*Marchman, Virginia A. and Elizabeth Bates. 1994.  Continuityin lexical and morphological development: a test of the critical masshypothesis.   Journal of Child Language 21, 339-366.

*!*Matsumoto, Yo. ms? date.  Abstract Motion and English and JapaneseVerbs.  Tokyo Christian University.

Simple manner of motion verbs cannot take a ni-marked goal argument, whilecomplex predicates "e.g., run-go" can:

*Taro wa eki ni hashit-ta. Taro-top station-goal run-past.

Taro wa eki-ni hashitte itta. Taro-top station-goal run went.Taro ran to the station

But these verbs CAN appear with other motion-related PPs:

Page 58: 1998: Annotated Bibliography - Princeton Universityadele/papers/Papers/Anbib98…  · Web view1998: Annotated Bibliography Adele Goldberg *!*Afarli, Tor. 1987. Non-Subject Pro-Drop

Taro wa  eki made /eki no hoo ni hashit-ta. Taro-top station as far as/station toward  run-past.Taro ran as far as the station/toward the station.

On Fictive ("ABSTRACT") motion:

Manner condition:  no property of the manner of motion can be expressed

Path condition:  Some property of the path of motion must be expressed.

\# The read walks through the center of the park \# Sono michi wa kooen no mannaka o ! anite iru.

Shows this nicely with climb which is ok only in its direction sense,not in its manner (clambering sense):

\eenumsentence From there on the trail climbs (up) steeply. \# From there on the trail climbs down steeply.

run is ok in both languages (and Chinese also), but in both languages runis used for variety of manners and primarly means motion or action.

The manner presupposes the presence of a moving entity, but no actual movingentity can be expressed.

Abstract motion is used to describe the configuration or extent of objectsthat extend in space, or over which the motion of thefocus of attention is evoked.  Thus the configuration or extent isessential to what is being conveyed and cannot be omitted:\eenumsentence *The road ran. John ran.

*!*McKoon, Gail and Talke Macfarland. in press. Language. (2000).Externally and Internally Caused Change of state verbs.

Internal cause : blister, bloom, corrode, deteriorate, erode, ferment, germinate, rot, rust, sprout,stagnate, swell, wilt, wither.

Page 59: 1998: Annotated Bibliography - Princeton Universityadele/papers/Papers/Anbib98…  · Web view1998: Annotated Bibliography Adele Goldberg *!*Afarli, Tor. 1987. Non-Subject Pro-Drop

External cause: abate, atrophy, awake, crumble, dissipate, explode, fade,fossilize, fray, redden, shrivel, splinter, thaw, vibrate.

Equated for overall frequency and for probability of occuring transitively.

Corpus difference: when internal cause verbs appear in transitive construction,they USUALLY have a natural force as subject.

Experimental difference: external cause verbs take longer to judgegrammatical in both intransitive and transitive sentences.

Authors attribute the difference to difference in complexity in their underlyingsemantic representation with only external cause verbs having a second"structural" argument.

They note results are consistent with ConGram since I also posit one senseunderlying uses in both transitive and intransitive sentences.(they suggest I could specifically say that for internal causation change of stateverbs, the agent argument is not profiled, and for external causation it is;I may have said something like that to them.  But really, I think I'dprefer to say that internal cause verbs have only one profiled argument (anyother would be contributed by the construction), while external causeverbs have two arguments, again, though, only one is profiled (in the case ofexternal cause verbs that can appear intransitively, in any case).)

But they suggest that their results argue in favor of lexical decomposition.

[I wonder whether the idea that one group is more likely to appear withvolitional agent and volitional agents can be involved in more complexchains of causation might be relavent here...]

*!* Medin, Douglas L., William D. Wattenmaker, and Sarah E. Hampson.  Family

Page 60: 1998: Annotated Bibliography - Princeton Universityadele/papers/Papers/Anbib98…  · Web view1998: Annotated Bibliography Adele Goldberg *!*Afarli, Tor. 1987. Non-Subject Pro-Drop

Resemblance, Conceptual Cohesiveness, and Category Construction. Cognitive Psychology 19 No. 2. 242-280.

Participants in their first 4+ experiments all sorted on the basis ofa single dimension and not according to family resemblances.(sorting bees with stripes, varying length tails, ears, patterns, headshapes into two piles).

Second set of studies considered the idea that interproperty relationshipsinstead of independent features serve to organize categories (propertiesare related by a theme or theory: e.g. earache and dizziness; or ability tofly).People DO sort by correlated properties if these are recognized, resulting in family resemblance category, especially when correlated propertiesare perceived as causally connected or when they are particularly salient.

`` the apparent use of family resemblance rules may be masking he use of a deeper rpinciple that some core factor or cause is present whcihprobabilisticallyleads to surface structure (family resemblance) features. " pg 273 (italicsin original).  ("core" properties are believed to give rise to "identification" properties)

*!*Douglas L. Medin and Stephen M. Edelson.  1988.  Problem Structureand the Use of Base Rate Information from Experience.  Journal of ExperimentalPsychology 117 No. 1. pg 68-85.

