The Relationships between Verbs and Constructions* *** Adele E. Goldberg University of California, San Diego 1. Introduction This paper focuses on the types of possible semantic relationships between a verb's semantics and a construction's semantics. In order to understand this topic, it is important to make clear what I have in mind by "constructional semantics". Sec-I tion 2 briefly reviews this idea and provides three arguments in support of recog- nizing constructional semantics. Section 3 explores the range of possible semantic relationships that the verb can bear to the construction. It is suggested that the analysis sheds light on the larger issue of which actions within a given scene are the most salient. I 2. Constructional Semantics % The basic idea I will be assuming is that the simple sentence types in English are t directly correlated with one or more semantic structures. For example, =■ I I, Dkransitive: | Subj V Objl Obj2 X CAUSES Y to RECEIVE Z • 2. Caused-Motion: | Subj VObj Obi - X CAUSES Y to MOVE Z | 3. Resultative: 5 SubjVObjPred X CAUSES Y to BECOME Z 1 4. Transitive: i Subj VObj X ACTS on Y; X EXPRESSES Y This idea is at odds with the widespread view that assumes that the basic sentence patterns of a language are determined by semantic or syntactic information specified by the main verbs. Thus, the sentence pattern in (1) appears to be determined by the specifications of put. (1) Pat put the ball on the table. That is, put is a verb which requires an agent, a theme and a location, and it appears overtly with those three arguments. Below 1 argue that while {]) repre- 2 E33W PabliahiBg : sfloni Collection (ESSCObast) - priced" «u ZfZO/lVld 8:45 via tiaSARTäflTtfl THtlVERZITÄ u H&: 3£D94E t JJrtftWeJT* £v*f Ytr Ki-dong, VeMpooi, Uujol^.: lexical and S^ncact^cuü. Canetxucfeicms and thm Csnstruct-ion e-f &feaniä£ : EW)CS*flings öf thft Si-ffiiK»! IOA [i.*. XCLC3 Ifeetlna. dr. älEnzcta&rqu«, Jul; 1955 3 kceuitt: 33431ETE.££ 384 ADELE E. GOl-DBERO sents the prototypical case, sentence patterns of a language are not reliably determined by independent specifications of the main verb. 2.1, Three arguments for constructional meaning Goldberg (1995: Chapter 1) lays out several arguments for distinguishing a verb's semantics from the semantics of the construction in which the verb can appear (cf. also Goldberg 1991, Goldberg 1992a,b). Although the particular implementations vary, there appears to be a growing consensus that it is necessary to distinguish a verb's "core" semantics from the semantics of the expression when the verb appears in different argument structure arrays (Fauconnier and Turner 1994, 1995; Pinker 1994, Hovav and Levin 1996; see Leek 1995 for an alternative view). Below, I briefly outline three motivations for distinguishing between the verb's meaning and the construction's semantics. 2. L1. Implausible verb settses are avoided The first argument for constructional meaning comes from the following types of sentences: (2) a. The train screeched into the station, b. Elena sneezed the foam off the cappuccino c. Pat smiled her appreciation. If argument structure were projected exclusively from the verb's semantics, we would need a special verb sense for each of the verbs in the above expressions. In particular, we would require a special sense of screech that would mean, "Y MOVES while screeching", a special sense of sneeze, "X CAUSES Y to MOVE Z by sneezing", and a sense of smile that would mean, "X EXPRESSES Y by smiling". Such senses are intuitively implausible; one does not find unique stems to code such meanings in other languages. For example, one does not find a unique stem Mick alongside an unrelated term for "sneeze" that means, "X CAUSES Y to MOVE Z by sneezing". Similar attested examples include: (3) a. If time is money then save yourself rich at Snyder's! (found by Mark Turner) b. The people of this small town ... have been unable to pray Mrs. Smith's two little boys home again, (found by Mark Turner}) c. She tried to avoid blinking the tears onto her cheeks. (Anne Tyler, 1992, Dinner at the Homesick Restaurant, NY: Knopf) Again, the senses that would be required if one wished to project the clausal pattern exclusively from the verb's lexical semantics, are intuitively implausible. ERSCOhojT; - piisnr-td jeet< -Co hfctpa;//*mr.oia.cam.'taras-^a£—line RELATTCWHIPS BETWEEN VERBS AND CONSTRICTIONS 385 2.1.2. A generalization about construction meaning is predicted A second motivation stems from an observation made by Pinker (1989). He points out that what he terms, "the syntactically relevant aspects of verb meaning^' are like the meanings of closed class items What he has in mind by the syntactically relevant aspects of verb meaning are the skeletal meanings such as "X CAUSES Y to MOVE ZtoT, "X CAUSES Y to RECEIVE Z", "X ACTS ON Y", etc Pinker observes that the elements of meaning involved, e.g., CAUSE, MOVE, ACT, PATH etc., are like the meanings of closed class items. On a constructional account, these semantic elements combine to form constructional meanings, and constructions are closed class items. Therefore, Pinker's observation is predicted: we expect constructions to have the semantics of closed class elements because they are closed class elements. 