Faber, Pamela, and Ricardo Mairal Usón. 1998. Towards a semantic syntax. Revista Canaria de Estudios Ingleses 36: 37-64.
TOWARDS A SEMANTIC SYNTAX' Pamela Faber Universidad de Granada Ricardo Mairal Us6n UNED, Madrid INTRODUCTION In the Functional Lexematic Model lexicon (hereafter FLM)2, the paradigmatic and syntagmatic information of each lexical entry are closely interrelated to the extent that a verb's syntax depends on its location in semantic space. In other words, a verb's combinatory possibilities and potential syntactic patterns are semantically motivated. This interface between syntax and semantics in verbs has also been observed by others (i.e. Dixon, 1991). Levin (1993: 5) states that various aspects of a verb's syntactic behavior are tied to its meaning. Pinker (1994: 395) also writes: ...in most languages a verb can appear in a family offorms, each with a distinct meaning component, plus a common meaning component that runs throughout the family. Nevertheless, the problem is that in the majority ofcases, syntactic coding has been used as the starting point to explain semantics. For example, words have been grouped in semantic classes on the basis of grammatical alternations, such as the following: (!) The harpoon plunged into the water. (2) The sailor plunged the harpoon into the water. [zero-derived causative form of plunge] (3) The hunter shot the tiger. (4) The hunter shot at the tiger. [conative grammatical alternation} This in itself is insufficient as a basis for the determination of semantic classes because syntax alone does not provide a satisfactory explanation of .semantic constraints. As Pustejovsky (1995: 10) points out: Revista Canaria de Esrudios lngleses, No. 36, 1998, p<'igs. 37-64 ...participation in one grammatical alternation does not sufficiently determine the semantic class of the verb. In fact, even once a complete cataloguing of participation in alternation classes is achieved, we must ask ourselves just what we have accomplished. The goal of any lexical semantic theory is to establish the premises upon which words can be organized into classes which to a great extent predict their syntactic and semantic properties. One might argue that this is a difficult task, but we believe that regularities in the lexicon can be mapped out by means of syntactic-semantic (synsem) parameters which operate throughout the lexicon in the various areas of meaning. Such parameters can be derived by analyzing the meaning definitions of semantic sets, as well as the complementation patterns characteristic of each subdomain in the FLM lexicon. We believe that they are the key to understanding lexical organization, and will ultimately serve as the basis for the set of lexical rules, which constitute the nucleus of domain-specific grammars (cf. Faber and Mairal (fc)). 2. MEANING DEFINITIONS AND SYNTACTIC PROJECTIONS: SYNSEM FEATURES Given the importance of semantics in syntactic representations, it is necessary to reconsider the ancillary status of meaning definitions in FG (Dik, 1978). We believe that such definitions, when considered in paradigmatic contexts, play an important role in determining the complementation of a given predicate, as well as the occur rence of certain grammatical features, such as the state of affairs of the predication and its co-occurrence with certain satellites. In this regard, the domains in the FLM lexicon are configured in such a way that the syntactic hierarchy in each subdomain can be mapped onto the semantic one. The previously-mentioned synsem parameters are one of the products of this explicitation of the interface between syntax and se mantics. Although all of them play a role in the generation of the actual clause struc ture, some have a greater effect on syntax than others. We have divided these parameters into three types, according to their scope of application: (a) LexicallyMrealized grammatical parameters are obligatory and are al~ ways explicit in a verb's complementation structure. (b) Lexically~realized optional parameters are semantically implicit, but are not always activated in the actual linguistic expression. (c) Lexically-realized contextual parameters are not syntactically projected, but serve as clues for contextual setting 2.1. LEXICALLY REALIZED GRAMMATICAL PARAMETERS Lexically-realized grammatical parameters are those which have a direct effect on a predicate's complementation structure, and act as a filter to block certain patterns while accepting others. Examples of such parameters are duration, temporal sequence, iteration, inception, achievement, cessation, causation, conation, negativw ity, andfactivity3• TOWARDS A SEMANTIC SYNTAX These factors are important for the simple reason that they are basic to our perception ofthe event that the verb m questwn encodes. When we experience an event, we take note of its "shape". In other words, we perceive when it begins and/or ends, how long it lasts if itis recurrent, what effect it has on us, and if it truly corresponds to the world or a state'of the world. Given that all of these questions are semantic as well as syntactic, it is logical that they should be reflected m a verb's meaning as well as its syntax. 2.1.1. Duration4 . Duration encodes how long an ev~nt lasts in our perception, and is thus the percepuon ofcontmuance. As one ofthe most Importantsynsem parameters it has two subtypes depending on whether the action or process happens over a longer' or shorter period of ttme. This dtstmctwn IS Important because whether or not an action is perceived as durattve or momenta?eous has very definite repercussions on a verb's syntax. The examples given are only a few of many that could have been chosen and illustrate JUSt how pervasive this parameter is in the lexicon because the verbs in question come from domains of quite diverse meanings. 2.1.1.1. Short duration The semantic parameter of short duration activates the following type of complementation in the domains of VISUAL PERCEPTION, CONSUMPTION and FEELING: (a) VISUAL PERCEPTION: The meaning of verbs such as spot, spy, sight and glimpse all contain the parameter of short duratwn. As hyponyms of see, all of them specify the basic meaning of the subdomain in different ways. To see somebody/something at a distance/briefly spot to see somebody/something with difficulty, EFFORT+ DIFFICULTY making an effort to do so. sight to see somebody/something suddenly, EFFORT+ SUDDENNESS making an effort to do so. glimpse to see somebody/something not very· PARTIAL VISION well/partially. spy to see somebody/something, after looking PREVIOUS LOOKING for them for awhile. Much of the complementation acceptable for other perception verbs is filtered o~t here bec~use of the parameter of short duration. A bare infinitive seems to trans mit the meamng ofmomentaneousness, and thus is incompatible with verbs ofVISUAL PERCEPTION such as spot, sight, glimpse and spy because the act of perception does not last long enough to adequately detect the action in the clause: (5) I saw_a movie actor walk across the street. [NP +bare inf.] (6) *I ghmpsed a movie actor walk across the street [NP +bare inf.] (7) I observed that it had become dark. [that-clause] (8) *I spotted that it had become dark. [that-clause] The complementation patterns that characterize this particular subdomain are an NP and an NP + ing-participle. However, there are instances when these complements also become unacceptable, and the reasons are purely semantic. (9) 1 spotted/spied/sighted/glimpsed a movie actor. [NP] (10) *I spotted/spied/sighted/glimpsed a movie actor for a long time. [NP] (ll) I spotted/spied/sighted/glimpsed a movie actor walking down Hollywood Boulevard. [NP ing-part.] (12) *I spotted/spied/sighted/glimpsed a. movie actor walking down Hollywood Boulevard for a long time. [NP