When it comes to verb meaning, everyone believes there is a difference between structural meaning (what I will call A-type meanings) and lexical conceptual meaning (what I will call B-type meanings). Researchers vary according to whether A-type meanings and B-type meanings are (i) divisions internal to cognition which are then reflected in lexicalization, (ii) divisions internal to a linguistic module which structures cognitive constructs for lexical labelling (the lexicon), or (iii) whether they belong to different modules altogether (syn-sem for type A meanings, and cognition for type B meanings).
Type A Meaning (‘Skeleton’) :
A hierarchically structured representation of abstract actional factors that are directly correlated with linguistic generalizations concerning argument structure realization in the syntax. This type of meaning is often assumed to be either cognitively or linguistically universal, although this is not necessary.
Type B Meaning (‘Flesh and Blood’) :
Encyclopedic and conceptually rich information that provides detailed expression to highly specific named events. It is always unsafe to assume that this type of meaning package is universal although it is drawn from common human cognitive proclivities.
This still leaves a number of open questions.
1

• (I) The Separation Problem : Which meaning components are part of the the Type A skeleton and which not?
• (II) The Lexical Specification Problem : Does a lexical root carry only information of Type B, or information from both domains?
If the former, then some mechanism of subcategorization or selection is required to put roots in particular frames since verbs are not fully elastic.
• (III) The Binding Problem : How do the two types of meaning combine to create verb meanings?
Of course, the answer one gives to (II) will be bound up in the answer one gives to (III). Approaches can be divided into information which is derivationally combined in series , or in parallel , with a further independent difference with respect to whether information is thought to be encapsulated within different modules. Here is a tentative classification of the different prominent views out there:
Lexicon Internal Unification : Type A and Type B meanings reside within the lexicon and are combined in parallel (i.e. unified) prior to insertion in the syntax. ( parallel combination, but Lexicon in series with Syntax)
Derivational (Syntactic) : Type B meanings alone are encoded in the lexical root and this is combined in series with Type A meanings encoded in the syntax (i.e. with the root at the bottom of the tree). ( Type B in
Lexicon, Type A in syntax, the two are combined in series )
Cross-Modular Unification : Type A meanings are encoded in the syntax, Type B meanings are represented in the conceptual-intensional part of mind/brain. Lexical root contains both types of information; combination proceeds by unification (in parallel ) at every stage of the derivation.
I assume the following as the subpart of the functional sequence that corresponds to the notion of ‘dynamic verb’ in earlier phrase structure models, after Ramchand (2008b). Each projection corresponds to a subevent with its own predicational subject position, and linked by the generalized ‘leads-to’ or ‘cause’ relation.
2

(1) init P ( causing projection)
DP
3 subj of ‘cause’ init procP (process projection)
DP
2 subj of ‘process’ proc res P ( result proj)
DP
1 subj of ‘result’ res
• init P introduces the causation event and licenses the external argument
(‘subject’ of cause = Initiator )
• proc P specifies the nature of the change or process and licenses the entity undergoing change or process (‘subject’ of process = Undergoer )
• res P gives the ‘telos’ or ‘result state’ of the event and licenses the entity that comes to hold the result state (‘subject’ of result = Resultee ) .
The semantics of structure delivers entailments about force dynamics, scalar structure and event participancy; fleshly meanings are given by lexical items with conceptual content. Lexical items come specified with categorial features that specify which parts of the skeleton they are designed to conceptually specify, so essentially they define a set of cross modular associations.
Conceptual content needs to be smeared all over the structure to express a specific event in the world because structural meaning is only an abstraction and can never be actualized in the absence of real flesh and blood.
(2) Exhaustive Lexicalization 1
Every node in the syntactic representation must be identified by lexical content.
(3) Non-Terminal Lexicalization :
Lexical items are bundles of conceptual information specified with a
1 The name and formulation of this principle emerged from collaborative conversations with Antonio F´abregas. See Fabregas (2007) for extensive discussion of its effects in the domain of Spanish directional complements.
XP
3