Key words: category, frequency

Used disease diagnosis as example. 

Most prototype models are insensitive to base-rate information.(size of category should not matter)

Exemplar based models predict appropriate use of base rate info.

Page 61: 1998: Annotated Bibliography - Princeton Universityadele/papers/Papers/Anbib98…  · Web view1998: Annotated Bibliography Adele Goldberg *!*Afarli, Tor. 1987. Non-Subject Pro-Drop

Tversky and Kahneman (1980) showed that in pencil and paper tasks, subjectsoften ignore base rate info.

But in studies that involve base rate info gained through experience, subjectshave been shown to use it.

Medin and Edelson argue that situation is more complicated:

a) retrieval failure associated with changes in context(only one of the symptoms instead of both are retrieved)

b) competition among predictors of a particular outcome.

Abstract representation of typical experiment

tabularlllrelative frequency in learning & Symptoms & Disease 3 & a,b & 1 1 & a, c & 2 3 & d,e & 3 1 & d,f & 4 3 & g,h & 5 1 & g,i & 6tabular

\noindentImperfect predictors: a,d,g

 Base rate would be used to predict disease 1 over 2 if a were given.Subjects DID do this.

\noindent! Conflicting tests: e.g., b\&c or b \& fbase rate info would predict that disease 1 should be predicted over 2 if b \& c weregiven, but Subjects DO THE OPPOSITE, and predict 2.

\noindentCombined test: e.g. a,b,c

Page 62: 1998: Annotated Bibliography - Princeton Universityadele/papers/Papers/Anbib98…  · Web view1998: Annotated Bibliography Adele Goldberg *!*Afarli, Tor. 1987. Non-Subject Pro-Drop

Base rate info was uses appropriately.

Interpretation: symptoms compete for attention on the basis of their ability to correctlypredict the associated disease.  Symptom a is a better predictor of disease 1 than ofdisease 2 (because of the higher frequency of disease 1).

a,b representational units suffer much more than the ac representational unit from the absenceof symptom a in the probe.  that's why the combined test doesn't show the pattern: a and bare both present.

That is, a \& b are more unitized and form a better integrated pattern than a and c.

``Each time a representational unit is retrieved and leads to a correct classification,properties (symptoms) in common between the probe and the retrieved unit are rehearsedor strengthened." pg 79ZZ

In other studies: ``causal linkages [bewteen symptoms] lead participants to attend toboth symptoms associated with a disease." pg 81.

*!*Murphy, Gregory L. and Hiram H. Brownell.  1985.  Category differentiationin object recognition: typicality constraints on the basic category advantage.Journal of Experimental Psychology: Learning Memory and Cognition. 11:1 70-84.

Atypical subordinate categories (e.g. racing car, boxing glove, electric knife are highly differentiatedand are therefore responded to as fast as basic categories in objectrecognition. 

category specificity: disfavores superordinates

Page 63: 1998: Annotated Bibliography - Princeton Universityadele/papers/Papers/Anbib98…  · Web view1998: Annotated Bibliography Adele Goldberg *!*Afarli, Tor. 1987. Non-Subject Pro-Drop

Differentiation is a continuum, not discrete basic/non-basic.

Invokes Grice:

*!*Naigles, Letitia R. and Erika Hoff-Ginsberg. 1998.  Why are some verbs learnedbefore other verbs? Effects of input frequency and structure on children's early verb use. JCL. 25 95-120.

Three properties of the input were relevant:  total frequency, final position frequency,and the diversity of syntactic environments in which the verbs appear (think aboutthis last--they use this to argue for syntactic bootstrapping).

*!*Ninio, Anat. 1999.  Pathbreaking verbs in syntactic development and the question of prototypical transitivity.  In JCL 26: 619-653.

"the more verbs children already know to combine in a certain pattern, the faster they learn new ones" pg 619.

"The 'pathbreaking verbs' that begin the acquisition of a novel syntactic rule tend to begeneric verbs expressing the relevant combinatorial property in a relatively purefashion: the same verbs that children first combine with direct objects,are typical grammaticalized markers of transitivity in many languages." pg 619 (abstract).

Following Tomasello, "get-it" is considered a single word.  Immediate imitationswere ignored.

The first verb is the only verb for a period of on average 42.6 days (ranging from17-104 days).

The speed-up of acquisition occurs both in VO and SVO patterns (independently--thatis, with different verbs): is "one of the central pieces of evidence suggesting

Page 64: 1998: Annotated Bibliography - Princeton Universityadele/papers/Papers/Anbib98…  · Web view1998: Annotated Bibliography Adele Goldberg *!*Afarli, Tor. 1987. Non-Subject Pro-Drop

that the speed-up is specificaly tied to the number of previously producedverbs in the same kind of combination" pg 629.

There is no indication that the later SVO pattern is acquired particularlyfaster than the VO pattern: little evidence of transfer of knowledge.  (by thetime the first SVO was produced, child uses on ave 25 different verbs in VOpattern) pg 630.