2. J. 3. Support from language acquisition The third motivation arises from a debate in the language acquisition literature. Landau and Gieitman (1985) propose that children use the syntactic frames that a verb is heard used with in order lo determine the verb's meaning. There is in fact experimental evidence that children do pay attention to the syntactic frames and that they can use that information to narrow down the choice of possible verb meanings (Naigles 1990, 1995; Naigles ct al. 1993; Sethuraman et al. to appear). However, Pinker (1994) points out that the syntactic frame cannot help the learner directly identify what is intuitively the meaning of the verb The syntactic frame corresponds to more abstract meanings such as "X CAUSES Y TO MOVE Z", "X CAUSES Y to RECEIVE Z" He notes that the conjunction (or disjunction) of all of the more abstract meanings will not lead to inferences about the "root" meaning of the verb.1 Pinker also correctly points out that such abstract meanings will not serve to distinguish between kick and throw or between give and hand. But the proponents of syntactic bootstrapping have not claimed that this type of finer grained learning takes place solely on the basis of syntactic frames. They assume that context plays a crucial role in acquiring these distinctions. For example, hearing (4a) should not lead the child to infer that rumble has a motion component to its meaning, since rumble can be used in expressions that do not entail motion as in (4b); (4) a. The bus rumbled down the alley, (motion) b. Elena's stomach rumbled loudly, (no motion) Once constructions are recognized to be distinguishable from verb meanings, we need not assume that the constructional frame directly reflects the meaning of the verb, Instead, the constructional frame can be used by the child in a more subtle way. Instead of the child assuming that the construction' semantics directly reflects the verb meaning, the child instead uses the construction to determine the general scene that is being referred to. That is, the construction encourages the learner to ZBSZChoi-c - pcin-ced on 2/10/S;95 SK via UHrTCB£I3i_ All cat a librae* -to fcttpa =//-r»nr.ek>a».eain/teEni3-D£-ua-e 386 ADELE E. GOLDBERG focus on the particular scene under discussion (see Fisher et al. 1994; Gleitman 1994, Sethuraman et al. 1996). On this view, the verb is recognized to code some salient action within the scene. For instance, the first time a child hears the verb kick as in (5), the child need not assume that kick itself lexically codes "transfer", the meaning associated with the ditransitive construction. (5) Pat kicked Chris the ball Instead, the construction focuses the child's attention on the scene of transfer being described. The verb can be used to pick out a salient action within that scene In this case, the action of "giving" is certainly one choice of a salient action, but another salient action is that of "kicking". If the child chooses the latter meaning, then she will have correctly determined the content of the verb upon a single exposure. Assuming this proposal for how constructions aid in the acquisition of verb meaning is roughly correct, the main subject of the present paper is to determine what actions are considered salient within a given scene; or, more concretely, what is the range of possible semantic relationships that verbs can bear to constructions? 