(continuous) set of categorial features which determines points of meaning unification with syn-sem structure.
(If we get this right, it can account for both flexibility and selectional rigidities)
Root and Event Schemas from Rappaport-Hovav and Levin (2008)
(4) manner −→ [ x act manner
]
(e.g.
jog, run, creak, whistle, . . . )
(5) instrument −→ [ x act instrument
]
(e.g.
brush, chisel, saw, shovel , . . . )
(6) container −→ [ x cause [ y become at < container > ] ]
(e.g.
bag, box, cage, crate . . . )
(7) internally caused state −→ [ x state ]
(e.g.
bloom, blossom, decay, flower, rot, rust, sprout , . . . )
(8) externally caused, i.e. result, state −→
[ x [ act ] cause [ y become < resultstate > ] ]
(e.g.
break, dry, harden, melt, open . . . )
(9) The lexicalization constraint: A root can only be associated with one primitive predicate in an event schema, as either an argument or a modifier.
“In English, most words are morphologically simple as there is no developed notion of stem: thus, manner/result complementarity is manifested in words. In contrast, in languages in which verbs are productively formed from stems and affixes, manner/result complementarity holds of the pieces of words, rather than the words themselves.”
4

They also claim there is a ‘manner/direction’ complementarity akin to
‘manner/result’ complementarity. Levin and Rappaport Hovav take directed motion verbs to be a type of result verb.
We need to ask ourselves whether the patterns here are absolute or tendential.
Detailed empirical work suggests to me that they are tendential. The reason we need to be careful about hardwiring this into a theory of Type A Type
B combinatorics is that there are a number of plausible functional reasons why there should be this tendency. I quote from Talmy on the different pressures that would conspire to make a language’s lexicon structure itself around a medium size of conceptual specificity, with some productive abstract combinatoric elements.
“If, indeed, the pattern with lack of conflation occurs rarely or never as the main system of a language, one explanation may be its relative inefficiency. The pattern calls for the reexpression of the same morpheme with the same fixed meaning . . . ” (Talmy
2007 pg 102)
“Conflation systems of this multi-component sort apparently never form a language’s major system for expressing Motion. The reason for such a prohibition seems straightforward for any system that would undertake to make relatively fine semantic distinctions: it would require an enormous lexicon. ”
In fact, Levin and Rappaport-Hovav end up restating manner-result complementarity in terms of scalar vs. non-scalar change, a move which possibly shifts the distinction from the B domain to the A domain, changing the nature of the restriction considerably.
“In summary, we have identified result verbs as verbs which lexicalize scalar change and manner verbs as verbs which lexicalize non-scalar change (and specifically, complex change). What we described as complementarity in the lexicalization of manner and result, then, is more accurately characterized as a complementarity in the lexicalization of scalar and non-scalar change.”
5

Table from Talmy (2007) (Table 2.12, pg 154).
Language/language family
The particular components of a
Motion event characteristically represented in the:
Verb Root
Motion + Path Romance
Semitic
Polynesian
Nez Perce
Caddo
Indo-European(not Romance) Motion + Cause
Motion + Manner
Chinese
Satellite
∅
Manner
(Figure/)Ground [Patient]
Path
Atsugewi Motion + Figure a. Path + Ground b. Cause
In our terms from above, the distinction here is between languages where verbs directly lexicalize the Type A meanings ( ‘path’) (Spanish), and where satellites lexicalize the Type A meanings (English, via particles and prepositions).
Important Open Questions :
• Is the difference between Motion and Change of State stated within the skeleton of Type A meanings, or is it simply a higher level set of cognitive categories within Type B flesh and blood?
• For the scalar vs. non-scalar distinction, is this encoded as complementary schemas within Type A, or are scalar changes a constructive ‘extension’ of non scalar change? Or is the distinction totally within the Type B domain?
2
2 See Son and Svenonius (2008) for a crosslinguistic study of directed motion which shows that the types of motion verbs need to be more finely distinguished than Talmy acknowledges; one of the dimensions in question can be interpreted as being what Levin and Rappaport Hovav called scalar vs. non scalar verbs of motional activity, in addition
6