The most facilitation is provided by the first verb: most of the facilitationis provided by the first and second verbs.  even the third verb has a verysmall additional facilitating effect.

First verbs:  want, get, give, take, bring, find, make/do, see, hear, eat, drink": almost always expressed the child's desire for an object she didn't have.

First verbs in SVO pattern is similar: want, make/do are used most frequently.

But get, give, bring, find (verbs of obtaining) do not occur as SVO verbs[BECAUSE the subject would be I and is omitted?!]: only 25.9% of the firsttwo utterances were requests, the rest were descriptive statements." The most frequent adn typical SVO utterance was a description of a sceneto do with the creation or consumption of objects." pg 633E.g., put used to describe the mother's giving the child a blanket; ride used to describe a puttpet 'possessing' a horse by mounting it [please!].

Individual children don't initiate their VO and SVO patterns with the sametwo verbs in the same order.  SVO emerges months after VO.

Since children don't start VO and SVO with the same verbs in the same order,overall frequency of the verbs cannot account for the data. pg 635.

[how reliable are the details really?? Chidlren were recorded at home

Page 65: 1998: Annotated Bibliography - Princeton Universityadele/papers/Papers/Anbib98…  · Web view1998: Annotated Bibliography Adele Goldberg *!*Afarli, Tor. 1987. Non-Subject Pro-Drop

for30 minutes once a week--augmented by parental report--but we know parentsdon't remember v's as well as n's.]

Travis [Tomasello's] is cited for NOT using WANT despite it's likely highfrequency, until 6 months fater she started VO combinations. "Apparently,high frequency in the input is not a sufficient condition" pg 636.

"High frequency may however be a necessary condition" pg 636.

Transitivity BS follows.

Equivocal english data was Travis' "Maria MADE this duck" "Big Bird RIDEhorsie" with only 10 days in between.

*!*Nosofsky, Robert. 1988.  Similarity, Frequency adn Category Representations. Journal of Experimental Psychology: Learning, Memory and Cognition.Vol 14 No. 1:54-65.

`` Classification accuracy and typcality ratings increased for exemplars presented with high frequency and for members of the target category that weresimilar to the high frequency exemplars.  Typicality deceased for membersof the contrast category that were similar to the high frequency exemplars." (from abstract)

Uses Medin and Schaffer (1978) context model : people represent categoriesby storing individual exemplars in memory and computing classificationsbased on similarity to stored exemplars.

N claims frequency doesn't play a role in original context model and adds it to the system.

Page 66: 1998: Annotated Bibliography - Princeton Universityadele/papers/Papers/Anbib98…  · Web view1998: Annotated Bibliography Adele Goldberg *!*Afarli, Tor. 1987. Non-Subject Pro-Drop

*!*Narasimhan, Bhuvaneswari. 1998.  The Encoding of Complex Eventsin Hindi and English.  BU dissertation.

>From Chapter 6:

She argues that Talmy's account of different lexicalization patterns isn'texplanatory, since Hindi (and Spanish, actually, Slobin 1996) does have manner of motionverbs, but still doesn't allow them to appear with directional prepositions.

The manner of motion verbs have similar entailments in the two langauges(discussed in chapter 3)

She observes that all languages (including Eng) seem to allow the adjunctconstructions: He went out of the room, hobbling, so that these shoulde considered the unmarked case.

She argues that the syntactic frames themselves are associated with fixed, complex meanings, such as causative change of state,directed motion, creation or communication: Extralexical Hypothesis.

The Lexical Mapping Principle is retained as the unmarked grammtical optionacross languages (pg 160) (cf. also Jackendoff 1990, 1997).

tabularlllllDirected Motion & directed motion & non-directed & non-translational & non-motion +goal & -goal & translational motion & motion &move & descend & dance & flutter & squeak come, go & sneak & toddle/mince & squirm & rattle reach & shift & hobble & wriggle & roar return & run, swim & drift && melt tabular

koo- marked dative NP as a diagnostic of a goal argument in Hindi.

Verbs of manner of motion can occur in directed motion

Page 67: 1998: Annotated Bibliography - Princeton Universityadele/papers/Papers/Anbib98…  · Web view1998: Annotated Bibliography Adele Goldberg *!*Afarli, Tor. 1987. Non-Subject Pro-Drop

constructions withatelic directionals (Spanish, Aske 1989; Italian, Napoli 1992, Slobin and Hoiting 1994 for a number of other languages, BN for Hindi).

BN: these are verbs that have a directed motion component in their conceptualstructure, but no goal argument.  None of them can occur with koo.The verbs entail an indefinite change in location: translational motion, butnot an endpoint.

Judgements vary on whether the class of non-directed translational motioncan occur with directional phrases.  Some coercion is possible.

*!*Osherson, Daniel N., Ormond Wilkie, Edward E Smith, Alejandro Lopez,and Eldar SHafir.  1990.  Category Based Induction.  PsychologicalReview 97 (2) 185-200.