3. Relating Verb and Construction 3.1. Elaboration It is clear that the most prototypical, the most common and the most universal case is one in which the verb designates an elaboration of the meaning of the construction.1 For example, if we assume that the ditransitive construction has roughly the meaning of transfer, "X CAUSES Y to RECEIVE Z" then it is clear that verbs such as give or pass lexically code this meaning. Similarly, if we assume that the caused-motion construction has roughly the meaning, "X CAUSES Y to MOVE ZpMh", then it is clear that put lexically codes this meaning. (6) a. Hana gave/passed Martin the salt. {ditransitive; X CAUSES Y to RECEIVE Z * "give", "pass") b. Laura put the book on the table. (caused-motion; X CAUSES Y TO MOVE e> "put" It is perhaps the prototypicality of this case, that the verb codes an elaboration of the meaning of the construction, that has led so many researchers to assume that the verb must always code an elaboration of the meaning of the construction; i.e., that the verb's meaning determines, or "projects", the meaning associated with the entire sentential frame. IE5COh.o-=<; - pxizveed oa l/ao7zai0 E;45 3H ria IGJiiS.tKii'Ji UMIVHRlITi. All -asm aub^tc-fc -to h£tpj://?nr>r.e&ac:D.ccui/-e«ma--o£'-uae RELATIOSHIPS BETWEEN VERBS AND CONSTRUCTIONS 387 3.2. Means More interestingly for the present purposes are cases wherein the verb does not itself lexically designate the meaning associated with the construction, Tor example, a common pattern in English, Chinese, and Dutch is that the verb codes the means of achieving the act designated by the construction (Talmy 1985). This is the case in each of the following examples: (7) a. Amy kicked Paul the ball. b. Elena sneezed the foam off the cappuccino, (from Kathleen Ahrens) c. Ken wrote his way to feme and fortune. Kicking is the means of achieving transfer; sneezing is the means of achieving caused-motkm, writing is the means of achieving metaphorical motion Pinker (1989) discusses the following example from Talmy (1985): (8) The bottle floated into the cave. He notes that this sentence is not felicitous in the situation in which the bottle is carried into the cave in a bowl of water. It is only acceptable in the case that the floating was the means by which the bottle moved into the cave. Croft (1991) similarly observes the difference in the following two examples: (9) a. The boat sailed into, the cave b. *The boat burned into the cave. He notes that (9a) is acceptable because sailing is the means by which the boat moves into the cave; (9b) is not acceptable because the burning is not the means of effecting motion.3 On the basis of these examples, Croft proposes that "individual lexical items appear to denote only causally linked events". It is an interesting question as to whether or not Croft's generalization is actually true for lexical items; It raises a number of issues including what should count as a distinct subevent within a lexical item's designation. T do not attempt to address this question here, but see Goldberg (ms) for a comparison of the range of possible relationships between a verb and construction on the one hand, with the range of possible relationships between subevents within a single lexical item's designation on the other. In the following section the question concerning the nature of the relationship between verb meaning and constructional meaning is addressed. 3.3. Causal relations hypothesis If we leave aside the question of whether individual lexical items only denote causally linked events, we can ask a different question: are the meaning of the verb EBSCCiko=t - pxia-f-d on 2/30^2010 E:45 2M -sria lULKlRYKCJi OKIVE&SIT?... Ml ssi swk^ccb t*» liCtpfl±//www.ebaco.coBi/tenns-o-f-aiaBs 38S ADULE E. GOLDBERG and the meaning of the construction always causally related? In answer, we might propose the following hypothesis: Causal Relation Hypothesis: *^ The meaning designated by the verb and the meaning designated by the construction must be integrated via a (temporally contiguous) causal relationship. In this section it is shown that verb and constructional meanings are typically related by some kind of causal relation if the verb does not simply designate an elaboration of the constructional meaning, la the following sections, however, it is argued that the Causal Relation Hypothesis must be broadened to allow several other types of semantic relations as well. A clue that the Causal Relations hypothesis may play some role is clear from the following pairs of examples: (10) a. The car screeched out of the driveway. b. The truck rumbled down the street. (Levin and Rappaport 1990) (11) a. The bird screeched out of the tree. b. Her stomach growling loudly, Sheila rumbled down the street. Notice the same verbs, screech and rumble, are used in both (lOa-b) and (11a-b), and yet the examples in (10) are fully acceptable to all speakers, while those in (11) are rejected by many. The difference seems to be that in the examples in (10), the sound is caused by the motion, whereas in the examples of (11), the sound is not caused by the motion, but is simply a co-occuring event. Thus verbs of sound