• What kind of evidence can we use to bear on this kind of question?
(10) Butt’s Generalization (Butt 2003, Butt and Lahiri 2005):
Unlike auxiliaries which may become grammaticalized over time to have a purely functional use, light verbs always have a corresponding full or
‘heavy’ version in all the languages in which they are found.
If light verbs were functional elements and their heavy counterparts were roots, there would be no ready explanation of the fact that certain types of meaning (event structural) carry over from one use to the other.
A Natural Interpretation :
-I take the specificity of ‘light verb’ entailments to indicate that all verbs come with both skeletal (Type A) and fleshy (Type B) meanings.
- I take the abstractness of the light verb uses to indicate that Type B meanings are systematically defeasible in these languages.
In the next few subsections I compare light with heavy versions of Verbs in a number of languages to show (i) that they are indeed nontrivially related and (ii) to attempt to get at what it is that remains constant across both types of meaning.
Claim:
The ‘light’ verb use of a verb and the ‘heavy’ verb use are systematically related. Type B meaning is always negotiable’ Type A meaning never is.
Kearns (1988) distinguishes two types of complex predicates in English based on morphological shape and syntactic behaviour.
True Light Verbs (TLVs): where the N is a bare nominal that is form-identical with a verbal stem.
Vague Action Verbs (VAVs): where the N is a derived nominal (although that’s not always easy to see. True Light Verb constructions are characterized to result vs non result paths. Son and Svenonius show a fine-grained division of labour between which lexical item identifies which parts of the Path structure of the event that goes beyond Talmy’s simple typology.
7

by the fact that the bare nominal complement does not undergo the following processes characteristic of full DP objects: passivization, Wh-movement, relativization, reference by means of a pronominal, modification by adjectives and use of definite article. TLVs are systematically worse than VAVs with respect to the possibility of these syntactic operations. The nouns that are form identical with the verbal stem are also quite bad at licensing internal
(of-phrase and genitive marked) arguments on their own.
(11) (a) *John’s wipe of the table was efficient.
(b)*The table’s wipe was efficient.
TLVs occur with have , take , make , do and give in English. In addition, there are VAVs with make and give .
The following is a list (not exhaustive) from Cattell (1984).
Give : bark, belch, chuckle, cough, cry, gasp, growl, guffaw, gulp, hiccup, howl, moan, quiver, screech, shiver, shudder, sigh, snarl, sneeze, sniff, snort, sob, squeak, squeal, twitch, yawn, ...
Make : appeal, approach, attempt, boast, bolt, bow, dash, demand, endeavour, grab, grimace, guess, joke, journey, leap, lunge, move, protest, remark, request, rush, start, try, turn.
Have : bath, dance, drink, fight, glance, taste, use, poke, joke, kick, peep, quarrel, rest, ride, shave, shower, sleep, lie-down, think, throw, try, wash.
Do : dance, shuffle, somersault, sprint, wriggle.
Give : bash, biff, crack, cuddle, hit, hug, kick, kiss, pat, pinch, punch, smack, squeeze, thump, thwack.
3.1.1
‘Give + Body Contact Verb’
(12) (a) John gave Mary a book.
(b) Mary gave John a kiss.
Note that Impactgive can be detransitivized systematically using got
(13) (a) Mary got a kiss/a look (from Bill). (b) Fred got a kick in the pants from Bill.
8

What Stays :
-Initiated punctual transition with result; Initiator is the inalienable source of the path. Overt Recipient.
What Goes Away :
Physical transfer; resulting final physical possession.
3.1.2
‘Give + Emissive Verb’
(14) (a) The apple trees gave a lot of fruit this year.
(b) John/the train gave a shudder/sigh.
(My judgement is that give in its emissive use corresponds to give out which does not allow the double object structure.)
What Stays :
-Initiated punctual transition without result; Initiator is the inalienable source of the path. No Recipient
What Goes Away :
Physical coming into existence.
3.1.3
‘Have + Dynamic Verb’
‘Ordinary’ verb have has a variety of uses even when not combined with another predicational root.
(15) (a) John has a dog.
(b) John has a sister.
(c) John has a nice face/graceful movements.
(d) John has a bruise.
These are all stative, with some general semantics of possession but with different degrees of inalienability of the complement. In addition, there is a more dynamic use of have which goes with some complements.
(16) John had a heart attack/John is having a heart attack.
(from Cattell 1984, chapter 4)
(17) (a) John had a dreadful shock yesterday.
(b) John had a mango from the bowl before going out.
It is this more dynamic use of have that seems to be at play in the TLV complex predicates in English. In many cases, take is substitutable for have with very little difference in meaning (my judgements).
9