ing-part.] Despite the fact that (10) and (I2) take the same complementation patterns as (9) and (11), they are ungrammatical because the adverbial phrase, for a long time, is in opposition to one of the basic semantic values of the subdomain. In contrast, adverbial modification that confirms this value (briefly/for a short time) is perfectly acceptable: (13) I briefly spotted/spied/sighted/glimpsed a movie actor walking down Hollywood Boulevard. (b) CoNSUMPTION: Another example of predicates which contain the parameter of short duration can be found in the domain ofCoNSUMPTION. Verbs of eating and drinking are interesting, not only because such activities are pleasant in themselves, but also because their meanings are largely motivated by social norms and pragmatic factors (Jimenez 1994). Drinking, for example, is an activity that has very definite meaning specifications. One can drink a liquid quickly/slowly and in large/small amounts. Curiously enough, there are no verbs which lexicalize drinking a liquid quickly in small amounts, or conversely, drinking it slowly in large amounts. (14) Consume to use (up) something until there is none left drink to consume liquid, taking it into one's mouth and swallowing it. imbibe to drink alcohol [formal] gulp (down) to drink something very quickly. quaff to drink something quickly [old~fashioned]. swig to drink something quickly in large amounts in a QUICKLY series of big swallows [informal]. IN LARGE swill to drink something quickly and greedily in large amounts AMOUNTS [informal]. guzzle to drink something (esp. alcohol) very quickly, greedily and noisily in an inattractive way. tipple to drink something (esp. alcohol) often secretly and in SLOWLY IN small amounts (informal). SMALL { sip to drink something slowly in very small amounts. AMOUNTS TOWARDS A SEMANTIC SYNTAX Verbs such as quaff; swig, swill and u . duration. This means that they canno't b g zzle all contam the parameter of short this inherent semantic value: e modtfted by adverbtal phrases which violate ( 15) ~1~~~dedd~~~:)q/uaff~d/swig.ged/swill.ed/guzzled four martinis. (16) long time. qua fed/swtgged/swilled/guzzled the martinis for a As manner-of-drinking verbs, uaff, swi . . . on the categorization and evaluati~ fth g, swlll and guzzle are not tehc, but focus carryi~g it out). Drink, the generic ~e~m o~~~:~~~~d~s well as of the person who is enter mto a construction typical of a 1d' . mam, IS the only one that can goa- nected actwn: (17) He drank himself to death (18) * He guzzled/quaffed/swi~ged/swilled/ himself to death. Drink and imbibe can be used without an e . . case, It is still present implicitly as a default ~phctt goalargument, though in that beverage of some sort. In contrast verb . ;ahue, whtch IS mvanably an alcoholic tion, s~ch as gulp, quaff; swig and ~will s wtt tk e meamng parameter of short dura en! object: ' must ta e an NP because they have no inher (19) ~ Pl,easel g/ive me a ~lass of something non-alcoholic I am afraid I on t gu P quaff/ swtg/ swill. · (20) Pl~akls~ giv.e me a glass of something non-alcoholic nn nnbtbe. ·I am afraid I don't d Of course, the goal argument of ul . . (preferably a drinkable one) Yet d g p,thquaff, swtg and swlll must be a liquid rapidly ingest any liquid in this w~ et6'tte e fact that, at least in theory, one can sound very strange in a collocation!;.,·!~ more r:egatr.vely evaluated verbs of this set ones like milk (22). And in () h 1 non-alcoholic beverages, especially healthy 23 activity, the sentence would hav~t:beret such] verbs are used m the sense of a daily able. e s rong y contextualtzed m order to be accept (21) He gulped down/quaffed/sw· go d/ ·n He gulped down/ uaffed/??1 e~o SWJ ed s~veral glasses of champagne. (22) 'J?He o l d j ..~wieged/??swilled several olasses of milk (23) . . ou ps own quaffs/swigs/swills milk every day a~ lunch. . . Indeed, more than encoding the act of dri k' twn of the act as well as the eval t' f h n mg, these verbs encode the percep , ua ton o t e agent. (c) FEELING: Short duration is not only a pro e t f . such as great joy, fear and surprise I~t~i~ o actftons, but also of very strong feeling . , area o meanmg, however, it takes the form of suddenness For example, panic is the way fear can be manifested when one loses control. It is something momentary, which cannot be maintained over a long period of time, and thus does not accept modification by adverbial time clauses. (24) After the accident happened, he panicked. [FEAR] (25) * After the accident happened, he panicked for a long time/all afternoon. Furthermore, in its participle form, it does not refer to the actual manifestation of fear, but rather is a sort of prelude to the actual loss of control (26). (26) I saw him panicking(~ about to panic). The perceiver is the judge of the emotional state of the other person. However, as this type of evaluation cannot be done momentaneously, verbs of visual perception such as glimpse, spot, sight, etc. cannot be used in this context, since they are of short duration and do not allow the perceiver enough time to make this type ofjudgement (27): (27) * I glimpsed him panicking. The same thing occurs with certain feelings of great pleasure (thrill, elate) and surprise (startle), which also have this element of suddenness: (28) She thrilled to his touch. [PLEASURE] (29) * She thrilled for a long time/all day to his touch. (30) His success elated him. [PLEASURE] (31) *His success elated him for a long time. (32) The noise startled him. [SURPRISE] (33) *The noise startled him for a long time. Moreover, as a momentaneous event, verbs with this semantic parameter cannot collocate with aspectual verbs that codify inception, continuance, and cessation. (34) *When they were together, he began/continued/finished to thrill to her touch. In fact they are over so quickly that it is virtually impossible to syntactiCally encode them as a process/state, and thus are not used in participle form: (35) * She spent the day being elated/startled. (36) * She was thrilling to his touch. (37) *I saw her thrilling/ being elated I being startled. (38) * I saw her thrill/ be elated I be startled. 2.1.1.2. Long duration In predicates affected by the parameter of long duration, the event/process takes place over a certain period of time. Such verbs cannot be modified by adverbs which TOWARDS A SEMANTIC SYNTAX violate this basic meaning( e.g. suddenly), and also sound strange in syntactic constructions with the bare infinitive. Verbs of this type are present in a number of domains throughout the lexicon, however the examples below are from the domains of PREPARATION, CHANGE and SPEECH. (a) PREPARATION (food) The verbs of cooking have been studied extensively (Lehrer 1974). The tiny segment below shows those which are affected by the parameter of long duration. (39) cook to be prepared for eating by the use of heat. (of food). stew to cook slowly in liquid in a covered container for long time. simmer to cook slowly in water at/just below boiling. poach to simmer (of eggs/fish). The above verbs of food preparation encode a process which occurs over a period of time, the result of which is a change of state. In the same way as more prototypical verbs of MOVEMENT and CHANGE, they have a zero-derived causative form: (40) I stewed the meat/The meat stewed. (41) I poached the eggs. I The eggs poached. The fact that the action denoted by stew, simmer, and poach takes place over a period of time filters out certain types of adverbial modification like (43) and (44) and even makes certain syntactic patterns ungrammatical ( 46). (42) *He suddenly stewed the meat. ( 43) *The meat stewed for a second. (44) *The eggs suddenly simmered/poached. ( 45) ? I saw the eggs poach /meat stew on the fire. (46) *I glimpsed/spotted the meat stew. In ( 42), the