The strength of a cateogrical argument increases with (a) the degree towhich the premise categories are similar to teh conclusion categoryand (b) the degee to which the premise categories are similar to membersof hte lowest level category that includes both teh premise and theconclusion categories.

Lions use X.Giraffes use X.-----------Rabbits use X

Lions use XTigers use X-----------Rabbits use X

*!*Pawley, Andrew and Frances Hodgetts Snyder. 1983. Two puzzles forlinguistic theory. in J. Richards and R. Smith (eds) Languageand Communication NY: Longmans.

Two puzzles are native-like selection and nativelike fluency.

Page 68: 1998: Annotated Bibliography - Princeton Universityadele/papers/Papers/Anbib98…  · Web view1998: Annotated Bibliography Adele Goldberg *!*Afarli, Tor. 1987. Non-Subject Pro-Drop

Why don't we say:

"The brothers of my parents were four. Their offering to soldier inlands elsewhere in the army of our country had occurred.  There is nota time when my remembering that Christmas will not take place, because o fhte coming ofthe brothers of my parents to our house, having put on their bodies the clothes of the army.  the eating of Christmas dinner by them i ourcompany was to happen and above that thin in importance the buying of the food by themwas to occur, a thing that was indeed unusual and indeed good."

*!*Pederson, Eric. ms. 1998.Event Realization in Tamil.University of Oregon.

\eenumsentence The brahmin broek the coconut but the coconut didn't break (Tam) He bought a book but iddn't get it. (Man) I loosened the knot but it was still tight (Jap) I burned the garbage, but it didn't burn (Tam)Similar phenomenon is Mandarin, Japanese.

Different pragmatics? (semantic entailments in tamil being simply weaker?): appealsto a vague difference for which we have no explanation. : Also pragmatics isgeneraly assumed to be universal.  Also, there are a few examples like thesein English, too ( strangle).

English: has a number of constructions which deny final realization : kick at, almost.

Different verbal semantics?   different aktionsart patterns?

In Mandarin, defeasibility is impossible when the perfective le is appended to thefirst verb.

In Tamil, almost any aux marking entails realization.

Page 69: 1998: Annotated Bibliography - Princeton Universityadele/papers/Papers/Anbib98…  · Web view1998: Annotated Bibliography Adele Goldberg *!*Afarli, Tor. 1987. Non-Subject Pro-Drop

Activity verbs watch do entail seeing

``sleep" in Tamil doesn't imply realization, but "fall asleep" does

[Parallel to instruments as verbs which do not entail the instrument was used?hammer with a stick, staple without a stapler...]

! why the Tamil simple verbs should have defeasible realizations, he asks.

Tamil verbs do not typically occur as simple verbs alone.  USually there is an adverbial participle indicating a sequence or are simultaneous.  If a verb is in adverbial partipcle form and is followed by another verb, realization IS entailed.

Adverbial participles may be perfective markers, but also may be reflexives.

Therefore EP suggests an Adverbial Participle construction

``Since Tamil verbs most typically occur in a constructoin which definitively entails realization of the first verb, when the verb is used without such a construction there is a temptation to assume that it might not entail realization."

[This raises another question, what is the analysis of the Adverbial Participle construction,and why should it have the entailment of realization no matter what the semantics ofthe adverbial participle?]

[Missed a few steps: principle of contrast; Horn's pragmatic implicature to distinguishsimple and periphrastic causatives]

A verb which ``really means" a realized event is useable for reference to a non-realizedevent by a process of metonymy: whole refers to part of the whole.

``If denial of realization is the result of a metonymic process rather than a baregrammatical fact, we shoud expect it in all languages...Lngs like English would simplyrequire much more extreme pragmatic motivation... He threw the ball to her but

Page 70: 1998: Annotated Bibliography - Princeton Universityadele/papers/Papers/Anbib98…  · Web view1998: Annotated Bibliography Adele Goldberg *!*Afarli, Tor. 1987. Non-Subject Pro-Drop

didn't actually let go of the ball at the last minute. [supposed to be ok!]." pg 10

*!*Pine, Julian M. and Elena V. M. Lieven.  Reanalysing rote-learned phrases:individual differences i the transition to multi=word speech. 1993. JCL 20.551-571.

Object-oriented children (combine single words) vs noun-preference (breaking downphrases into slots).  A strategy involving the breaking down ofunanlysed phraes may be used by all children, to varying degrees.

*!*Pizer, Karin.  1994. Perception Verb Complementation.  A Construction-basedAccount.  CLS 30.

*!*Posner, Michael I and Steven W Keele. 1968.  On the Genesisof Abstract Ideas. Journal of Experimental Psychology.  Vol 77. No 3 Part 1.353-363.

The prototype of learned dot patterns is more easily classfieid than control patterns which are alsowithin the learned cateogry.  As the variability among the memorized patternsincreases, so does the abiltiy of Ss to classify highly distorted new instances.That is, prototypes are abstracted with high efficiency.  The experiments areinconclusive about whether the prototype is abstracted during learning orwhether it is computed quickly upon presentation.

