Cognitive Linguistics Research 1 Editors René Dirven Ronald W. Langacker Mouton de Gruyter Berlin - New York Concept, Image, and Symbol The Cognitive Basis of Grammar Ronald W. Langacker 1991 Mouton de Gruyter Berlin - New Yorkvi Contents: 4. The English pas 1, Grammar and analyzability 11, Economy 12. Components 1.3. The symbolic nature of grammar 1.4, Semantic structure 15. Grammatical morphemes 1.6. Overiness of grammatical structure 1.7. Continuum of lexicon, morphology, and syntax 2. Descriptive framework 2.1. Linguistic organization 2.2. Syntagmatic combination 3. The passive construction 3.1. The perfect participle 3.2. The passive be 3.3. By-phrases 34, Related phenomena 5. Abstract motion 1. Basie concepts and assumptions 2. The characterization of verbs 3. Objective motion 4. Subjective motion 5. Avenues of semantic extension 6. Grammatical valence 1. Canonical instances 2. Noncanonical instances 3. Further departures from the canon 4. Scope and morphological layering 7. Active zones 1. The phenomenon, 2. Analysis 3. Grammatical implications 8. The Yuman auxiliary 101 102 102 104 105 108 10 13 us 116 116 120 27 129 135 139 143 149 149 152 155 157 160 165 167 174 179 183 189 189 193 196 ‘Transitivity, case, and grammatical relations 1. The conception of actions and events 2. Unmarked linguistic coding 2.1. Some basic grammatical constructs 22. Selection 2.3. Heads and wails: 24. Complexities 25. Subjective asymmetry 2.6. Subject and object 2. Marked coding 3.1. Lexical options 3.2. Voice 3.3. Setting vs. participants 4. Case 4.1. General comments 42. Role archetypes 4.3. Correlation with grammatical relations 44, Brgativelabsolutive systems 45. The structure of events 46. Ergativity 4.7. Other case phenomena 5. Causative constructions 5.1. Complex events 5.2. Causatives derived from intransitives ‘5.3. Causatives derived from transitives 10. A usage-based model 1. Two conceptions of generality 2. The network conception 3. General applicability 4. Distribution 5. Conclusion 11. Autonomy and agreement 1. The autonomy issue 2. The symbolic alternative 3. Grammatical markings 4, Agreement S, Conclusion 209 2u1 an 213, 216 219 22 224 226 226 29 230 236 239 241 243, 249 254 254 256 261 261 266 22 278 288, 289 292 301 313,Figure 17. Observe that the postulated constitency of this sentence, as depicted in Figures 13 and 17, is perfectly compatible with the one suggested by phonological and morphological considerations. I] and [HIM] are added only after the entire complex verb has been assembled, as reflected by their status as separate words noo poy ee-vicu-ni-g. Nevertheless, the correspondences which figure in the various valence relations (and those’ internal to certain predications) properly establish the speaker as the trajector of the causative relationship, and the third person singular individual as the trajector with respect to both the wanting and the leaving, All of this is accomplished without deriving the sentence from a hypothetical underlying structure posited solely 10 accommodate its supposed prodicate-argument configuration (cf. Figure 1(b)), and without special rules like Predicate Raising and Equi-NP Deletion (Figure 1) As a general matter, this framework promises to reconcile semantic and ‘grammatical constituency with the constituency one is led to posit on phonological grounds, Basic grammatical relations are not defined with respect t specific Constiuent structure configurations, but rather through correspondences and the lwajectorlandmark alignment intemal to every relational predication, The same corespondences, and hence the same grammatical relations, can be established through alternate compositional paths leading to the same overall composite structure, This is why the Conventions of a language often permit alternate word forders or phonological phrasings for otherwise equivalent expressions. Comespondences are essential to grammatical valence relations, but constituent structure is to some degree incidental and variable. 7. Active Zones ‘This chapter investigates a certain phenomenon, illustrated in (1), pertaining to the ‘compatibility of a relational predication with its arguments.* (1) a. David blinked. b. She heard the piano, €.1'm in the phone book. All are perfectly normal expressions, yet they exhibit an apparent peculiarity when ‘we think about what they actually seem to say. The subject of (1a) names a person, but the activity of blinking is not something carried out by a whole person--only the eyelid realty does anything. In (1b), we encounter the seeming. anomaly of a person hearing a physical object, when the only things we can in fact, hear, quite obviously, are sounds. If (Ic) is taken literally, moreover, we can only ‘conclude that the speaker is either very small or quite cramped. Examples like these have often been noted in one connection or another, but I am unaware of any comprehensive and unified treatment. Their implications are quite substantial, and go right to the heart of critical grammatical issues. I will Claim, for instance, that the