adverbial modification suddenly is in direct opposition to the time parameter basic to the semantics of these verbs. As a result, poach and stew sound rather bizarre in subordinate clauses with the bare infinitive (which to some degree, encodes the meaning of immediacy), whereas they can be used with an ingparticiple: ( 47) I saw the eggs poaching /meat stewing on the fire. They are also -telic, which is shown in these examples in which the tests for telicity are applied: (48) The meat stewed for an hour(* in an hour). (49) * It took Peter three hours to simmer the milk. (50) ? He almost poached the eggs. (b) CHANGE The verbs in this domain encode processes that may also have a time span implicit/explicit in their definitions. (51) to become different by becoming greater develop (jromlihto) to become larger or more complete gradually mature to develop completely ripen to mature, becoming ready to be eaten (of fruit/crops), to be eaten/dnmk (of cheese/wine), or complete (of qualities). For example, when mature and ripen refer to natural processes, no parameter of suddenness (either adverbial or syntactic) can be attached to them. Examples (54) and (55) sound weird enough to project us into the realm of science fiction. (52) *The apple tree ripened suddenly. (53) *The wine matured suddenly. (54) ? I saw the apple tree ripen /wine mature. (55) ? I saw the apple tree ripening. Nevertheless, when these processes are extended to human beings as well as abstract entities, rapid maturing/ripening becomes possible in certain contexts because of metaphorical extension. (56) He suddenly matured after he left home. (57) I saw their plans suddenly ripen. (c) SPEECH Complaining is an area where examples of the parameter oflong duration as well as iteration abound. (58) To complain continuously nag to complain continuously so that somebody will do something. gripe to complain continuously and forcefully (informal). grouse to complain continuously and loudly (informal). bellyache to complain continuously and loudly for no reason (infor mal). In the same way as the previous examples, this parameter filters out certain types of adverbial modification, as well as syntactic constructions. As can be seen in (59), complain, as the generic term of the set, takes the greatest variety of subcategorization: (59) Typology of syntactic patterns --NP that-clause PP (to) PP (about) PP (of) Quote _:omplain !_ + -+ + .. + + + nag + --+ + + gripe + + --+ -+ grose ---+ + -+ bellyache + --- + TOWARDS A SEMANTIC SYNTAX However, there is one pattern which is not included in the inventory ofcomplain. Both nag and gripe can take NP's, whereas complain does not: (60) She was nagging me endlessly.[= annoying] (61) His behavior really gripes me.[= annoying] This is primarily because in this type of context, the meaning of nag and gripe is close to that of annoy in the domain of FEELING. Consequently, the fact that both can take an NP of this type signals a meaning extension to another lexical domain because such a complement is typical of that particular meaning area. As verbs which encode repeated and continuous complaining, nag, gripe, grouse, and bellyache often appear in ing-forms: (62) He kept nagging me endlessly. (63) She was griping about the lack of heating. (64) Stop your bellyaching! It is considerably less frequent for them to appear in a bare infinitive construction such as the following: (65) *I glimpsed him gripe ( 66) *He spotted her nag. (67) ? I saw him gripe/nag. In a similar way, they do not accept adverbial modification such as suddenly or briefly, which violates their basic semantic parameters: (68) *He nagged/bellyached/griped suddenly. These verbs are also -telic: (69) He griped/nag for an hour(* in an hour). (70) *It took John three hours to bellyache/nag. 2.1.2. Temporal sequence The parameter of temporal sequence appears in verbs whose meaning depends on a temporal context For example, the action encoded in all of the verbs listed below is sequential to a previous action/event of the same sort. (a) SPEECH rephrase to phrase something again in a different way. For example, the act of rephrasing can only take place after another action of a similar type, something which makes the following example unacceptable: (71) * He rephrased his question for the first time. Since this type of meaning is signaled by the prefix re-, there are many other possible examples, such as redo, repossess. rewrite, replay, etc., whtch belong to other domains as well as speech. (b) POSSESSION PossESSION is a lexical domain imbued with pragmatic meaning. The act of giv ing is firmly rooted in a complex variety of social relations which stipulate who can give what to whom and in exchange for what. Certam verbs, such as bequeath, also denote when such a transfer of possessions is acceptable. To give something to somebody after one's death leave to give for use after one's death by leaving written instructions to do so. bequeath to leave something, handing it down to them when you go away or die. The use of bequeath generally takes place after the death or disappearance of the possessor of what is being handed over: (72) He bequeathed me his fortune. (73) I bequeath you my fortune (= You shall have my fortune after I am dead.) (74) ?? I bequeath you my fortune, and here it is. For that reason, (73) can be understood as a future transfer of possession, but (74) sounds extremely odd because the possessor is still alive. 2.1.3. Iteration The parameter of iteration encodes a repeated action, something which manifests itself in different ways depending on the particular area of meaning in which it is operating. Examples of this parameter can be found in CONTACT, UsE and SOUND. (a) CONTACT . In CONTACT, this parameter is present in the set of verbs that encode phystcal punishment, all of which basically share the same complementation patterns and argument structure. The selection restrictions mainly pertain to the entity affected by the action. One can beat and batter an inanimate object as well as an animate one, but in the other verbs, the scope of this argument narrows and only accepts living beings. TOWARDS A SEMANTIC SYNTAX To hit somebody/something many times especially to hurt/damage them beat to hit somebody/something hard and continuously over a long period of time. batter to beat somebody/something with great force many times with one's fists/other object. clobber to beat somebody up [informal] whip to beat somebody with a whip as punishment. NP lash to whip somebody (slave, sailor) as punishment NP PP (to death) [old-fashioned]. NPC flog to whip somebody (slaves, sailor) with great force NP PP (with) as punishment [old-fashioned]. flagellate to whip somebody as a religious act of penance (formal). birch to whip somebody with a birch. cane to beat somebody with a cane as punishment. The fact that whip, birch, and cane are denominal verbs derived from the instrument employed to carry out this action eliminates the PP (with) from their inventory, unless the mstrument specified in the phrase modifies the default value because it is of a different type or is more highly specified. (75) * The overseer whipped the runaway slave with a whip. (76) The overseer whipped the runaway slave with the whip in the shed/a cat o'nine-tails. (77) *The schoolmaster caned the boy with a cane. (78) The schoolmaster caned the boy with a long, thin cane. Needless to say, the type of meaning encoded in all of the verbs of the subdomain i~ hitting somebody repeatedly with something. The parameter of iteration is princtpally what connects the verb in question to sociocultural frames of institutionalized punishment. Iteration is always present in this particular set of verbs, and consequently, even when the action is specified as once, it does not mean one stroke