Hinsley (1963): pretraining on a prototype is superior to pretraining on a peripheralpattern.  I.e. knowing the schema can aid later learning.o

Experiment I:  4 prototypes, two groups:

One group had small distortions from the prototype

Other group had large distortions.

Page 71: 1998: Annotated Bibliography - Princeton Universityadele/papers/Papers/Anbib98…  · Web view1998: Annotated Bibliography Adele Goldberg *!*Afarli, Tor. 1987. Non-Subject Pro-Drop

Tested on list of patterns which were more highly distorted than those in eitherof the two original lists.

If a clearly definied schema was of primary importance, a small distortion group should show better tranfer.

If variability is more important, the larger distortion group should show bettertransfer:  this is what occured.

Demonstrates that instances were stored since it should be easier torepresent the prototype only given smaller distortion.

Experiment 3:demonstrates the schema pattern is, on the whole,  as well recognized as the exemplars from whichit is abstracted.

RT to teh schema is longer than to the old distortions on the first presentation of hte transfer list may indicate that the subjects calcuate on the basis of storedinformation OR it could mean the abstracted information but it is notas clea or completely definied as info about individual exemplars.

*!*Rapoport, Tova. 1999.  Structure, Aspect and the Predicate.  Language 75 4: 653-677.

Data:

Jones fried the potatoes naked.Jones cut the bread drunk.

Johns fried the potatoes raw.Jones cut the bread hot.

Jones-i phoned Smith sad-i.*Jones phoned Smith-i sad-i.

The host of a depictive predicate must be a subject in AS structure (not necessarily in surface structure).

[I'm not sure I understand how this is supposed to work]

Page 72: 1998: Annotated Bibliography - Princeton Universityadele/papers/Papers/Anbib98…  · Web view1998: Annotated Bibliography Adele Goldberg *!*Afarli, Tor. 1987. Non-Subject Pro-Drop

******

AS focus is the foregrounding or emphasis of a particular structure orpart of a structure with the consequent backgrounding or deemphasis of otherparts of that structure. (pg 659; see Erteschik-Shir and Rapoport 1999 for discussion).

In "Jones ran to school" As focus is possible only on the endpoint of the action.(diagnostic: can appear with "in an hour")

In "Jones ran", the focus is on the activity itself.(diagnostic: can appear with "for an hour")

Accomplishments allow either focus or a focus on the initial state: Jones painted the picture for an hour/in an hour. Jones ran to the store for three minutes/in three minutes.

Different AS structures force different AS foci, which may or may not be compatible with the focus forced by different modifiers.

*!*Rehder, bob and Brian H Ross. Abstract Coherent Categories.  2000-ms.UIUC

categories based on how the world works, not just on how it appears: abstractcoherent categories.

Abstract in that they are independent on any fixed set o features or stimulusdimensions.

Knowledge serves to relate or link features of a category together.

Certain categories are coherent because their features "go togehter" based ontheoretical or causal knowledge. (Murphy and Medin 1985).

Murphy and Allopenna (1994): novel coherent categories with features such as"drives on galciers, made in Norway, heavily insulated" were differentiated

Page 73: 1998: Annotated Bibliography - Princeton Universityadele/papers/Papers/Anbib98…  · Web view1998: Annotated Bibliography Adele Goldberg *!*Afarli, Tor. 1987. Non-Subject Pro-Drop

from members of another category like "drives i the jungle" "made in Africa""lightly insulated" faster than control condition that learned todiscriminate more random features from the same domain.

The accelerated learning (and better transfer performance) is attributed to the relations among featuressupplied by prior knowledge (cf. also Pazzani 1991; Wattenmaker et al 1986; Murphy and Wisniewski 1989).

Wattenmaker et al 1986: participants had to identify a "house painter" categorythat consisted of either indoor or outdoor painters, with no shared features.

Present paper investigates the idea that features of some categories are not preset, even probabalistically.  Instead category is defined by a system of relations and membershipis determined by the extent to which a particular combination of featuressatisfies the system of relations. pg 6 They mention that many superordinatecategories are  examples of this type of category. [sounds like language!]

Morkels:  "machines that gather a certain type of pollution, in an area wheresuch pollution is likely to be found, with a device suitable for collection of that type of pollution" pg 6

Incohrent morkels combine features of coherent morkels but in inconsistent ways.

Real abstract, coherent categories include things like "political revolution""family" "corporation" "academic department" "murder" "Marriage" etc.

*!*Rice, Sally. 1988.  Unlikely Lexical Entries.  BLS 14

`` The notion of default interpretation is very important to thephenomenon of object omission'' pg 203

Page 74: 1998: Annotated Bibliography - Princeton Universityadele/papers/Papers/Anbib98…  · Web view1998: Annotated Bibliography Adele Goldberg *!*Afarli, Tor. 1987. Non-Subject Pro-Drop

Omitted objects are restricted to complements with a low degree of semantic independenceform the verb.  Default interpretation is the verb's prototypical, basic levelargument.