sentences in (2) illustrate the same basic phenomenon as those in (1). (Q)a. Wombats are tough to catch. Tom is likely to succeed. ¢.1 believe Donovan to be an honorable man. ‘The sentences in (2) are commonly analyzed as manifesting the syntactic moverient (or relation-changing) rules of Object-to-Subject Raising, Subject-to- Subject Raising, and Subject-to-Object Raising, respectively. I will suggest, on the contrary, that a reasonable account of the normal connection between relational predications and their nominal arguments, required even for single-clause expressions, obviates the need for special syntactic rules ofthis sot. 1. The phenomenon Let us take the sentences in (3) as our point of departure. In each instance it can be argued that the central relational predication[APPROACH], [BEYOND], or [NEAR}--profiles a relationship in which its trajector and landmark participate as integral, undifferentiated wholes. As the spacecraft approaches Uranus, for example, every part of it becomes progressively closer to every part of the planet,190 Active zones and no particular subpart of either one has special status in this regard. (3)a. The spacecraft és now approaching Uranus. . Goleta is beyond Santa Barbara. . Your dog is near my cat. ()a. Your dog bit my cat. 1, Your dog bit my cat on the tail with its sharp teeth, ©. "Your dog bit my cat with its teeth. ‘When we examine (4a), a rather different picture emerges. Here it is evident that the relational predication [BITE] designates an interaction between its trajector and landmark that directly involves only selected aspects of these entities. For instance, the teeth of the dog are pivotal to the act of biting, but the contribution of its tail and kidneys t0 this process is at best negligible respect to the landmark, [BITE] itself does not specify which particular part of the cat is privileged to participate directly, but [BITE] strongly suggests that only restricted portions are affected ‘Those portions of a tajector or landmark that participate directly in a given relation will be referred t0 as its “active zone" with respect to the relation in question. For some relational predications, like those in (3), the active zone of the {rajector and/or landmark is coincident with the whole. For others, like (BITE), the active zone of the trajector and/or landmark is limited to a proper subpart of the whole, The active zone should not be thought of as a discrete or sharply founded region within the overall entity-it is more accurate to think of it as the focal area of the relationat interaction, the participation of a region becoming, more tenuous the farther it les from this focus. For example, the trajector's participation in the Process designated by [BITE] is not strictly limited to its teeth: also involved are the jaws, the operative muscles, the nervous system, and so on. Pushing things to the extreme, one could argue that every part of the trajector is involved in the act of biting in one way or another, however marginal it might be, if only because all ‘Portions of the body arc interconnected, so that no portion is totally unaffected by ‘what happens to any other, I would not dispute this argument. ‘The only point crucial here is thatthe participation of certain regions is obviously more direct and mote central to the relational conception than that of others. Linguistic form is often blind to these subteties. Precisely the same expressions, namely your dog and my cat, are used 10 describe the trajector and. primary landmark in sentences (3c) and (4a), even though these entities participate as integral wholes only in the former. On the assumption that the nouns dog and at designate the entie animal (9 that dog, for instance, would be an inappropriate expression to describe a disembodied set of teeth), we can observe that (4a) manifests a notable discrepancy between what is profiled by the trajector and landmark of {BITE] and the active zones of these entities with respect to the The phenomenon 191 process. This discrepancy sometimes has overt linguistic consequences; it is possible, for example, to spell out the active zones explicly by means of Periphrastic expressions, such as the prepositional-phrase complements in (4b). Generally, though, such periphrasis is optional, and one resorts o it only when this additional specification is informative. This is the case in (4b), since (BITE) is vague about the ative zone of its landmark, and since the adjective sharp provides information about the active zone of the tajector that would not otherwise be available. However the prepositional-phrase complement in (4c) adds nothing to the content conveyed by the verb, so this sentence is needlessly redundant, ‘The existence of a substantish discrepancy between the entity profiled by an expression and its active zone with respect to a given relational predation is not all unusual. In fact, bt of reflection reveals that a discrepancy between profile and active zone represents the NORMAL situation. 