emission provide tentative support for the Causal Relation Hypothesis: that the meaning of the verb and the meaning of the construction must be causally related (at least for speakers who reject [1 la-h]).4 In order to look more closely into whether the Causal Relations Hypothesis plays an important role, I looked for novel uses of verbs in various constructions. Many of my examples were drawn from Clark and Clark (1979), "When nouns surface as verbs1'. Consider the following sorts of examples: (12) a. Ken Houdini'd his way out of the mailbag. b. Arthur wristed the ball over the net. c. The trainer kennelled the dogs d. The Jones' summered in New Hampshire. The first thing to notice is that each of these cases involves a metonymy. In (12a), a general metonymy, Agent stands for the Action associated with that Agent, licenses Houdini to be used to convey the action(s) associated with EBSCOhase - prinBed en 0:45 SK vis* CSITE&ZIIi. ill m« siabjftct: to fctrpa ://»w.ebaco.«nn/tems-of-uae RELATIOSHIPS BETWEEN VERBS AND CONSTRUCTIONS 389 Houdini. This metonymy is at work in many cases involving both proper and common nouns, including: (13) a. Nixon the tapes v b. Reagan the public c. butcher the cow d. jockey the horse Example (12b) involves the metonymy, Instrument stands for the Action associated with that Instrument. In this case, wrist is interpreted as an instrument conveying the action associated with the wrist. Other examples include (examples again from Clark and Clark 1979): (14) a. Criscothe pan b, sweater the child c badge the members There are also cases which have become conventionalized, so that what was initially an on-the-fly metonymy has become a lexicaiized verb meaning: (15) a Azita elbowed her way through the crowd. b. Dave hammered the metal flat. c, Kathleen handed Dave a letter. Example (12c), (The trainer kennelled the dogs), involves the metonymy: Result stands for the Action leading to that Result. In this case kennel is the result of the trainer acting on the dogs. The question arises as to how it is that the result involves the dogs being in the kennel I would like to suggest that the containment relation follows from general world knowledge about kennels and dogs, Support for this idea can be found in an experiment described by Clark (1973), She asked 18 month olds to "Do this", as she placed a block in a crib. The children dutifully did as they were asked. But the experiment continued with Clark placing the block beside or under the crib and asking the child to "Do this". In these conditions as well, the child put the block in the crib. That is, given a block and a crib, or a movable object and a container, children construe the natural relation between the two to be one of containment. It is claimed here that adults do the same. As we saw was the case with instrumental denominal verbs, there is some conventionalization in this class as well. That is, many denominal result verbs are highly conventionalized. These include not only cases in which the verb codes the resulting location (16a-b), but also cases in which the verb codes other types of result, e.g. the resulting shape (16c) ZESCCha-c - printed on S/IO75<]30 3H "i* Ha3&RYKCtfl. nHTVZRZITi. ill us« stihjeee -co Jrcrpa ://ww.efc»co-ccHi/t-i303"Of-iuit 390 ADELEE. GOLDBERG (16) a They housed the furniture. b. She booked reservations at the Bed and Breakfast. c. The chef cubed the meat, More generally, verbs often designate the result achieved by some action, and they do not need to be denominal: (17) a. Herb melted some butter, b. Ann burped her grandchild. To summarize, examples (1 ia-c), repeated in (18a-c), each involve a particular metonymy: (18) a. Ken Houdini'd his way out of the mailbag. Metonymy: Agent for Action performed by Agent b- Arthur wristed the ball over the net. Metonymy: Instrument for Action performed with Instrument c. The trainer kennelled the dogs. Metonymy: Result for Action leading to that Result If we consider a simple causal event an event that has an agent, an optional instrument and a result, it becomes clear that the denominal verbs fell into a pattern. These verbs can metonymically stand for the action associated with the agent, the action associated with the instrument or the action leading to the result. Thus each of the denominal verbs in (ISa-c) is used to stand for a critical pan of a causal event. (^agentj^)-K^^^str^^-»(^^esult/^) Figure /. Causal Sequence The verb can also code the causally related notion of enablement, as noted by Fauconnier and Turner (1994); (19) Carolyn let the water out of the tub. This case also falls squarely within the purview of the Causal Relations Hypothesis. The fact that many of these verbs are conventionalized should not take away from the fact that the larger generalization also exists: verbs are often related to a construction by lexicalizing a critical part of a causal event: the action, the instrument or the result. EE3COfc-a.