Verbs that encode manner tend to resist object omission:Celia ate/*nibbled/*bit/*chew.

Of cousre, some of hte predicates are acceptable minus objects when occuring innonfinite verb forms or in the imperative (John always nibbles before dinner,Memorizing is how she studies best, Don't guzzle!).

If the omitted object refers to some whole entity, it is more likely to be leftout than if it refers to the part of some entity.

There is some back and forth here:

John squinted (his eyes).

``Because the sentences...readily evoke general semantic frames or scenarios, the particularobject is fairly unimportant as hte pragmatic focus is on the activity itself":

\eenumsentence% Hemingway ate, drank and smoked too much. Martha cooked and cleaned while Mary entertained. Billy Jo washed adn Bobby Jo dried.% John finally married.% Horowitz practises daily. Scot hammers and saws like a pro.% Bill always interupts. He paints, she pots; he sculpts, she draws. Those who can, do, those who can't teach. 31-39 pg 206

``a particular semantic frame tends to evoke objectless verbs:\eenumsentence Typical restaurant script: The man entered, he ordered,he ate, he paid, he left.

Page 75: 1998: Annotated Bibliography - Princeton Universityadele/papers/Papers/Anbib98…  · Web view1998: Annotated Bibliography Adele Goldberg *!*Afarli, Tor. 1987. Non-Subject Pro-Drop

``Collectively the individual objectless clauses are fine, especially when strungtogether, because the identity of each of hte omitted objects is easilyinduceed from the context of the larger script or fromassociations enegnderd by other lexical items in the string." pg 206

*!*Ritter, Elizabeth and Sarah Rosen. 1996. In 1998.  TheProjection of Arguments. Ed by M.Butt and W. Geuder.  CSLI Publications

"syntactic structure plays a role in determining the realization of arguments and the interpretation of the sentence" pg 136.

Aspectual notions are relaized through aspectual projections in the syntax (Borer 1994,Guilfoyle 1995 Travis 1994 are cited).

Borer (1994): causer role and event measure (delimitation or endpoint of an event)are assigned by Functional Projections (FPs).

Erteschik-Shir andRapoport (1995a) also propse that event structure interpretation is compositionally (syntactically) determined.

"event" or "event-structure": encoding of event information in the phrase structure of a clause. pg 137:  defn is "purely syntactic": syntactic events are necessarily delimited events (D-events) [oh, please!]

"what distinguishes predicates syntactically and semantically is the difference between Delimted-events and all others.

"Only predicates that have a syntactically encoded event structure are those that are delimited, i.e., those that have an endpoint." pg 137.

" The lexical semantics of the verb doesnot determine the event interpretation" pg 137.(may determine the number of args selected and does determine some portion of the thematic roles).

" acivity vs achievement vs acopmlishment is largely syntactic" Pg 138

Page 76: 1998: Annotated Bibliography - Princeton Universityadele/papers/Papers/Anbib98…  · Web view1998: Annotated Bibliography Adele Goldberg *!*Afarli, Tor. 1987. Non-Subject Pro-Drop

Activities

Result-oriented Verbs: lexically select a result and are lexically specified for delimitation: must map onto a D-event: initiators can be added to result verbs:

D-events: may include event role of delimiter; may include initiator event role.

Activities have no inherent event structure specification: walk does not syntacticallyselect a D-event.

*!*Rosch, Eleanor and Carolyn B. Mervis.  1975.  family Resemblances:Studies inthe Internal Structure of Categories. Cognitive Psychology 7, 573-605

Members of categories that are considered the most prototypical hsare the mostattributes in common with other members of the category and the leastattrivutes in common with other categories.:  Pototypicality is a function ofthe total cue validity of the attrivutes of items.

*!*Rosch, Eleanor, carolyn B. Mervis, Wayne D. Gray, David M. Johnsonand Penny Boyes-Braem. 1976.  Basic Objects in Natural Categories.  CognitivePsychology 8. 382-439.

Basic categories carry the most information, hve hte highest category cue validity,and are therefore the most differentiated from one another.

Have significant numbers of shared attributes

Have motor programs that are similar to one another

Have similar shapes

Page 77: 1998: Annotated Bibliography - Princeton Universityadele/papers/Papers/Anbib98…  · Web view1998: Annotated Bibliography Adele Goldberg *!*Afarli, Tor. 1987. Non-Subject Pro-Drop

Can be identified from averaged shapes of members

*!*Sadock, Jerrold M. ms-2000.  Grammatical TEnsion.  U of Chicago

Hi Jerry, THanks for sending your paper a while back.  I made some notes to myself, that I thought I'd pass along...

 Why do you pay homage to OT in the abstract and OT?  Your system seems much more like harmony theory or some earlier version of OT, since you seem to add upthe violations instead of ranking them.   You also allow for graded judgements, which OT doesn't predict.  You mention that NEARNUM seems to be increasing in weight, which is also not something that OT can explicitly deal with, as far as I know.

 You might be able to quantify some of hte factors--do you know of Bock and Cutting 1982?  They vary several factors experimentally. 