1t is in fact quite difficult to find convincing examples like those in (3), where all aspects of the designated entity participate equally in a relationship. In the averwhelming majority of instances the various facets of the profiled entity participate in a relationship to different degrees and in different ways. The examples in (5) afford some initial appreciation of the ubiquity of this phenomenon, (5) a. Roger blinked. . Roger ate an apple, c. Roger heard a noise, 4. Roger walked faster. ©. Roger is digesting. £. Roger figured out the puzzle g. Roger whistled. h, Roger peeled an orange, i, Roger licked the popsicle J. Roger breathed hard. Considering the trajector of the main verb, we find that only selected facets ofthe designated individual participate directly in the process, and that these facets differ from one expression 10 the next. Roger's eyelids are the primary participants in ‘Ga; in (5b), his hands, mouth, teeth, tongue, and the upper parts of his alimentary canal are more diecly involved than, say, his kneecaps, his ears and central nervous system are pivotal in (Sc); while the whole body moves in (Sd), the legs in particular are of prime importance; and so on. In those cases where the landmark istinct from the trajector, there is comparable variation. ‘Thus the puzzle in (Sf) ‘avticipates holistically in the verbal process, but presumably only part ofthe apple is eaten in (Sb) (most people do not consume the core), while in (Sh) only the ‘outer surface ofthe landmark object is affected. Some discrepancy between active zone and profile is thus the rule, not the ‘exception, Our conception of grammatical organization must cherefore be able to‘accommodate it as a normal situation rather than a pathological one, preferably without special apparatus. In fact, the problem is even worse than I have so far indicated. For one thing, even for those predications which appear to relate the Urajector and landmark as integral wholes there is often an active-zone/profile discrepancy in specific instances. Consider {IN], whose trajector would seem to Participate holistically in the inclusion relationship with its landmark. Yet this is very commonly not the case: (6) a, Abernathy is inthe bathtub, . Susan has a cigarette in her mouth, He has an ace in his hand. ‘A normal construal of all three sentences situates only a portion of in’s rajctor within the confines ofits landmark. ‘A second exacerbating factor is that the active zone is often not even a subpart of the entity designated by a nominal expression. Frequently it is something merely associated with the designated element in some characteristic fashion, as ‘we saw above in (Ib)-(Ic). In (1b, the subject does net fea the piano per se, a8 2 Physical object, but rather the sound emitted by the piano~canonicelly this would be the musical sound produced dy playing the piano, but in context it could also be the crashing sound it makes when dropped from a helicopter. In (Lc), of course, it is not the speaker as a physical object who occupies the phone book, but rather a symbolic representation of his name, address, and phone number. (7) provides some further examples, (Datsmelt a cat . The balls yellow. ©. need ared pen 4. This red pen is yellow. might say (7a) when opening the garage door in the morning; what I actually smell is che odor emitted by certain excretions ofthe cat. In (7b), it is not the ball as a physical object that is capable of interacting with color space (a range of possible color sensations)~rather it isa color sensation associated with the ball's ‘outer surface. ‘The phrase red pen is ambiguous, and its ambiguity hinges precisely on the choice of active zone for pen with respect to the color predication. On the ‘one hand, the active zone may be the color sensation associated with the outer surface of the pen (parallel to (7b)); on the other hand, it may be the sensation associated with the marks left on the page when the pen is used as a writing implement. By taking the notion active zone into account, we can explain why (74) is meaningful and noncontradictory. The active zone of the pen with respect to the precication [RED] is the color of the marks it makes, while for [YELLOW] itis the color of the pen’s outer surface. The phenomenon \3S Obviously, the permiued discrepancy between profile and active zone greatly increases the flexibility ofa linguistic system, Ifthe two were always required to coincide precisely, i.e. ifthe trajector and landmark of a relational predication had 40 be expressed with full accuracy and specificity, the result would be a vast proliferation of highly cumbersome locutions a sample of which is offered in (8). (8) a, Roger's eyelids blinked. b. Roger's mind figured out the puzzle. ©. Roger's lungs and oral tract whistled. «Susan has the end ofa cigarette in her mouth . He has a portion ofthe handle of an axe in his hand. The color sensation associated with the outer surface of his pen the color sensation associated with the marks created by which is red, is yellow. The tolerance of profile/active-zone discrepancy is further quite natural in cognitive terms, for it permits the designation of linguistic expressions o focus on conceived entities that have substantial cognitive salience. The following principles can be seen at work in various examples: (i) a whole is generally sore salient than its individual parts; (i) discrete physical objects are generally more salient than abstract entities; and (ii) humans and (to a lesser extent) animals are generally more salient than inanimate objects (other things being equal). 