of the whip or lash, but rather several ( in sufficient number to constitute one punishment). (79) The overseer whipped him once (=hit him several times with a whip). Another salient feature of these verbs is that the action they encode is goal-directed. Therefore the following constructions are possible: (80) The ships captain beat/battered/clobbered/flogged/etc. the sailor sense~ less. (81) The ships captain beat/battered/clobbered/flogged/etc. the sailor to death. These are all examples of resultative constructions in which the agent (in this case, the ships captain) exercises an action on the patient (the sailor). The resulting state of the patient is codified by an adjectival complement (senseless) or a prepositiona! phrase (to death). This type of construction is very frequent with iterative verbs, in which the repetition of the action results in the attainment of a final state. (b) CONSUMPTION Other examples of the parameter of iteration can be found in verbs like chew, masticate and gnaw, all of which verbs convey the meaning of iterative biting. bite to use one's teeth to cut through something (82) chew to bite something repeatedly usu. in order to taste/eat it. masticate to chew [formal]. gnaw to bite something repeatedly. Chew and masticate differ in degree of formality. The agent argument of both verbs is necessarily an animate being and the goal argument is solid food (soft enough to be eaten). Gnaw is slightly different in that the goal is something hard, which gets worn away little by little through continuous biting. All three of these verbs have an inherent instrument (i.e. teeth). However while the instrument can be lexicalized in gnaw and masticate, it cannot in chew. (83) * He chewed/masticated the food with his teeth. (84) The prisoner gnawed at the rope with his teeth. The parameter of iteration also makes the following example anomalous: (85) * She chewed/gnawed the meat once. (c) SouND Another example of iteration can be found in verbs of repeated laughter, such as giggle, titter, snigger and snicker: laugh to make the sound expressing happiness or amusement. giggle to laugh softly and repeatedly. titter to uigcle nervously, esp. to express embarrassment. iteration snigger ~o ~ggle s.oftly ~n a s~cret,.disrespectful way. { snicker to sntgger tn a htgh-pttched way. In the same way as chew and masticate, verbs of iterative laughter do not appear in collocations with once I one time. (86) * She giggled/tittered/sniggered/snickered one time. This parameter also filters out the collocation with at, which is possible with laugh. At, in this sense, indicates the specific target towards which the agent directs his/her action. TOWARDS A SEMANTIC SYNTAX (87) The children laughed at the teacher. (88) * The children giggled/tittered/sniggered/snickered at the teacher. It is to be noted that in this sense, laugh acquires a meaning which projects it into anothe~ domam.~t becomes more than JUSt sound, and in this sense, signifies mocking or makmg fun of someone/something. It thus becomes an event, something which is impossible for the manner-of-laughing verbsc 2.1.4. Inception .As its name indi~ates, inception refers to the beginning of an event. The most obVIOUS examples of It can be found in the domains of CHANGE and EXISTENCE. (a) CHANOE Verbs of CHANGE are similar to MOVEMENT, though instead of movement from one place to another, they encode movement from one state to another. It is thus an event that occurs over a period of time, and as such, it has a number of phasal distinctions. The followmg verbs of CHANGE encode the parameter of inception: (89) become to begin to be different in the way that is stated. get to become (fairly informal). turn \i~to) to.become something different in nature/quality/con dttwn (with emphasis on the suddenness of the result). gro~ to be~ome something different in nature/quality/condi tiOn (with emphasis on gradualness). go to become, changing to a particular state/condition. In this subdomain, the process of change is slow because in this non-causative sense, time is often the implicit agent: (90) Her parents became old. This filters out constructions in which this parameter is violated: (91) * Her parents stopped becoming old. In _fact, even when the implicit agent is obviously not time, constructions with stop still sound strange because the process of change in this initial phase is conceived as something that really cannot be stopped at will: (92) The frog turned into a prince. (93) ?The prince stopped turning into a frog. (b) EXISTENCE Lexical structure in the domain of EXISTENCE reflects the duality of our understandrng of this concept. Accordingly, something can exist (for us) in two ways: more generally because we believe it has a counterpart in the outside world, and more specifically, because it enters our field of perception. Therefore, existence in the objective world runs parallel to existence in the perception of others. In both types of existence, life can be conceptualized as an event, a temporal segment that can be broken down into a number of phasal distinctions such as beginning, continuing and stopping. The following lexical set includes those verbs which belong to the phase of inception (Faber and Mairall997: 135-6): (94) appear to begin to exist in the perception of others. dawn to begin to appear [formal] surface to appear, becoming obvious or known, especially after a period of remaining unseen. materialize to appear, taking bodily form/becoming reality. form to appear, beginning to be visible and having a clear shape/ outline. This subdomain is closely linked to that of PERCEPTION, since when something appears, it begins to exist (not in itself), but for that perceiver. Something can appear because it moves into somebody's field of vision, or it can appear because of a change in contextual condition which facilitate visual perception (i.e. sufficient light, absence of barriers). Consequently, the above lexemes refer to a process ofchange which takes place in the perceiver who moves from a state of unawareness to awareness of the perceived entity. In the phase of inception, this means that such verbs cannot appear in constructions which go against this basic synsem parameter: (95) *The idea stopped dawning/materializing. (96) * The outline of the ship stopped forming on the horizon. 2.1.5. Cessation In comparison with inception, cessation is much more extensively lexicalized because ending seems to be more perceptually salient than beginning. Examples of this parameter can be found in EXISTENCE, LIGHT, and POSSESSION. (a) EXISTENCE Cessation is encoded in to stop existing in the perception of others, which includes the following verbs, all of which describe the state of affairs of not being able to seem something any more: (97) disappear to stop existing the perception of others fade to disappear gradually from sighUhearing/memory, often remaining to some extent. vanish to disappear suddenly and completely, especially in a way that cannot be explained. dissipate to disappear completely, as if by scattering. evaporate to disappear like vapor. TOWARDS A SEMANTIC SYNTAX In vanish and dissipate, the disappearance is complete, so what is lexicalized is the manner of disappearance (suddenly, as ifby scattering). Interestingly enough, in vanish there is also a secondary connection with CoGNITION in that the suddenness of the process causes a state of perplexity in the perceiver. The fact that a sudden disappearance is something that needs an explanation signals a default value in this subdomain regarding the relative speed with which something is expected to move outside the scope of our perception in order to be considered a normal state of affairs. In contrast, in fade, the process is so gradual that there is usually some trace left