=i= - pf.in-ced c-a Ej4S £H via l&£A£L¥ftQU± USFJERZIXii. All use atijec.fr -to bfitpa ://ww.efeaw.ccm/tttrs»3--o£~iise RELATTOSH1PS BETWEEN VERBS AND CONSTRUCTIONS 391 f Examples (12d) = (20a) and the similar (20b) seem to pose a clear challenge to the Causal Relations Hypothesis; (20) a. The Jones' summered in New Hampshire. b. They were going to weekend in Catalina. Clark and Clark describe these cases as involving denominal verbs of duration. Let us assume that the construction involved, Subj V Obi, has roughly the semantics, X BE-LOCATED Y. The verbs live, reside, stay can be considered semantic elaborations of this meaning. It is clear that the duration of a stay is not causally related to being in a location. Thus the appearance of denominal verbs such as summer and weekend appears to pose a challenge to the Causal Relations Hypothesis. However note that this pattern is severely restricted: (21) a. *She was going to summer where she lived. b. *She was going to weekend at work. c. ??She intended to lunch-break at the mall-Notice further that not every denominal verb of duration is acceptable: (22) a. *The Jones' were going to month in New Hampshire, b. They were going to week in Catalina. What seems to be critical is that the noun be construed as conventionally associated with travel, for example, a vacation, and not a temporal duration. Summer and weekends are conventionally associated with vacations. Weeks and months are not If the noun itself designates a vacation, the expressions are as expected, folly acceptable: (23) a. They planned to vacation in Spain. b. Azita and Peter were going to honeymoon in Greece. c. Carolyn was going to holiday in France. Other nouns that designate periods in which travel can be expected are also acceptable: (24) a. The Goldstein's sabbaticalled in England, b. Nina sojourned in Italy. Once it becomes clear that the denominal verbs must designate periods of expected travel, it is clear that they are in fact causally related to the meaning of the EBBCOhorrt; - printrf on E:4S 3M * MSi&HTKCVS. TTSTVERZITS.- m -nse subject to ktt^://m. CONSTRUCTIONS 393 f In these cases involving verbs with negative import, the claim as to how verb meanings relate to constructional meanings must be adapted slightly. The construction does not serve to pick out a scene and the verb a salient causey related action within that scene. Therefore the Causal Relation Hypothesis does not apply straightforwardly. Still, there is a sense ih which the negation of a proposition is closely related to its assertion. It has often been noted that sentences with negative import typically presuppose that the corresponding positive assertion is "on the table" (Givon 1979; Horn 1989 and references therein). That is, the scene associated with the construction is accessible in the discourse Support from this idea comes from a number of sources, Horn (1989) discusses this idea at length. He cites an early reference from Baldwin (1928: 146-48, ched by Horn 1989: 68): In order that a negative statement may have any value, there must have been some reason to suppose thai the affirmative statement of which it is the exact denial was true, either that it had been proposed for our acceptance by an interlocutor, that it had been part of our stored-up knowledge or purported knowledge, or that wc had in mind What we took at the moment to be sufficient ground for its acceptance ... Negation is a secondary function of thought, which presupposes the existence of positive judgments, Givon similarly states, "Negatives in general are uttered in a context where the corresponding affirmative has been discussed, or else where the speaker assumes... the hearer's bias toward or belief in - and thus familiarity with - the corresponding affirmative" (1979: 139).' The close relationship between a positive assertion and its negation is also evident in the interpretation of sarcasm. For example, I'm sure the cat likes you pulling its- tail, can be used to convey that the cat does not like having its tail pulled: The close relation between positive and negative is also attested in certain diachronic meaning shifts, e.g. words like bad and nasty coming to be used to mean "good". It is in this way that the meaning of the verb is related to the meaning of