2. Analysis ‘The apparent difficulty posed by profile/active-zone discrepancies is that what a sentence literally says conflicts with how it is actually understood, so that its ‘compositional semantic value is either inappropriate oF logically inconsistent. This is the type of situation for which generative theory has commonly posited logically ‘consistent underlying structures, from which the surface form of an expression is derived by the application of transformational rules or some comparable device. For example, if it is granted chat one can only hear sounds, not physical objects, the seeming anomaly of (Ib) might be accommodated by taking (Qa) as a deep structure and postulating an optional transformation that deletes the boldface Portion of it (9) a, She heard the sound of the piano, b. The color sensation associated with the outer surface of the ball is yellow. c. Roger’s hands, mouth, teeth, tongue, and the upper parts of his ‘alimentary canal ate an gppie. Let us call chis type of approach the “linguistic paraphrase analysis": anWA Active zones expression derives fom an underlying structure that-when manifested without \deletions~-provides a logically accurate paraphrase of is actual meaning. ‘The linguistic paraphrase analysis is offered only as a straw man, not as a serious analysis on my part or anyone else's. There are obvious problems with it While reasonable enough for cases like (9a), where a simple and obvious Paraphrase readily suggests itself as a deep structure, its plausibility quickly evaporates when one tres to extend it to a representative array of instances: few linguists would be attracted by the deletions in (9b) and (9c) forthe derivation of simple sentences like The ball is yellow and Roger ate an apple. ‘The choice of a particular paraphrase to serve as underlying structure is arbitrary, moreover, and ‘any paraphrase that is chosen is itself likely to prove inaccurate in mare suble ‘ways; (9¢), for instance, does not indicate that the various body parts mentioned patticipate in the process wo different degrees and in different ways. Furthermore, ‘be linguistic paraphrase analysis teats as problematic--as something t0 be remedied by abstract constnicts--something that in actuality represents the normal situation rather than anything exceptional, A more natural account would be one in which the phenomtenon is not a problem at al Actually, { would argue that the perception of the apparent difficulty rests on certain tacit and ill-founded assumptions. On what basis does one conclude that a), for example, is more accurate ot logically valid than (1b), She heard the iano, a8 a characterization of the conceived situation? It is by virtue of the assumption-which I mysel€ exploited in presenting the data--that hear has precisely the same meaning whether sound or piano occurs as its direct object; consequently the collocation of hear a piano should be anomalous, since a sense of hear which specifies that its direct object isa sound cannot combine felicitously with an object nominal designating a physical object. This tine of thought Drejudges the semantic value of the verb, however, treating it as nonpolysemous and atibuting certain specific properties to its single meaning, The validity of ‘iese assumptions is not self-evident, Instead of saying that hear has a single ‘meaning, designating the interaction between the perceiver and a sound, one could perfectly well say that hear has two semantic variants: the first designates the interaction between a perceiver and a sound, while the second designates the anteracsion between a perceiver and a sound-emitting object (the emission of sound ‘being the basis for this interaction). (8a) and (Ib) then involve different ‘redications, both symbolized by the phonological sequence hear, and in neither expression does the meaning of the direct-object nominal conflict with the nature of the eelasonal lanmark implied by the verb. ‘The linguistic paraphrase analysis Avoids the necessity of postulating two separate predications associated with hear, ‘but this is at best a Pyeshic victory, as it engenders the need for highly abstract and highly problematic analyses like those in (9). Moreover, I take it as established that polysemy is whe normal state of affairs for common lexical items. The truly heroic effors that would be required 10 eradicate it in favor of abstract derivations inall instances would most certainly be radically misguided. Analysis 195 ‘The analysis is rendered a bit more explicily in Figure 1 (where AZ labels an active zone), The predication skeiched