over a period of time. All of these verbs are intransitive and there are virtually no selection restrictions on the argument. Unlike inception, cessation does allow constructions with begin. In other words, one can talk about the beginning ofthe end much more easily than the end of the beginning. It is only ungrammatical in vanish primarily because of the parameter of achievement. (98) The bruise on her arm gradually began to fade. (99) His enthusiasm for a new car began to dissipate when he saw what it would cost. ( l 00) Her desire for him began to evaporate as his real personality came to light. (101) *The spot on the carpet began to vanish. Alternatively, however, these verbs are ungrammatical in constructions with stop because in the examples below, cessation in fade, dissipate, evaporate and vanish is cancelled out rather than reinforced. (102) *The bruise on her arm gradually stopped fading. (103) *His enthusiasm for a new car stopped dissipating when he saw what it would cost (104) *Her desire for him stopped evaporating as his real personality came to light. ( 1 05) * The spot on the carpet stopped vanishing. (b) CHANGE with secondary connections with LIGHT put out to cause something to stop burning. extinguish to put out something (lighUfire) [formal]. quench to extinguish a fire with water [old-fashioned]. smother to extinguish a fire by covering it. (106) ?They started to quench/extinguish the fire. (1 07) * The water started to quench/extinguish the fire. ( l 08) ?They started to smother the fire. (I 09) * The blanket started to smother the fire. Cessation in this sense also has an important element of achievement. The event in this case is also affected by short duration or momentaneousness. Consequently, it cannot be perceived as having a beginning and an end, but only an end. 2.1.6. Achievement Achievement is present in various areas of meaning and is manifest in the following type of construction. The following resultative constructions have unconscious, sick, flat, blind and silly as culminating predicates. (110) She beat him unconscious. [CONTACT) (111) The thought of his going away worried her sick. [FEELING] (112) They knocked him flat and stole his wallet [CONTACT) (113) Roldan robbed the government blind. [PossESSION) (114) She laughed herself silly. [SOUND) As Pustejovksy (1995: 15) observes, these can be classified as stage-level predicates. These contrast with individual-level predicates, which are properties that an individual possesses more or less throughout a lifetime (Carlson, 1977 and Kratzer, 1988 apud Pustejovsky, 1995). Such predicates cannot appear in resultative con-. structions: (115) *John ate himself handsome. (116) *Paul worried himself tall. Another type of resultative construction can be seen in the following examples where death, insanity and exhaustion are conceptualized as the final destinations of the affected argument. (117) They starved him to death. [EXISTENCE) (118) She drank herself to death. [CONSUMf'TION] (119) The Lone Ranger galloped Silver to exhaustion. [MOVEMENT) (120) She nagged him to death. [SPEECH) ( 121) His indifference drove her to insanity. [MovEMENT] Certain of these verbs belong to the domain of MOVEMENT, but others such as starve, drink, and nag belong to other domains in which such movement is understood more figuratively. Evidently, achievement can be expressed in a variety of different ways. Although all of the examples refer to completed action, they are different in that in some of them, the action referred to affects the first argument and in others the second. TOWARDS A SEMANTIC SYNTAX 1st Argument 2nd Argument -· CONTACT *They knocked themselves flat They konked him flat * She beat herself unconscious She beat him unconscious - CONSUMPTION She ate/drank herself to death. * She ate/drank him to death. FEELING She worried herself sick. She worried him sick. POSSESSION * Roldan robbed himself blind. Roldan robbed the government blind r--- EXISTENCE She starved herself to death. He starved her to death. --,---~~~·-~ MovEMENT He ran himself to death. He galloped his horse to exhaustion. ··SPEECH *She nagged herself to death. She nagged him to death. -~~~ - SOUND She laughed herself silly. She laughed him silly. 2.1.7. Causation Virtually all of the dimensions in MOVEMENT have causative subdimensions in which the same verbs appear: MovEMENT IN LIQUID (NON-CAUSATIVE) MOVEMENT IN LIQUID (causative) · To move/go downwards To cause to move downwards dive 1 to go into the 'water, jumping head--., dive2 to cause somebody/something to go first with one's arms straight above into the water. one's head. plunge] to dive quickly and force--., pltmge2 to cause somebody/something to fully. go into the water quickly and forcefully. sink] to go slowly downwards below the-~ sink2 to cause somebody/something to go surface of the water. slowly downwards below the surface of the water. submerge 1 to go under the surface of a liq--7 Submerge2 to cause somebody/something uid, esp. at a planned speed. go under the surface of a liquid, esp. at a planned speed. immerse to cause somebody/something to go down into liquid, covering them/ it completely. dip to cause somebody/something to go down into liquid for a short time. dunk to dip something into a liquid for a particular purpose. (a) MOVEMENT (in liquid) Many of these verbs can be used both transitively and intransitively. (122) He plunged/sank/dove into the water. (123) He plunged/sank/dove his hand into the water. Dive can be considered a special case in that its second argument is more semantically specified than that of the other verbs, and is necessarily a part of the body: plunged } (124) He ~ank the knife into the water { "'dove Some of the verbs within this particular area of meaning, however, are only causative, and cannot be used intransitively, at least, in their sense of movement inion liquid. (125) *The ship immersed/dipped/dunked. Many of these verbs also have other meanings which refer to generic downward movement. With the exception of dunk, immerse, and submerge, the others (dive, plunge, sink) can be used for general movement as well as movement in air. (126) The seagull plunged into the ocean (127) The missile plunged downward (in the air). (128) The drill plunged into the earth. Sink, as the generic term, can be used to designate downward movement in wide variety of contexts: (129) Ralph sank into the water/chair/mud. However, the fact that its prototypical sense is movement in liquid can be seen in the following example, where the medium is unspecified and the default value is obviously in liquid: (130) He sank the ship/* plane/* drill. (b) L!GHT A second example of a subdomain marked by a causative parameter can be found in verbs of LIGHT. In the preceding subdomain, causation was more evenly balanced with non-causation. However, the domain of LIGHT is quite different perhaps because of its generic semantic value in which light is emitted from a natural source. This is a good example of a domain of intransitive verbs which have no zero-derived causative TOWARDS A SEMANTIC SYNTAX form. With the exception of shine and flash, which are in both subdomains the other verbs are only in one. ' LIGHT (NON-CAUSATIVE) LIGHT (CAUSATIVE) to give offlight To cause sth to give off light Shine I to give off lighllto be bright. '""Shine2 to cause something to give offlight/ glow to shine with a soft, warm light (of to be bright. a low fire, stars). blaze to blaze suddenly. flare to blaze suddenly. flash] to shine with a sudden, bright j-7 flash2 to cause something to shine with light (esp. quickly and regu-a sudden, bringht light (esp. larly). quickly and regularly). glint to shine brightly, giving off small brighten to cause something to shine flashes of light (esp. a metallic more. surface). illuminate