the construction. The verb serves to negate the positive meaning of the construction. As noted above, these cases do not conform to the Causal Relation Hypothesis in a straightforward way. However, like causal notions, the negation involved is in each case "force-dynamic" in the sense of Talmy (1985b). Force dynamic relations are those relations that involve causes, forces, counterforces and tendencies. Thus, we might broaden the Causal Relation Hypothesis to include all force-dynamic relations as follows. Force Dynamic Relation Hypothesis: The meaning designated by the verb and the meaning designated by the construction must be integrated via a (temporally contiguous) force-dynamic relationship. ZBSCOfcose - prixiCtd oc 2/2072320 £:4S IsM via UMvXUZTTa.. ail uac aulject. to hctps://wir.±fcico.ciw/tuii3^o£-iue 394 ADELE E. GOLDBERG This reformulation allows us to include both causal relations and denial under the broader notion of force-dynamic relations. However, in the following two sections, two more ways in which verbs can be related to constructions £fe discussed. Neither of these is a force-dynamic relation. Interestingly, however, each of these cases is somewhat limited in its application and judgments often vary across speakers. Thus, it seems there is a hierarchy of possibilities, with elaboration being the most prototypical, force-dynamic relations being the next prototypical, and the following cases being possible, but non-prototypical, 3.4.2. Preconditions If we assume that the ditransitive construction has roughly the meaning of transfer, i.e., "X CAUSES Y to RECEIVE Z" (e.g., Goldberg 1992b), then we find that this construction allows the verb to designate a precondition of transfer, namely, the creation or preparation of the transferred entity, For example: (28) Dave baked Elena a cake, Here the preparation of the cake is a precondition for Dave's transferring the cake to Elena. Transferring something from an agent to a recipient is associated with certain frame-semantic knowledge. In particular, we know that what is transferred from one person to another is often prepared for that purpose. The preparation or creation of the transferred entity can thus be viewed as a salient action within our frame semantic knowledge of transferring. Interestingly, for many speakers, the verb does not designate a precondition as readily in other English constructions. For example, for a theme to move in a direction requires the precondition that the theme be free of physical restraints. In the following construction which designates caused motion, the verb designates the precondition of removing constraints that will enable motion. However, judgments on the following examples vary, with speakers ranging from finding them fully acceptable to clearly unacceptable: (29) a. % The warden freed the prisoner into the city, b. % Pat unleashed the dog into the yard. The reason that the precondition, of preparation in the scene of transfer may be more available than the precondition that restraints be removed in the scene of caused motion may be simply that the preparation preceding transfer may be a more frequent occurrence in our experience than removal of restraints enabling motion. In transferring something from one person to another, it often happens that the transferred goods have to be prepared or created for the purpose. On the other hand, it is generally not necessary to remove any restraints before causing an entity to move: most movable entities in the world are unrestrained. RELATI0SH1PS BETWEEN VEKUS AND CONSTRUCTIONS 395 f- J. 4.3. Co-occurring activity The way construction allows the verb to designate a co-occurring activity that is not causally related to the action designated by the construction, For exajuple, (30) "He seemed to be whistling his way along." (Oxford University Press Corpus) Certain speakers find the same relation possible with the intransitive motion construction (recall also examples [1 la-b]): (31) % He whistled out of the room. Notice a co-occurring activity is not possible in the resultative (32) or ditran-sitive (33) constructions; (32) *She whistled the meta! flat, (To mean, she caused the metal to become flat while whistling.) (33) *She whistled him a box. (To mean, she gave him a box while whistling.) On the one hand, a co-occurring activity in a scene can be salient. In a collection of attested examples of the way construction, a good percentage of the cases involving a non-causally related co-occurring activity actually had conjoined verbs, making their status as salient all the more apparent. For example: (34) "...youth