(18) (a) John had/took a swim.
(b) John had/took a lie-down.
(c) John had/took a read of Mary’s novel.
(d) John had/took a try of Mary’s cocktail.
Although, they are not always substitutable.
(19) (a) John had/*took a scribble in his notebook.
(b) John had/*took a play on the swings.
And some activity verbs do not work at all with either have or take .
(20) (a) *John had a work in the garden.
(b) *John had an eat of the apple.
(c) *John had a write of his dissertation.
Wierzbicka (1982) makes an attempt at describing what is semantically invariant about complex predicates with have in English.
(21) (from Wierzbicka (1982), pg 793)
X had a V =
For some time, not a long time
X was doing something (V) which could cause something good to happen in him that nobody else would know of he was doing it not because he wanted anything to happen to anything other than himself.
He could do it again.
Wierzbicka and Cattell point out that the have a V construction is systematically odd with (presumed) cognitively unaware entities.
(22) (a) John had a hit of the baseball.
(b)*The truck had a hit of the garage door.
(c) John had a kick of the expensive new football.
(d)??The horse had a kick of the ball.
It seems plausible that light verb have is the same as heavy verb have as used in the following type of sentence in (23), repeated from above.
(23) John had a mango from the bowl before going out.
What Stays :
-(Self)-Initiated experience, durativity dependent on boundedness of ‘nominal’ complement.
What Goes Away :
-Physical possession
10

3.1.4
‘Make + Dynamic Verb’
What about ‘make’ has a causational use but is non-emissive. Initiator is not the source of the undergoer path.
I am going to concentrate on the TLVs with make here, and ignore the obviously derived nominals. I include only those which are used with the indefinite article.
(24) (a) John made a dash for the border.
(b) John made an appeal to Mary.
(c) John made a grimace and turned away.
(d) John made a start on his dissertation.
(e) John made a move towards the door.
(f) John made a lunge towards his opponent.
(g) John made a quick turn on his heel and left the room.
All the latter sentences are paraphrasable with the eponymous verbs.
They also all conform to the argument structure of the verb make on its own in its heavy use without a benefactive argument.
(25) John made a cake.
These TLVs do not allow an indirect object, even though make does.
(26) (a) *John made Mary a dash for the border.
(b) *John made the border a dash.
(c) ??John made me a grimace, and turned away.
(d) *John made himself a move towards the door.
This makes TLV make similar in some interesting respects to Emissivegive . And in fact, some verbs can go with both make and give .
(27) (a) John made a grimace/gave a grimace.
(b) John made a lunge/gave a lunge....
(c) John made a groan/gave a groan....
More of the Emissivegive ones are good with make than the other way around.
Make seems more restrictive on the one hand though, in requiring an intentional causing actor in this usage. It is less restrictive in its choice of complement however. Any action seems to do for make , provided it can be done in a delimited amount of time. Emissivegive seems to require the sister verb-noun to have the initiator as its inalienable source.
11

What Stays :
-Initiated intentional causation
What Goes Away :
-small clause result and beneficiary.
3.1.5
‘Do + Activity Verb’
It’s surprising how few of these there are. They are a small subset of the unergative verbs of motion. Of the five that Cattell cites, I actually only find do a dance and do a somersault any good.
What Stays :
-Initiated non-scalar change
What Goes Away :
-nothing
Notice there is also systematic selection between the light verb and the nominal deverb that it goes with. I argue elsewhere that this is due to matching for certain categorial features (Ramchand 2008a). This is another source of evidence for the structural properties of LVs, but I put aside the technical details of how to implement matching here.
Light verb constructions in Bengali are monoclausal from the point of view of agreement, control and anaphora, and in addition exhibit integrity with respect to scrambling and adverbial modification (see Butt 1995 for a detailed examination of the equivalent construction in Hindi/Urdu). They always give rise to a telic interpretation. The light verb is marked for tense and agreement, and the main verb is explicitly marked as a perfective/conjunctive participle.
There are a number of different light verbs that are used in this construction in Bengali (between 8 and about 16 depending on how one counts), but
I will only consider the most common and productive here.
3
The three light verbs that I will focus on for this discussion are: jaoya -
‘go’, ot.ha
-‘rise’, phæla -‘drop/throw’. My data also includes the extremely common pOr.a
-‘fall’ and t.ola
-‘lift’, but for the purposes of the ‘matching’
3 My thanks to Tista Bagchi for her patience and good nature during extensive informant sessions, and for her careful descriptions of contexts of use.
12