to cause something to be sparkle to shine brightly giving off bright by shining light on it. small, quick flashes of light light to cause something to begin to give (esp. jewels, stars). off light/be bright. twinkle to sparkle continuously, changing from bright to faint (esp. stars). scintillate to sparkle ( esp. of wit, personality) (formal] [literary]. glitter to sparkle in a flashy, artificial way. dazzle to shine in somebody's eyes with a sudden very bright light, making them unable to see. glare to shine with a very bright light that is difficult to look at. flicker to shine unsteadily with sudden changes in brightness. shimmer to shine with a soft, changing light (esp. a moving surface such as water, silk, etc.). glimmer to shine with a faint, unsteady light. gleam to shine brightly by reflecting light (esp. a clean, shiny surface). glisten to shine brightly (esp. a smooth/ wet/oily surface). beam to shine brightly and constantly (esp. in a certain direction). p. In the causative subdomain, an agent causes a light source to give off light or alternatively, causes a space to be bright. It follows that these verbs are those which have a transitive use, whereas the verbs in the parallel subdomain are one-place predicates which do not: (131) The janitor shone/flashed the flashlight from the window. (132) *The janitor twinkled/flickered/blazed/flared/sparkled/etc. the flashlight from the window. 2.1.8. Factivity The parameter ofjactivity emphasizes the different degrees of the thinker's commitment towards the truth or falsity of a proposition within the complement phrase. There are a range of subjective, epistemological operators that act as the basic structuring parameters of an important sector of the domain of cognition, organizing its subdomains so that the verbs within them encode different degrees of certainty'. This commitment on the part of the speaker has important syntactic repercussions. For example, a semantic value such as Certain (the speaker's evaluation that the proposition holds in every possible world) is in consonance with topicalization, a syntactic· construction involving the fronting of a non-subject NP interpreted as the top to a sentence-initial position. This topic fronting applies to definite NPs, and thus, is compatible with this type of operator, which offers no alternatives. However, it is incompatible with others which apply other available options, as can be seen in the following examples: (133) That she did a good job, he regrets. (134) That he left, she knows. (135) That she played well,* she believes. (136) That he go there, she demanded. (137) That he go to the dentist, *he suggested. As can be observed in example (133) and (134), topicalization is possible because the speaker's evaluation of the propositions (that she did a good job/that he left) are modified by the operator Certain. However, this is not the case in examples (135) and (137) because the complement proposition is not modified by this operator. In essence, a syntactic construction involving definiteness tends to occur with a type of meaning that offers no alternative. A subjective epistemological operator like Probable, which signals the speaker's evaluation that the proposition is expected to be the case, is in accordance with syntactic transformations known as "root transformations" (Hopper and Thompson, 1973). Such transformations cannot operate on questions, reduced clauses, and presupposed clauses, since it would not make any sense to emphasize constituents in a sentence whose proposition is presupposed or already known. From this, it follows that these syntactic constructions will be compatible with a probable meaning and its possible alternatives. This type ofmeaning will therefore be sensitive to constructions, such as complement preposing, VP preposing, etc. TOWARDS A SEMANTIC SYNTAX (138) He wants to leave the country, he says. [complement preposing] (139) The trade unions are obsolete, the president concluded. [complement preposing] (140) The group vows that disrupt the elections they will. [VP preposing] (141) Ken announced that marry Jane he would. [VP preposing] 2.1.9. Conation This parameter encodes the speaker's attempt to carry out an action. It is lexically marked by try, which appears as part of the genus of all of the verbs. In the same way as other types of semantic markers of this sort (start, stop, continue, etc.), it can either appear as the genus itself (142), or alternatively as a modifier of the genus (143) and (144). (142) try to make an effort to do something. attempt to try to do something difficult endeavor to try very hard to do something (formal) strain to try very hard, especially making a great physical effort strive to try to do something, making a very great effort over a period of time. The focus of this semantic value is thus on the initial phase of the activity without explicitating whether the action is actually carried out. It is explicitly marked by try, and appears marked in a variety of different areas of meaning. (143) coax to try to persuade sb to do sth in a gentle, pleasant way. (144) hawk to try to sell something by shouting in the street I taking it around to various people (informal) The presence of try as a semantic constituent of the definition has important syntactic consequences insofar as it encodes a -telic state of affairs, and consequently, this sort of construction is compatible with the test for telicity (cf. Dik, 1998: 92-94). (145) I coaxed him for an hour(* in an hour). (146) *It took me three hours to coax him. (147) I hawked the jewelry for an hour(* in an hour). (148) *It took me three hours to hawk the jewelry. However, a change in the complementation ofcertain verbs brings about a change in the state of affairs, as can be observed in the following examples: (149) I coaxed him into going to the store /to go to the store. (150) *It took me three hours to coax him to go to the store. Here it can be seen how the inclusion of an infinitive/prepositional phrase can cause the canonical meaning of coax to vary6 and acquire the meaning of achievement. However, the same is not true for hawk which does not accept this type of complementation. 2.1.2. Lexically realized optional parameters Lexically-realized optional parameters explain why certain arguments, though semantically present, are not syntactically prominent in the actual linguistic expression. The most obvious examples ofthis can be found in PossESSION, in the subdomain which encodes transfer: (151) I bought/purchased a new car (152) I bought/purchased a new car from Tom (153) I bought/purchased a new car from Tomfor Mary. (154) I bought/purchased a new car from Tom for Mary for $5000. In the above examples, the only argument that is strictly necessary for the syntax of the sentence is a new car. The others are all potential in the structure of this type of verb, but not necessarily activated all at once. In fact, when they all appear together, the result even sounds awkward. The arguments in the above examples are true arguments because they are syntactically realized parameters of the verb in question. However, Pustejovsky (1995: 63-4) also distinguishes between default arguments and shadow arguments. Default arguments participate in the meaning, but are not necessarily expressed syntactically: (155) She made the wedding dress out ofsilk. (156) The second little pig built his house out ofsticks. Shadow arguments are semantically incorporated in the lexical item. As we have seen, they can be expressed only by operations of subtyping or discourse specification: (157) *The overseer whipped the runaway slave with a whip. (158) The overseer whipped the runaway slave with the whip in the shed/a cat o 'nine*tails. (159) *The schoolmaster caned the boy with a cane. (160) The