 Both the Q and R principles are functional: one to the listener and one to the hearer.  Functionalists routinely invoke both.  (There's a great discussion by Haiman about languages as diagrams in his 1985 book).

*!*Schultze-Berndt, Eva. 1998. draft.  Making sense of complex verbs:generic verb meaning and he argument structure of complex predicates in Jaminjung.  MPI

Jaminjung coverbs clearly point to the need to keep the semantic and theconstructional level separate: coverbs demonstrably have semantic participantroles which determine the argument structure of the complex verb jointlywith the verb, but they can only be integrated into a syntactic argument strcture through combination with a generic verb.

*!*Siegel, Laura. 2000.  Semantic Bootstrapping and Ergativity.

Page 78: 1998: Annotated Bibliography - Princeton Universityadele/papers/Papers/Anbib98…  · Web view1998: Annotated Bibliography Adele Goldberg *!*Afarli, Tor. 1987. Non-Subject Pro-Drop

LSA meeting, Chicago.

*!*Smith, David J and John Paul Minda.  1998.  Prototypes in the Mist: The EarlyEpochs of Category Learning.  Journal of Experimental PSychology: learning, Memoryand Cognition. Vol 24, No 6, 1411-1436.

In learning larger more differentiated categoires, the prototype model shows an early advantage that gives way slowly to exemplar based model.

[at Beckman]

*!*Sovran, Tamar. 1992.  Between Similarity and Sameness.  Journalof Pragmatics 18. 329-344.

*!*Thompson, Sandra A. and Paul J. Hopper.  Ms. to appear.Transitivity, Clause Structure and Argument STructure: Evidence from Conversation.  In Bybee and Hopper (eds) Frequency and the Emergence of Linguistic Structure. Benjamins.

They observe that most of language involves INTransitive clauses.Conversation is very low in transitivity.

*!* Tomasello, Michael. (to appear-1999?).  Do Young Children Have AdultSyntactic Competence?  Cognition.

Entrenchment works early from  3;0 or before.

Preemption and semantic subclsases work later, perhaps not until 4;6.

Tomasello suggests that laugh is not likely to be overgeneralizedbecause it is well-entrenched.  giggle is a good candidate forovergeneralization.  chortle is learned once verb classes arein place and therefore is not likely to be used transitively.

Cites Clahsen's (1996) discussion of three approaches that assumethe continuity hypothesis (that children's competence does not differsignificantly from adults):

1) differences are attributed to external factors, such as memory andprocessing limitations or pragmatic limitations.  Tomasello's studies

Page 79: 1998: Annotated Bibliography - Princeton Universityadele/papers/Papers/Anbib98…  · Web view1998: Annotated Bibliography Adele Goldberg *!*Afarli, Tor. 1987. Non-Subject Pro-Drop

with Akhtarwith novel vs known verbs argue against this, since children were

2) maturation : "is basicaly an unconstrained 'fudge factor'" pg 31

3)the grammar develps gradually throug the interaction of availableabstract knowledge and the child's learning of the lexicon.T notes that this undermines the whole point of generative accountsince much more than a minimal triggering event is required.  Also, noneof the proponents has attempted to work out precisely how the childgoes about linking up item-specific lin knowledge with UG Pg 33.

The problem of learning linking generalizations given the cross linguisticvariation and the non-obvious cues, for e.g., subject is discussed.able to produce transitives for known but not knew verbs.

See Benson, Benson and Ilson 1997: dictionary of many of the tens of thousands of idiosyncratic English collocations.

How children get from here to there: 1) learning culturally (culteralimitation not to be confused with mimicking which involves littleor no understanding), 2) making analogies, and 3) combining structures.The tool experiment is cited:  Nagell, Olguin and Tomasello 1993.

Re: analogy and structure mapping:  Gentner and Markman 1997 for review.Gentner 1993 claims that when relations are mapped acrosssituations: specific object properties are ignored, relations amongthe objects are preserved, and conected systems of relations are moreliekly to be transferred.  "This is of course exactly the kind of cognitive abilityneeded for chidlren to create a verb island schema across different arrangements ofobject participants. pg 48

*!*Tversky, Amos.  1977.  Features of Similarity.  Psychological Review 84: 4. 327-352.

Constrast Model:  similarity between objects is a linear combination of themeasure of theiir common and distinctive features.  Both semantic

Page 80: 1998: Annotated Bibliography - Princeton Universityadele/papers/Papers/Anbib98…  · Web view1998: Annotated Bibliography Adele Goldberg *!*Afarli, Tor. 1987. Non-Subject Pro-Drop

and perceptual stimuli.

Used to explain presence of asymmetric similarities, and context effects.

*!*Waxman, Sandra R. and Dana B. Markow.1995. Words as Invitations to Form Categories: Evidence from 12-13 monthold Infants. Cognitive Psychology 29, 257-302.

Waxman and Hall 1993 studied 15 month olds.

Current paper studies 12-13 month olds.