that I found, ‘fall’ patterned like ‘go’ and ‘lift’ patterned like ‘drop’. In the interest of space and clarity, I present only the data from ‘go’, ‘rise’ and
‘drop’ here. All of these light verbs have both a ‘light’ verb use and a ‘heavy’ verb use, conforming to Butt’s Generalization .
jaoya -‘go’
(28) (a)ritu bar.i gælo
Ritu house go.
past3
‘Ritu went home.’
(b)gelaˇs-t.a bhem-e gælo glasscl breakperfpart gopast3
‘The glass broke.’ phæla -‘drop/throw’
(29) (a)ritu hat theke boi-t.a
phello
Ritu hand from bookcl drop.
past3
‘Ritu dropped the book from her hand.’
(b)ritu kaj-t.a
kor-e phello
Ritu workcl doperfpart drop.
past3
‘Ritu finished her work.’
Ot.ha
-‘rise/climb’
(30) (a)ritu gaˇche ut.hlo
Ritu treeloc climb.
past3
‘Ritu climbed the tree.’
(b)baloon-t.a
ut.hlo
balloonclass rise.
past3
‘The balloon rose.’
(c)ritu hOt.hat heˇs-e ut.hlo
Ritu suddenly laughperfpart rise.
past3
‘Ritu burst out laughing.’
Trans : init, proc, (res)
Unerg : init i
, proc, (res)
Unacc : proc, (res)
Considering the three light verbs under discussion here, I make the following assumptions about their make-up as full verbs:
13

jaoya -‘go’ is unaccusative ot.ha
-‘rise/climb’ is ambiguous between an unergative and an unaccusative reading phæla -‘drop’ is transitive .
3.2.1
Selectional Facts: ‘go’ and ‘fall’
Looking first at the light verb jaoya -‘go’, I found that it was exceptionlessly good with intransitive unaccusative verbs as in (31), exceptionlessly bad with unergatives like ‘speak’ and ‘dance’ (32), and also with transitives (33).
(31) dorja-t.a
khul-e gælo doorclass.
opene go.
past3
‘The door opened.’ unaccusative
(32) *O kothat.a
bole gælo he wordclass.
speake go.
past3
‘He spoke’ (intended) unergative
(33) *jon bar.it.a
ban-iye gælo
John houseclass builde go.
past3
‘John built the house’ (intended) transitive
The light verb pOr.a
-‘fall’ behaved the same way, but was only good with a (large) subset of the unaccusative verbs and seemed to require in addition, a sense of suddenness and/or unpredictability that was absent with ‘go’. I put aside the encyclopedic connotations of the light verbs here, and concentrate on the most productive exemplars of each group.
4
3.2.2
Selectional Facts: ‘throw’ and ‘lift’
(34) *dorja-t.a
khul-e phello doorclass.
opene throw/drop.
past3
‘The door opened.’ (intended)
(35) ram bol-e phello
Ram speake throw/drop.
past3
‘Ram blurted something out.’
(36) O bar.i-t.a
ban-iye phello he/she houseclass.
builde throw/drop.
past3
‘He accomplished/completed the building of the house.’ unaccusative unergative transitive
4 There is also a more ‘aspectual’ use of jaoya -‘go’ as a light verb which has quite distinct semantics along the lines of ‘went on V-ing’, and on this meaning it was able to combine with verbs of all types and valency. I put this reading aside here.
14