swore and snarled and sloganised its way round the lovely arcades..." (Oxford University Press corpus) At the same time, however, there is by definition no inherent relationship between the the co-occurring activity and the scene designated by the construction, except for a temporal overlap. Therefore insofar as the co-occurring activity is deemed salient, attention would seem to be naturally drawn away from the scene designated by the construction. This may be why this possibility is so limited.4 4. Conclusion We started out by framing the question of what types of semantic relationships verbs could bear to constructions in terms of a more general question, what types of actions are most salient within the event-types designated by simple clauses. The generalization that emerges from the data discussed here involves the following hierarchy of possibilities: EBSCCfcofS - printed £-ii3e 396 ADRIJ. E, GOLDBERG Hierarchy of Ways Verbs can be Related to Constructions (35) Elaboration > Force-dynamic Relation (means, instrument, result, denial) > Precondition, Co-occurring activity In particular, the most prototypical case is that in which the verb codes an elaboration of the construction That the verb's meaning and the construction's semantics be related force-dynamically is the next most prototypical possibility. Correspondingly less common are cases where the verb codes a precondition or a co-occurring activity. These last possibilities appear to be somewhat construction specific, with a fair amount of variation among speakers. More specifically, we can state the following hypothesis relating to the above hierarchy: For any two relations in the hierarchy such that A > B, we predict that if a construction allows B then it allows A; It follows that if a language allows B then it allows A. Endnotes * I would like to thank Mark Tamer, Michael Israel, an anonymous reviewer for this volume, my graduate students as UCSD, and the audience at the 1CLA in Albuquerque for helpful discussion on this topic. Some of the issues addressed here are discussed in compact form in Goldberg (1995: 61-65). The analysis here supersedes that analysis, 1. Pinker also correctly points oat that such abstract meanings will not serve to distinguish between kick and throw or between give and hand. But the proponents of syntactic bootstrapping have not claimed that this type of finer grained learning takes place solely on the basis of syntactic frames. They assume thai context plays a crucial role in acquiring these distinctions- 2. Goldberg (1995) refers 10 litis case as instance instead of elaboration. 3. As an anonymous reviewer points out. example {3 5b) is of course acceptable if it is construed to mean that the boat embedded itself into the cave by burning However, as the reviewer notes, this is what is predicted since on thai construal, the burning is the means of achieving tlie relevant effect 4. The dialect which accepts (35a-b) is addressed in Section 3.4.3. 5. The fact that the scene designated by the construction does not hold in the real world requires that novel verbs that are related in this way are not learned by simply recognizing a salient action within a scene as suggested in Section 2.3.1. in fact in many other cases as well as these cases with negaave import, the scene designated by the construction does not have a real world physical correlate in the context. For example, we can and often do talk about hypothetical . past or future events.' Tn all of these cases, the scene designated by the construction is one that is interpretable in the discourse context, but is not directly reflected in the immediate real world context. It is possible that in these situations, the constructional meaning does not EfiaCCiorfc - pxiaited on 2/20V£020 B^iS XU vi* nSTUEEim. Ill aa* aabjact to hrtua.-//wtr.*baco.-=iin/t*=aia-o£-iiM RELATIOSHIPS BETWEEN VERBS AND CONSTRUCTIONS 397 aid in the acquisition or verb meaning, although it still would be available when the scene designated by the construction is instantiated in context. More likely, speakers may be able to use the scene as construed in a given discourse (instead of in the real -world) as a way of determining the meaning of the verb in much the same way: the verb would be assined to code a salient action associated with the scene designated by the construction, whether or not thai scene or the action itself arc directly instantiated in the non-linguistic context. 6. It could be that the co-occurring activity is possible with the way construction and, for some speakers, win the intransitive motion construction because of a similarity with the intransitive plus motion adjunct: (i) She whistled all the way to the bank. References Baldwin, James M. (ed). 