in i(a) corresponds to sentences like (9a), while that in 1(b) is illustrated by sentences like (1b). If the former is somebow ‘more “basie™ than the latter, as the argusnent for a transformational derivation OF (1b) assumes, I would interpret this not as a matter of logical consistency, but rather as reflecting the greater enisenchment and. cognitive salience of (HEAR), from which (HEAR’] may well originate as a semantic extension; all of this is perfectly consistent with the view of polysemy outlined earlier and is consonant with the general character of lexical items. Both predications have the same base, and differ only in the profile they impose on this base. Common to bath predications, in other words, isthe knowledge system that involves the full array of relevant concepts: that of sounds, including their emission from physical objects (or otter sources); that of perceptual experience, implying a perceiving individual; and the knowledge that such experience relies on certain auditory apparatus within the perceiving individual. ‘auditory spparatus (AZ) ‘audttory apparatus (AZ) Figure J ‘The contrast between the two predications-~the two senses of hear--comes down {to a matter of imagery. Specifically, the predications differ as w the substructures they single out for special prominence as the trajector (gure within the relational conception) and primary landmark (the most salient entity other than the figure). However, the relative prominence of substructures, though itis definitely an aspect of meaning and constitutes a semantic difference between the two predications, is independent of what one might call the “content” of the predications as defined by their common base, In both instances, the perceiver as an integral whole is profiled as the relational trajector, even though selected facets ofthis individual are own 10 function more directly in auditory experience than others, Furthermore, in both predications the sound is conceived as mediating the relationship between tis trajector and the sound source, regardless of whether it is Wie sound in particular or its source that is profiled as the primary landmark. [HEAR] focuses tm the direct imeraction between the sound and the perceiving individual, while196 Active zones {HEAR’] profiles instead the mediated interconnection between the perceiver and the sound source, but both relationships are part of the meaning of beth expressions, despite their differing salience. Observe that the two predications share a profilefactive-zone discrepancy with respect to their tajector. [HEAR’) also exhibits such a discrepancy with respect to its landmark, whereas for [HEAR] ‘he profile and active zone of the landmark coincide. In short, the cognitive grammar framework accommodates the phenomenon With no special apparatus whatever. All of the construsts needed for this purpose are independently established features of the model, most notably its treaunent of lexical polysemy, the profle/base conception of semantic structure, and the tajector/landmark asymmetry among the participants in a profiled relationship. Profiling and trajector/andmark alignment are matters of the relative prominence cf substructures within a domain, and to some extent these aspects of semantic organization can vary independently of how the various substructures are inwinsically structured and interconnected. ‘The entities accorded the status of trajector and primary landmark can therefore deviate from those that participate ‘most directly and critically ina relationship, without this affecting its “‘content™ or its “logical properties”. 3. Grammatical implications Since the nature of a predication’s profile determines its basic grammetical category, lexical polysemy of the sort just described can extend across ‘grammatical classes. A term like yellow, for example, can designate either a ‘bounded region in color space, in which case the expression is nominal, or else a relationship of coincidence between this region and alight sensation, in which case the expression is adjectival. There are actually multiple adjectival senses. ‘They differ as to whether the light sensation is itself selected as the trajector (relational figure), or whether that honor is accorded to some other entity, typically a physical object whose outer surface is the source of this sensation, ‘The contrast between the yellow of yellow flash and that of yellow ball is therefore not unlike the one between the two senses of hear. Another example of polysemy across grammatical classes is provided by ‘modifiers like fas that can be used as either adverbs or adjectives (e.g. work fast vs. fast car). ‘The adverbial predication is sketched in Figure 2(@). Is domain is the conception of a rate scale, and the region on this scale that lies beyond the neighborhood of the norm (n) functions as the primary landmark. Processes can be situated at various points along this scale, andthe trajectorof the predication is one such process, specified as being situated within the landmark region. What about the adjectival