3.2.3
Selectional Facts: ‘rise’
(37) boroph jom-e ut.hlo
snow/ice freeze/accumulatee risepast.3rd
‘The ice/snow accumulated.’ unaccusative
(38) O bol-e ut.hlo
he/she speake rise.
past3
‘He/she spoke up.’ unergative
(39) *O bar.i-t.a
ban-iye ut.hlo
he/she houseclass.
builde rise.
past3 transitive
‘He accomplished/completed the building of the house.’ (intended)
(40) O bagh-t.a-ke lathi mer-e ut.hlo
he/she tigerclass acc kick hite rise.
past3
‘He unexpectedly kicked the tiger.’ transitive
To summarise the results of this section, a light verb in Bengali imposes a requirement on the argument structure of the main verb it combines with
(independent of any other additional semantic constraints is may impose).
Generalization :
• jaoya -‘go’ (and pOr.a
-‘fall’ ) can only combine with unaccusative main verbs.
• phæla -‘drop’ (and t.ola
-‘lift’) can combine only with transitive and unergative main verbs.
• ot.ha
-‘rise/climb’ can combine with either unaccusative or unergatives, and even one or two transitives.
(41) The Light Verb Constraint :
A verb can be used as a light verb when all of its category features
Agree with some other verbal element in its complement domain.
If two lexical items combine to lexicalize an event structure skeleton, then their conceptual contents must be able to unify without contradiction
15

What Stays :
-Association with particular subevental heads, inceptive vs. completive transition.
What Goes Away :
-physical motion, transfer and possession.
In Persian, light verbs come in transitive/intransitive pairs. In addition, the choice of light verb determines the choice between durativity and punctuality. (Karimi-Doostan 1997, Meegerdomian 2002, Folli et al. 2005, Pantcheva
2009).
(42) causative LVs inchoative LVs zædæn ‘hit/strike’ xordæn ‘collide’ kærdæn ‘make’ aværdæn ‘bring’ shodæn amædæn
‘become’
‘come’ dadæd ‘give’
ændæxtæn ‘throw’ didæn ‘see’ oftadæn ‘fall’
The transitive versions do not necessarily give rise to two overt arguments, rather the resulting complex predicate for the LVs in the left column seem to require volitional agents (data from Pantcheva 2009)
(43) (a)bæchche qælt=zæd child tumble=hit
‘The child tumbled over (intentionally)’
(b)bæchche qælt=xord child tumble=collided
‘The child tumbled over (by accident).’
Meegerdomian (2002) notices that verbs using ‘hit’ as a light verb have a punctual reading, whereas the same verb with the light verb ‘pull’ gives a durative reading.
(44) Punctual Durative dad zædæn dad keshidæn ‘to shout’ cry hit cry pull næfæs zædæn næfæs keshidæn ‘to breathe’ breath hit breath pull
16

Folli, Harley and Karimi (2005) is a good example of a constructivist treatment of complex predicates. They embrace the natural temptation to turn all light verbs into little v and argue that Persian complex predicates are just phrasal spell-outs of Hale and Keyser’s lexical syntactic structures.
(45) vP vP
DP
John v ′ DP
Kimea v ′
(46) v do vP
N work
N gerye -‘cry’ v kard -‘did’ vP v become
AP AP v shod -‘became’
DP the door
A open
DP
Kimea
A bidar -‘awake’
Folli, Harley and Karimi try to derive the properties of the resulting CPs from the nature of the complement of little v alone. However, this still leaves the problem of ‘selection’ and of the non-accidental relationship between the properties of the light verb in its heavy use, and the properties of the complex predicate that results.
What Stays :
-initiated vs. non-initiated event structures, punctual vs. durative.
What Goes Away :
-physical motion and physical transfer.
From the evidence of light verbs, it looks like causation, event structure and abstract path related meanings are skeletal Type A, while motion in space is Type B.
17