N label highlights category of objects

(studied how long children play with new kind of  toy in N, A and No Label conditions)

N label had no effect on basic level categories, presumably becuase categorizationwas already at ceiling.

N and A had same effect: both focused attention on object categories.

They stay agnostic about direction of facilitation and big linguistic questions.

Show infants with ave voc of 9 words show evidence of object bias.

See also Gleman 1988, Waxman 1991, 1994

``words serve as invitations to form categories" pg 298.

*!*Wolfe-Quintero, Kate. (University of Hawaii) 1998.  The connection between verbs and argumentstructures: Native speaker production of the double object dative. Applied Linguistics 19 225-257.

Investigates teh "strength of the connections" between the ditrans (DOD)and English dative verbs in production studies.

Contrasts Goldberg with Pinker is some mysterious ways.

Page 81: 1998: Annotated Bibliography - Princeton Universityadele/papers/Papers/Anbib98…  · Web view1998: Annotated Bibliography Adele Goldberg *!*Afarli, Tor. 1987. Non-Subject Pro-Drop

Cites studies that have shown an effect of narrow classes:

Bley-Vroman and Yoshinaga (1992): manner of speaking verbs were lesslikely to be used ditransitively.

Inagaki (1995) and Sawyer (1995) : speakers differentiated ballistic motion verbs fromcontinuous verbs, even when verbs were novel.

Trueswell et al. (1993): sentences were processed faster when the complement of the verb marched the preferred complement in production.

"in lexical processing accounts, the consensus seems to be that a lexicalrepresentation consists of a verb and a pattern of connections topossible argument structures, with varying strengths of activation encodedfor each connection.  This view is consistent with Goldberg's (1992) lexicalsemantic account." (235)

EXPERIMENT I:free production, given pictures of a family (husband, wife daughter and dog)  17 verbs were provided andeach verb was supposed to be used in only one sentence.

A majority of subjects produced give and offer in theditransitive, but none produced take, kick.

In other experiments as well, results show that certain verbs are more likelyto be used in the ditransitive than other verbs.

Concludes that hte study supports my proposal, since the relative strengthsseem to be verb particular in production.

*!*Wolff, Phillip.  Causal Chains and Verbs.  submitted toto Cognition.  University of Maryland, College Park.

Page 82: 1998: Annotated Bibliography - Princeton Universityadele/papers/Papers/Anbib98…  · Web view1998: Annotated Bibliography Adele Goldberg *!*Afarli, Tor. 1987. Non-Subject Pro-Drop

(reviewed by me: see review for exceptions).

discusses the idea that volitional agent's intention can allow morecomplex construal of single event: (Schlesinger 1989; Talmy 1976, 1988: page 14 in ms);

Leslie (1994, 1995): proposes that people have two paradigms of causation mechanical and teleological (pg 15)   (also Carey 1985; Keil 1995)

Woodward, Phillips and Spelke (1993), Speklke, Phillips and Woodward 1995; Golinkoff, Harding, Carlson and Sexton 1984): infants distinguish intentionaland mechanical causality.  7 month old infants assume that casuationbetween non-sentient beings must involve phsyical contact (principleof contact), but don't make the assumption with sentient causers pg 15.

Lexicalization is possible if either:

1) simple chains in which causer and causee make physical contact"contact criterion" been suggested by Ammon 1980; Gergely and Bever 1986; Pinker 1989; Shibatani 1976...)

2) extended chains in which they do not make physical contact, butwhich are initiated by causers capable of intention.(cites also Delancey 1983; Cary, Hilton, Keil, Morris , Spelke and Talmy 1995;Schlesinger 1989)

Pharoahs built the pyramids.The policeman stopped the car.The landlord evicted the tenants.

[BUT:  The new law evicted teh tenants]

"Causal chains that can be lexicalized by a verb are those that can be construed as single events."  (see also Croft 1991; Delancey 1983; Dechaine 1997;Goldberg 1995; Haiman 1983; Kiparsky 1997; Rappaport Hovav and Levin 1997; Shibatain 1976; Wolf and Gentner 1996).

Experiment involved marbles hitting each other or a  person hitting a marble

Page 83: 1998: Annotated Bibliography - Princeton Universityadele/papers/Papers/Anbib98…  · Web view1998: Annotated Bibliography Adele Goldberg *!*Afarli, Tor. 1987. Non-Subject Pro-Drop

References of causation studied from psych perspective given (pg 3)

*!*Zhang, Ning.  1998.  The interactions between constructoin meaning andlexical meaning.  Linguistics 36-5. 957-980.

Cross-linguistic DIFFERENCES:

Ditransitive construction in English, Chinese and Spanish: 

the default interpretation of the indirect object is goal in Eng, source in Chinese and either goal or source in Sp. 

Verb classes also differ as a result:

creation verbs: Englishconsuming verbs (and not creation verbs): Chinesecreation, consuming verbs: Spanish

In Chinese: transference verbs can be compounded with either "gei" (give)or "-zou" (leave)

[ very explicitly and in a flattering way, adopts my approach.]

[ See also Ahrens on this topic!]document