schoolmaster caned the boy with a long, thin cane. 2.1.3. Lexically realized contextual parameters Lexically realized contextual parameters, though not syntactically projected, serve as clues for contextual setting. They are of different types and elaborate the schema in contrasting ways (location, instrument, intensity, movement). Examples of this are in the domains of POSSESSION, VISUAL PERCEPTION, POSITION, and SPEECH: (a) POSESS!ON Peddle and hawk in the domain of PossESSION activate a schema of location and of movement because in both cases, the activity is can·ied out while moving from one place to another. TOWARDS A SEMANTIC SYNTAX (161) To give somebody something in exchange for money sell to give something in exchange for money. vend to sell [formal]. peddle to sell small things by going fr?m place to place. . . { hawk to try to sell something by shoutmg m the street I takmg It around to various people [informal]. Given their basic semantic parameters, it thus sounds strange for someone t? eddie/ hawk something in a department store. Nor is it normal to hawklpeddle a pair ~f matched Rolls Royce or a mansion in Malibu. (162) The traders hawked their wares on the street corners. (163) ?The shop attendant hawked the perfume in the department store. (164) The man peddled his wares from door to door. (165) ?The man peddled his wares in th~ department store. (166) ?He went from door to door peddling Rolls Royces. Whereas examples (163), (165) and (166) sound very strange, they are possible. However, the parameter that cannot be violated is that of sound m hawk: (167) The traders loudly hawked their wares in the market place. (168) *The traders silently hawked their wares in the market place. (b) VISUAL PERCEPTION . . Another example of contextual activation can be found m the domam of VISUAL PERCEPTION. Most of the manner-of-staring verbs below are very general and tell us more about the perceiver than the what is being percerved. For example, goggle, gape and gawk reflect negatively on the intelligence of the expenencer. Glare and glower give us information about his emotional state. However, ogle can be smd to activate a specific context. stare to look at somebody/something for a long time with wide open eyes. goggle to stare at in surprise. gape to goggle esp. with an open ~outh. gawk to stare in a stuP,id, unthin~ing way [mfonnal]. glare to stare angrily, man unfnen?ly way. glower to glare for a long time. \ ogle to stare with se~ual interest] The semantic parameter of manner codified in this verb e~okes a very definite context, which greatly narrows the scope of the goal argument. (169) The soldiers ogled the girls sitting at the bar. (170) The ladies ogled the male striptease danc~r. (171) The boys ogled the pictures in the magazme. The reason for this is that with the exce tio -- patterns, sexual interest of the Agent is g p 11 n~~ certam unprototyp!cal behavior the opposite sex_ A non-human ar umen~ni~~ y !fected towards similar entities of dependmg on the type of animal in~olved d h~s posltlon IS margmally acceptable, ness can be attributed to it. , an t e extent to whtch human conscious (I 72) The male gorilla ogled the I :ookeepers had put in his cag:ew y-arnved female gorilla that the (173) . The male cockroach ogled the newly-arrived female cockroach. The selection restrictions of the ar ume h cally odd/unacceptable sentences suchg , thnt~ tilus block the occurrence of semanti as e 10 owtng: (174) : Peter ogled the aardvark/tarantula. (175) The aardvark/tarantula ogled Peter. (I 76) ?Peter ogled the beautiful aardvark. (177) : Peter ogled the beautiful tarantula. (178) . Peter ogled the table for a long time (179) *The table ogled Peter · (I 80) * Peter ogled the beautiful table. However, if the aardvark in exam les (174) man-like in the sense ofpossessin ~---, (175), and (176) is made more huof a scale of human attractivenes; ';~a !ties whwh contnbute to a positive evaluation interpreted as a kind of marginall , en~lxample (176), though odd, can at least be modification of acceptability occJr!i~s~;7;)~ehavwrpattern__ In contrast, no similar consciousness (eyes-ears-nose schema hum ec~use It !S difficult to attribute human Wise examples (178-180) d : an-hke reactiOns) to a tarantula Like - o not expenence any m dT -- ammate entity (albeit an attractiv ) -o I !catwn either because an in- e one IS generally not an object of sexual interest (c) POS!T!ON Inthe domain of PosmoN, verbs such as lod cohabu Wlth the base meaning ofto I -ge, board, squat, camp, bivouac and All of these verbs have live as th say m one place, all belong to the same subdomain --- e1r genus and the d --- de tmtwnai structure refer f to time . d ' escuptive parameters in their with the subject: peno ' payment, location, and person(s) living (181) live /o stay s~mewhere, having it as your home. od~~~~-hve somewhere esp. for a short period of time and paying board to live somewhere (usu in sb's h .. return for payment. · orne and rece1vmg meals) in cam:;. to live somewhere for a short time in a tent/caravan tvouacat~e~~~p in a bivouac (a temporary shelter whi~h is not squat to live somewh ( , out having a I , ler~ husu. pubhe land/unused building) with ega ng t to do so. TOWARDS A SEMANTIC SYNTAX room to live in a room/rooms in sb else's house. cohabit to live together, having a sexual relationship (of unmarried people) [formal], · shack (up) to start to live together, having a sexual relationship (of unmarried people) [informal]. The lexical contextual parameters within this subdomain are the following: (182) Lexical contextual parameters L Short period of time [During]: lodge 2. Payment [in exchange for]: lodge, board 3. Particular location [at/in]: camp, bivouac, room, squat 4, Company [with]: cohabit, shack up Although semantic parameters codifying sociocultural norms are different, depending on the domain and subdomain they are embedded in, it is relatively easy to deduce what such norms are because they are precisely what is not lexicalized_ In language, we tend to find words for what draws our attention, and what surprises us is generally what differs from our expectations or deviates from the socially-accepted norm, Squat, for example, activates a whole sociocultural context of property rights and others who make use of that property against the law_ From its original sense as a verb of body position, more specifically a rather unorthodox kind ofsitting (on one's heels instead of on a chair/ground), squat has a meaning extension to another subdomain, to stay at a particular location, In this case the location is an unused building/land which the subject occupies illegally (without permission and without paying rent)_ (d) SPEECH Biological norms are also codified in the lexicon, An example of this can be found in one of the subdomains of SPEECH: (183) to soy sth (speak) with difficulty stutter to speak with difficulty (with short stops one cannot control). stammer to speak with difficulty (hesitating and repeating sounds and words), lisp to speak with difficulty (using "s" sounds which are not clear). bumble to speak with difficulty in an awkward, uncontrolled way. This subdomain encodes deviations from the biological norm_ If we stutter, stammer or lisp, it is more difficult to communicate because of physical impediments, This kind of speech is negatively evaluated, not so much for its content, but for the quality of its sound, which is the result of the speaker's lack of control of the speech act Consequently the following examples are unacceptable: (!84) *He stuttered/stammered/lisped/bumbled welL The definitions of these verbs tell us that speech should prototypically be fluid without stops or hesitations, Sounds should not be repeated, and phonemes (such as I s/)dshould not be unnecessarily emphasized. The norm