(47) DOMAINS OF CONFLATION I
Syn-Sem Lexical Encyclopedic Identifiers
Cause non change vs. change
-manners of causation
(instruments, degree of volitionality)
-specific properties and state descriptions non scalar vs. scalar change
-manners of change types of dynamicity generally
(qualities of motion, speed, attitude, shape and orien multivariate vs. bivariate transition types of scalar changes source of scale result of change
(properties, ordered locations, deictic information) specific properties, locations for start and end of scalar path
Lexical Encyclopedic Identifiers can be full verbs or small morphemes within verbs, and/or various satellites like particles. I assume that LEC content can in principle smear all over Syn-Sem structure, and that there is no enforced limit to what you might choose to do with a single memorized lexeme. Some are general, and some are particular; some smear over just one subevental head, some smear over several.
However, we still have a number of major questions to address before we can have a satisfying story for Butt’s Generalization and the pattern of loss that turns main verbs into light verbs.
• Why is it possible to remove Type B fleshy meanings from some verbs and not others? Making this a systematic possibility for natural language would grossly overgenerate.
• Why is it so easy to subtract physical motion in space and physical transfer from a light verb’s semantics?
• Indeed, even from the evidence of other phenomena, why is it so easy to conflate with motion?
Many verbs that get used as light verbs do have quite impoverished lexical semantic content. But it’s not quite zero. If it were zero, then we wouldn’t have to have rules for when Type B meanings were negotiable, we would just use the most abstract syn-sem identifiers as light verbs, whose ‘content’ would always unify unproblematically with whatever other subevental
18

information was around in in the form of non-finite verbal forms. But over and over again the verbs in question are verbs of generalized movement and transfer in space (attached to different path properties).
What if those verbs really did have zero lexical semantic content after all?
What if motion, location and transfer in space are cognitive defaults that don’t appear in these lexical items at all?
If that were true, then there would be a whole other dimension of meaning that consistently unifies with syn-sem structure.
(48) DOMAINS OF CONFLATION II
Syn-Sem Cognitive Defaults
Cause caused positional transfer
Lexicon non change vs. change non scalar vs. scalar change multivariate vs. bivariate transition source of scale result of change locations
-manners of change change of location
EVERY
THING
ELSE
Butt, Miriam. 1995.
The structure of complex predicates in Urdu . Dissertations in linguistics. CSLI Publications, Stanford, Calif.
Butt, Miriam. 2003. The morpheme that wouldn’t go away. Handout, University of Manchester seminar series.
Butt, Miriam and Aditi Lahiri. 2005. Historical stability vs. historical change.
Ms., University of Konstanz.
Cattell, Ray. 1984.
Composite Predicates in English , vol. vol. 17 of Syntax and Semantics . Academic Press, New York.
19

Fabregas, Antonio. 2007. The exhaustive lexicalization principle.
Nordlyd
Tromsø Working Papers in Linguistics 34 2.
Folli, Raffaella, Heidi Harley, and Simin Karimi. 2005. Determinants of event type in Persian complex predicates.
Lingua 115: 1365–1401.
Karimi-Doostan, Gholamhossein. 1997.
Light Verb Constructions in Persian .
Ph.D. thesis, Essex University.
Kearns, Kate. 1988. Light verbs in English. Ms, Massachusetts Institute of
Technology.
Meegerdomian, Karine. 2002.
Beyond Words and Phrases: A Unified Theory of Predicate Composition . Ph.D. thesis, University of Southern California.
Pantcheva, Marina. 2009. First phase syntax of persian complex predicates: argument structure and telicity.
Journal of South Asian Linguistics 2 1:
51–70.
Ramchand, Gillian. 2008a. Lexical items in complex predications: Selection as underassociation.
Nordlyd Tromsø Working Papers in Linguistics 35:
115–141.
Ramchand, Gillian. 2008b.
Verb Meaning and the Lexicon . Cambridge University Press.
Rappaport-Hovav, Malka and Beth Levin. 2008.
Reflections on manner/result complementarity.
In Syntax, Lexical Semantics and Event
Structure , edited by Malka Rappaport-Hovav Edit Doron and Ivy Sichel, chap. 2, pp. 21–37. Oxford University Press.
Son, Minjeong and Peter Svenonius. 2008. Microparameters of cross-linguistic variation: Directed motion and resultatives. In Proceedings of WCCFL 27 , edited by Natasha Abner and Jason Bishop, pp. 388–396. Cascadilla Press,
Somerville, Ma.
Talmy, Leonard. 2007. Lexical typologies. In Language Typology and Syntactic Description. Second Edition.
, edited by Timothy Shopen, vol. III, chap. 2, pp. 66–168. Cambridge University Press.
Wierzbicka, Anna. 1982. Why can you have a drink when you can’t *have an eat?
Language 58: 753–799.
20