1928. "Entries for "Negation1 and 'Negative' " In Dictionary of Philosophy and Psychology, vol 2,146-49 New York: Macmiilan. Carboncll. Jaime. 1982. "Metaphor An Inescapable Phenomenon in Natural Language Comprehension". In W.G. Lehnert and M. Rhiglc (cds.), Strategies for Natural language Comprehension. Eribawn. Clark, Eve V. 1973. "'Nonliiiguistic strategies in the acquisition of word meanings". Cognition!: 161-182. Clark, Eve V. and Herb H. Clark. 1979. "When nouns surface as verbs". Language. Croft, William, 1991. Syntactic Categories and Grammatical Relations. Chicago: University of Chicago Press. Fauconoier, Gilles and Mark Turner. 1994. "Conceptual Projection and Middle Spaces". Department of Cognitive Science Technical Report 9401. — 1995. In A. Goldberg (ed), Conceptual Structure, Discourse and Language. CSLI Publications. Fisher, Cynthia, D. G. Hali, S Rakowitz, arid L. Gleitman. 1994, "When it is better to receive than to give: Syntactic and conceptual constraints on vocabulary Growth", In L Gleitman and B. Landau (eds.), The Acquisition of the Lexicon. Cambridge, MA: MIT Press, 333-376. Genrner. Detdre 1983. "Structure-mapping: A theoretical framework for analogy"'. Cognitive Science 7'. Giv6n. Talmy. 1979. On Understanding Grammar New York: Academic Press. — 1984, Syntax: A Functional-Typological Introduction. Amsterdam/Philadelphia: Benjamins. Gleitman, Lila. 1994 "The Structural Sources of Verb Meanings", hi P. Bloom (ed), language Acquisition; Core Readings. Cambridge: MIT Press, 174-221. Goldberg, Adele E. 1991. "A Semantic Account of Resuhatives". Linguistic Analysis 21 (1-2): 66-96. — 1992a. Argument Structure Constructions. Ph.D. dissertation. University of California, Berkeley. Goldberg, Adelc E. 1992b. ''The Inherent Semantics of Argument Structure: The Case of the English Ditransitive Construction". Cognitive Linguistics 3(1): 37-74. E&KCbos* - pri»Md on Sz45 AH via M£5aRYKCV3i IfimSEftZIIfc. ill U3- *ubj«t to i-v.^p.i://*-xv .«-fc*co. cot/nexus-of-iiae 398 AOELE E, GOLDBERG — 1995. Constructions: A Construction Grammar Approach to Argument Structure, Chicago: University of Chicago Press. — MS. Semantic Principles of Predication. University of California, SanJDiego. Horn, Laurence R. 1989. A Natural History of Negation. Chicago; Chicago University Press. Hovav, Malka Rappaport and Beth Levin. 1996, Ms. Building Verb Meanings. Bar Dan University and Northwestern University. Landau. Barbara and Lila R. Gleitman, 1985, Language arid Experience: Evidence from the Blind Child. Cambridge. Mass.: Harvard University Press. Leek, Frederikc van dcr. 1996. "Rigid Syntax and Flexible Meaning: The Case of the English Ditransitive". In A. E, Goldberg (ed.), Conceptual Structure, Discourse, and language. Stanford: Stanford University Press. Levin, Beth and Malka Rappaport Hovav. 1990. 'The Lexical Semantics of Verbs of Motion; The Perspective from Unaccusativity". In 1. M. Roca (ed.), Thematic Structure: its Role in Grammar. Berlin: Mouton dc Gmyter, 247-269. — 1995. Unaccusativity: At the Syntax-Lexical Semantics Interface. MIT Press. Naigles, ljCtitia. 1990. "Children use syntax to leam verb meanings". Journal of Child Language 17: 357-374. Naigles, Letitia. 1995. Using Multiple Syntactic Frames in Learning Verbs, Presentation at University of California, San Diego. Naigles, Letitia, Lila Gleitman and''Henry Gicitman, 1993. "'Children Acquire Word Meaning Components from Syntactic Evidence". In E Dromi (cd.), Language and Cognition: A Developmental Perspective. Human Development, vol. 5, 104-140. Pinker, Steven 1994. "How Could a Child Use Verb Syntax to Leam Verb Semantics?" Lingua, — 1989. Leamability and Cognition: The Acquisition of Argument Structure. Cambridge, Mass: MIT Press. ScthuramaiL Mtya, Adeie E. Goldberg and Judith Goodman- to appear, 1996. "Using the Semantics Associated with Syntactic Frames for Interpretation without the Aid of Context". In E. Clark (ed), Stanford Child Language Research Forum. Talmy, Leonard, 1985a. Lcxicalization Patterns: Semantic Structure in Lexical Forms. In T. Shopen (ed), Language Typology and Syntactic Description, vol 3: Grammatical Categories and the lexicon. Cambridge: Cambridge University Press, 57-149. — 1985b. "Force Dynamics in Language and Thought". Chicago Linguistic Society 21.1, Parascssioii on Causatives and Agentivhy, 293-337 ras^Cid-jt — pxiireftd o±i S/UV2020 E:45 SU via HASSSEfK&JSi. U3TIVERZIX?... AL1 i»e siibje-ct. to hctp9:y/vw.eb4ca.eDn/teoi7-o£-iue