that dominates all of the pre ce mg ones rs that of speaker control. (e) SPEECH: codification of power relations Another important type of _norm is codified in the power relation between arguments, as well as the cont~xt thrs action can take place in, something which is clearl dehmtted Wtthm tts meanmg. A case in point can be seen in the predicate upbraid: y (185) To say that somebody has done something bad scold to .say somebody has done something bad, criticising them an~ gnly. chide to scold somebody gently [old-fashioned]. /upbrazd to scold somebody for a long time (formal]./ sermomse.to scold sorn~body, giving them unwanted moral advice. morabse to sermon1se somebody, telling them what is ri(rht/ 0 wrong. berate to scold somebody loudly [formal]( old-fashioned]. repnmand to scold somebody severely and officially. Upbraid has an agent, a patient, and optionally a reason for scoldin someone in that way. However, the argument slots can be filled by different types 0~entities: Agent p t' . . . (186) The teacher } { a lent Actwn: actlvtty, behavior (187) He t~e st?dents thetr poor marks (188) ? The soldiers upbraided hts Wife } for {h~rextravagance (189) ? The children the general his bad temper then parents cornmg home late A ~~~plesh(186-189) show us that one of the selection restrictions for arguments an lS aut onty. In e~amples (1 86) and (187), the power relation between teacher-students and husband-wife gtves A the right to upb ·dB I · dd d · ' raz . n contrast, example (189) rs ~-.' an m need of contextualization because children do not prototypically have su '?tent authonty to upbraid their parents. The same is true for the inverted 0 relatton betwe~? soldiers and the general in (188). The third argument (C), the fea;;~ for bemg upbrmded, ts mvanably a behavwr/activity/attr'tude wht'ch A· t · me t 'th F h · rs no m agree- n WI · urt ermore, m contrast with other verbs in the same subdom · t grounds for being upbraided are usually definite and justifiable (poor ma aln, he travagance, lack ofdiscipline) etc. r s, ex (200) ?The pharma?ist upbraided his children for always attending church (201) The pharmaciSt berated his children for always attending church. · ~!though bo~ upbraid and berate unprototypically have a positively valued elemen1 mpo~ttwn ,(200) ts less acceptable than (201) because upbraid has a negative axw ogtca. evaluatwn bmlt-m to lts third argument, whereas berate does not I sonance wlth the whole subdomain, the activity lexica!ized in all these verbs·· n conliVely valued, thus the unacceptability of the following example: ts nega- TOWARDS A SEMANTiC SYNTAX (202) *He upbraided/berated/chided her well. 3. CONCLUDING REMARKS In this article, we have shown that in a lexical subdomain, semantic parameters constrain and filter syntactic projections. Synsem parameters, such as duration, iteration, temporal sequence, achievement, conation, andfactivity are a determining factor in the actual process of constructing an underlying clause structure. Obviously, this brings meaning definitions to the forefront in FG because semantic information is projected onto syntax. In this way, it becomes evident that the semantics encoded in a lexical subdomain is in consonance with its syntactic potential. Accordingly, a verb's complementation is not random, but is systematically coherent with its semantics. The distribution and activation of synsem parameters is also principled in that the greater the semantic scope of the lexeme, the greater its syntactic variation. The configuration of both the semantic and the syntactic information is hierarchical, and the two resulting hierarchies are not independently motivated, but constrain and interact with one another. Notes 1. This research was carried out within the framework of the project, Desarrollo de una l6gica lixica para la traducci6n asistida por ordenaddr a partir de una base de datos lixica ingles-francis-alemdn-espafiol multifuncional y reutilizable, funded by the DGICYT, code n" PB 94/0437. 2. For a detailed presentation of this lexicological model, we refer the reader. 3. The inventory of features presented here is by no means exhaustive. 4. For a complete discussion of the role of semantic parameters in the lexicon, see Martin Mingorance (1985b) . More specifically, we refer the reader to Felices (1991) and Portero (1997) for thoroughout discussions of specific semantic parameters like axiology and intensification respectively. 5. The following discussion is based on Hopper and Thompson (1973) and Ransom (1986). 6. As is well known this falls under the predicate formation rules in FG. However, our approach somewhat defers from this insomuch as we believe that this type of variation can be captured in terms of a set of lexical rules. For further discussion, see Faber and Mairal (forthcoming). Work Cited Dik, S.C. Stepwise Lexical Decomposition. Lisse: Peter de Ridder Press, 1978. -The Theory ofFunctional Grammar. Part I: The Structure ofthe Clause. Dordrecht: Foris Publications, 1989. Dixon, R.M.W. A New Approach to English Grammar on Semantic Principles. Oxford: Clarendon Press, 1991. Faber, P. and R. MairaL "The Paradigmatic and Syntagmatic Structure of the Lexical Field of EXISTENCE in the Elaboration of a Semantic Macro net." Studies in Language 2l.l (1997): !19-154. -"Lexical Structure and Syntactic Projections: Towards the Design of a New Linking Algorithm." Eighth International Conference on Functional Grammar [Forthcoming]. Felices, A. El componente axiol6gico en ellenguaje: Su configuraci6n en los adjetivos / que expresan emociones y conducta en la lengua inglesa. Granada: Servicio de Publicaciones de la Universidad de Granada, 199!. Hopper, P.J. and S.A. Thompson. "On the Applicability of Root Transformations." Linguistic Inquiry 4.4 (1973): 465-498. Jimenez Hurtado, C. El componente pragmdtico en el lexicon verbal del espaiiol, aleman e ingtes. Universidad de Granada, Diss. 1994. Levin, B. English Verb Classes and Alternations: A Preliminary Investigation. Chicago: U of Chicago P, 1993. Martin Mingorance, L. "Lexical Fields and Stepwise Lexical Decomposition in a Contrastive English-Spanish Verb Valency Dictionary." Lexeter '83 Proceedings. Ed. R. Hartmann. Tlibingen: Max Niemeyer, 1984. 226-236. -"La semantica sintagmatica del adjetivo. Parametres para la organizaci6n de un lexicon ingles/espafiol de valencias adjetivales." Aetas del II Congreso de A.E.S.L.A. (1985a): 329-340. -"Bases metodol6gicas para un estudio contrastive dellexico derivado." Revista Espanola de LingUistica Aplicada l (1985b): 37-54. -"Classematics in a Functional-Lexematic Grammar of English." Aetas del X Congreso Nacional de AEDEAN. (1987a): 377-382. - "Semes, Semantic C1assemes, and Dimensions: The Lexicological and Lexicographic Perspectives." X!Vth International Congress of Linguists. (l987b): 10 15. -"Pragmatic Features in the Lexicon of a Functional Grammar." International Prag matics Conference (l987c): 17-22. -"Functional Grammar and Lexematics." Meaning and Lexicography. Eds. J. Tomaszczyk & B. Lewandowska-Tomaszczyk. Amsterdam: John Benjamins, 1990. 227-253. -"Lexical Logic and Structural Semantics: Methodological Underpinnings in the Structuring of a Lexical Database for Natural Language Processing." Panorama der Lexikalischen Semantik. Ed. Ulrich Hoinkes. TUbingen: Gunter Narr, 1995. 461-474. Pinker, S. "How Could a Child Use Verb Syntax To Learn Verb Semantics?" Lingua 92 (!994): 377-410. Portera, M.C.lntensificaci6n: estudio de un clasema. Universidad de COrdoba, Diss. 1997. Pustejovsky, J. The Generative Lexicon. Cambridge (Mass.): The MIT Press, 1995. Ransom, E. Complementation: Its Meaning and Forms. Amsterdam: John Benjamins, 1986. Tomaszczyk, J. and B. Lewandowska-Tomaszczyk, eds. Meaning and Lexicography. Amsterdam: John Benjamins, 1990.