fast? A rate scale is clearly relevant, but a physical object like a car cannot, per se, interact with this scale, ‘The active zone of a physical object with sespoct toa rate scale must be some process in which this object participates, This, Grammatical implications 197 is made explicit in Figure 2(b), which has exactly the same base as 2(a), ‘The contrast lies solely in the profiling, and specifically in the choice of overall ‘ajector: the tajector of the adjectival [FAST] is not the process that occupies the landmark region of the sate scale, but rather the trajector of that process, The process is nevertheless a crucial part of the meaning af the adjective, as it mediates the relationship between the overall trajector and landmark. @) | | ©) mist (Ae) Figure 2. We saw previously, in (4), that when the active zone of a presication diverges from its profile the former can be spelled out periphrastcally if there is ‘communicative motivation for so doing. This is true for the implicit process that functions as the active zone of the adjectival [FAST]. When the modified noun strongly implies @ particular type of process, specifying this process Periphrastically is superfluous, as seen in (10). (10) That {barberirunnericarisurgeon) is fat. (11) a. When it comes to sweeping out the shop, that barber is really fas. b. That surgeon is fast a solving a Rubik's Cube Periphrasis becomes necessary, however, when the process that functions as active zone is not the one derivationally oF canonically associated with the modified noun, as in (11). ‘A very similar analysis readily accommodates the seemingly unrelated data cited in Newmeyer 1970. The following examples are representative: (12) a, He began eating dinner. 'b. He began dinner. (13) a. The orchestra started playing the next song. b. The orchestra started the next song.198 Active zones (U4) a, The author finished writing a new book. . The author finished a new book. ‘The problem is to account for the second sentence of each pair, where an aspectual verb like begin, start, or finish takes a simple direct-object nominal instead of the verbal complement we would “logically” expect. Newmeyer proposes something akin to the linguistic paraphrase analysis, but T would simply say that these verbs ‘manifest a pattem of lexical variation hinging on a permitted discrepancy between active zone and profile, ‘The basic analysis is presented in Figure 3. @ O ° Figure 3. ‘have nothing of a detailed nature to say about the semantic structure of verbs like begin, so instead of trying to diagram this strature {have simply employed an abbreviatory notation for a process predication. In a sentence like (122), [BEGIN] Profiles a process in which the tajector is a thing (spelled out by the subject ‘rominal) and the primary landmark is another process." fn a sentence like (12b), ‘on the other hand, the related predication {REGIN'T profiles a process in which ‘both prominent participants are things: the tajector is the same, but the primary landmark is not the initiated process as a whole, but rather the landmark of that process. The initiated process remains as a pivotal facet of the base-it isthe active zone of the landmark with respect (0 the inceptive process designated by this predication—but it need not be spelled out explicitly when its characses is apparent from context or the other lexical items inthe Sentence. 1 would emphasize that the analyses presented so far in this section require no specitl apparatus, They employ precisely the same constructs introduced earlier and needed to handle the active-zone/profile discrepancies of simple sentences like those in (1). The only novel aspect of these latter analyses~one that is really nothing more than a special case of the general phenomenon~is the notion that the active zone of a thing with respect to a relationship may be a process in which that thing participates. This is necessary to account for the adjectival sense of fas, a well as examples like (12)-(14). Neither of these is generally thought wo be associated with the raising constructions of (2), but it may already be apparent that Grammatical implications 199 the present analysis establishes a connection. In fact, ihe analysis € prapose for raising constructions should by now be quite obvious. Let us focus on Object-to-Subject Raising. The standard analysis derives the semences in (15) from the respective deep structures that also underlie those in 6). (15) a. Hondas are easy to fi bo. Landscapes are tough to paint ©. Monopoly is fun to play. (16) a, To fx Hondas is easy. . To paint landscapes is tough. ¢. To play Monopoly is fun, Main-clause predications like easy, tough, and fun are claimed to have precisely the same meaning in the constructions of (15) and (16), one that selocts a clausal suibject at the deep-structure level. Object-o-Subject Raising then accounts for the superficial divergence from this pattem in cases like (16). ‘There are two basic classes of arguments that are generally advanced 10 motivate this type of analysis. One class-involving “idiom chunks” (c.g. headway) and “dummy” elements lke it and there—I will ignore here, except 10 note in passing that they depend on certain assumptions that 1 do not accept (c Langacker 1987). Instead 1 will concentrate on the second type of argumeni, which appeals to selectional restrictions and “logical” grammatical relations, In presenting the raising analysis to a clas, 1 have often said something very much like the following: “*What does (15a) mean? Logically, itis not Hondas that are easy, but rather the process of fixing them. The superficial form of the sentence obscures this relationship, but such relationships are captured explicitly at the Aeep-structure level if sentences like (15) derive by transformation from underlying stracuses like (16).”” Important and persuasive as such arguments wese in establishing the transformational model, 1 would claim in retrospect that they are fallacious. The raising rules exemplify the linguistic paraphrase analysis, and the argument from Jogical grarmmatical relations is subject to the same criticism advanced in section 2 with respect to the putative derivation of (Ib), She head the piano, from (9a), She heard the sound of the piano: the argument prejudges the semantic value of the ‘governing lexical item, assuming quite gratuitously that it has precisely the same sneaning in both constructions, and that this meaning can be determined on some kind of (ill-defined) “logical” grounds. Instead I suggest that fough, for instance, is polysemous, having the two semantic variants sketched in Figure 4 (among others). The base for both predications is a conceived scale of difficulty along which processes can be situated, The primary landmark is the region of dhs scale thet lies beyond theWY Active emer neighborhood of the norm. In the case of (TOUGH), probably to be regarded as the more deeply entrenched variant (from which the other is extended), the trajector isa process that is located in this andmark region. AM ofthese elements are present if the base of (TOUGH’], which differs only in its choice of overall trajector: the trajector of this predication is equated not with the entire process situated in the landmark region, but rather with the landmark of this process in ppaticular. The full process is nevertheless crucial to the semantic valut of [TOUGH], for it constitutes the active zone of the overall trajector with respect to its interaction withthe difficulty scale, ® rgrrosoe TE oy Figure 4. In some respects my lexical variant analysis resembles the syntactic raising analysis. However the former straightforwardly accommodates certain types of 4ata that cannot be handled in @ purely syntactic account. I teat the zoclauses in (15) as periphrastic speciticaions ofthe trajectr's active zone with respe:t 10 the scale of difficulty (or pleasure, et.); parallel sentences lacking these clauses are thus not unexpected, but they are problematic in the raising analysis, since the ‘missing clauses are the putative source ofthe subject nominal, (17) a, Landscapes are tough. b. Monopoly is un cc. When it comes to fuing them, Hondas are easy, Sentence (174) would be perfectly natural in the context of a painting class, where the process functioning as the active zone is too obvious to require periphrastic specification; the operative process is similarly quite apparent even out of context in (170). Moreover, the lexical variant analysis is consistent with alternate types of Periphrasis (though a to-clause is standard). Sentences like (I7c) are thus ‘anicipated, but they are problematic for the standard raising analysis The analysis of Subject-o-Subject and Subject-to-Object Raising differs only in specifics. Slightly different senses are thus attributed to likely in (18a) and (186): a process serves as the overall trajector of [LIKELY] in (182); the rajector of this Gramenatical implications 201 rocess is specified instead as the overall trajector of [LIKELY’1 in (18b), but the process remains in the base and functions as the active zone for the interaction of the overall trjecior with the probability scale. The £o-clause in (18b) elaborates, the active zone periphrastically, but on occasion it can be omitted, as seen in (49), (18) a, For the dog to escape is likely 'b. The dog is likely wa escape. (19) a, A war is likely b.Q Do you think anyone will come to the party? A: Well, Tom is likely. (20) a. would expect for the Clippers 0 lose again. b.L would expect the Clippers to lose again. €. When do you expect Tom? In similar fashion, slightly different predications are associated with expect in (20a) and (20b). A process functions as the overall landmark of {EXPECT} in (20a), but the wajector of this process has this satus in the case of [EXPECT in 200), The landmark process of (201) is nevertheless an active zone in (20b), ‘here itis elaborated periphrastically by the fo-clause. While such elaboration is ‘generally obligatory, sentences lke (20) show that itis optional with expect when the process is understood to involve nothing more than arrival on the scene,