U 4 ARGUMENT STRUCTURE

19
440A / 2002 / 42 MORPHOLOGY I LING-440A - FALL 2002 Prof. Jonathan David Bobaljik UNIT 4 ARGUMENT STRUCTURE Review: Fromkin, Victoria A. (ed.) Linguistics: An Introduction to Linguistic Theory, pages 94-116. Available at McLennan/Redpath Reserve Desk. Prelude: One of the things we’re looking for is a deeper understanding of a very productive kind of compounding in English, namely, the kind represented by truck-driver, etc. (in which the first noun corresponds to the object of the embedded verb). But it will take a little time to get there… 4.1 DERIVATION, VERBS AND ARGUMENT STRUCTURE (1) lockable, understandable, drinkable, movable … * dogable, *stoolable, *largeable, *blackable -able: attaches to verbs, (makes) Adjectives Adjective NLC Verb NLC lock able Verb Adj Adjective NLC Verb NLC drink able Verb Adj This door is lockable. This water is drinkable. Our conventions (WSTs, NLC) can capture the fact that –able makes verbs into adjectives. However, there are facts about this derivation which are not captured by our Wd-Structure Trees. (2) a. I locked the door. b. I drank the water. (3) * I am lockable. * I am drinkable. [cf. I am drinking.] The “subject” of the adjective corresponds to the object of the verb. Why?

Transcript of U 4 ARGUMENT STRUCTURE

440A / 2002 / 42

MORPHOLOGY ILING-440A - FALL 2002Prof. Jonathan David Bobaljik

UNIT 4 ARGUMENT STRUCTURE

Review: Fromkin, Victoria A. (ed.) Linguistics: An Introduction to Linguistic Theory,pages 94-116. Available at McLennan/Redpath Reserve Desk.

Prelude: One of the things we’re looking for is a deeper understanding of a very productivekind of compounding in English, namely, the kind represented by truck-driver,etc. (in which the first noun corresponds to the object of the embedded verb). Butit will take a little time to get there…

4.1 DERIVATION, VERBS AND ARGUMENT STRUCTURE

(1) lockable, understandable, drinkable, movable …* dogable, *stoolable, *largeable, *blackable

-able: attaches to verbs,(makes) Adjectives

AdjectiveNLC

VerbNLC

lock ableVerb Adj

AdjectiveNLC

VerbNLC

drink ableVerb Adj

This door is lockable. This water is drinkable.

Our conventions (WSTs, NLC) can capture the fact that –able makes verbs into adjectives.However, there are facts about this derivation which are not captured by our Wd-Structure Trees.

(2) a. I locked the door.b. I drank the water.

(3) * I am lockable.* I am drinkable. [cf. I am drinking.]

• The “subject” of the adjective corresponds to the object of the verb. Why?

440A / 2002 / 43

• Not all affixes behave alike:

(4) employ-er employ-ee

NounNLC

VerbNLC

employ erVerb Noun

NounNLC

VerbNLC

employ eeVerb Noun

(5) a. I employ you. í I am the employ-er.You are the employ-ee

b. I pay my landlord. í I am the pay-er.My landlord is the pay-ee.

(6) The Word-Structure Trees for –er and –ee are the same. Yet, there is a systematicdifference in the way the referent of the noun relates to the verb. Therefore, we aremissing something. And this something will be important in understanding the structureof the truck-driver compounds…

4.2 ARGUMENT STRUCTURE I

Along with their category feature (V), verbs are also associated with argument structures. This isa list of phrases which the word must (or may) appear with in a syntactic structure.

(7) a. Intransitive Verbs swim <Agent>

swim: *The fish swam me.The fish swam.

b. Transitive Verbs bite <Agent <Theme>>

bite: The dog bit me.*The dog bit.

c. Optionally transitive verbs. sing <Agent <(Theme) >>

sing: The soloist sang the aria.The soloist sang.

d. Ditransitive Verbs give <Agent <Theme, Goal>>

give: You gave a book to Jody.*You gave a book.*You gave Jody.*You gave.

The labels we give to the arguments are called “Thematic Roles” or “q-Roles” (see O’Grady &Dobrovolsky [Intro Textbook] pp250 ff and/or Fromkin ed. reading on reserve).

440A / 2002 / 44

These tell you what semantic function the subject, object etc… of a verb have in relation to thatverb. Here is a list of the ones we will make the most use of in class.

(8) Agent Ag The thing that is responsible for doing the action [also Actor]Theme Th The thing that is moved, changed, described [also Patient]

(The external argument of a simple Adjective is usually a Theme)Goal Go The thing where an action ends (or for which the action was done).Location Loc The place where the theme is.Referent R The thing a noun refers to. (we’ll see this below)Event Ev Something that takes place, that happens. (we’ll see this below)

(9) Typically, one of these arguments is special in that it appears outside the phrase (maximalprojection) that contains the verb plus all its other arguments. This is called the subject orexternal argument. (In English, this is usually the Agent if there is one). The externalargument is the subject of a simple clause:

(10) a. The fish swam.b. The dog bit me.c. The soloist sang (a song).d. You gave a book to Jody.

• If you look at the examples in (7), you see that the Agent is the external argument, and that itis contained in the outermost layer of angled brackets (by convention), thus:

(11) Argument Structure: < External Argument < (Internal Arguments) >>

(12) The external argument can also usually be identified by its place in a construction like:

They consider external argument [PREDICATE (to be) X ].or They made external argument [PREDICATE X ].

(13) a. They consider the dog [to be biting me].b. They consider me [to be biting the dog]. “me” ≠ Thc. They made the dog [bite me].d. They made me [bite the dog]. “me” ≠ Th

• Return to the -able examples above:

(14) They consider this door [ (to be) lockable].*They consider me [ (to be) lockable]. (not grammatical as ‘me’ = theme)

(15) I am glunkable. í It is possible [(for someone) to glunk me].* It is possible [for me to glunk (someone)].

(16) The external argument of an adjective made with –able corresponds to the object of theverb.

440A / 2002 / 45

(17) They consider me [(to be) the employer].*They consider you [(to be) the employer]. [i.e. ≠ (5)]

(18) *They consider me [(to be) the employee]. [i.e. ≠ (5)]They consider you [(to be) the employee].

(19) The external argument of a noun with –er is the subject of the verb.The external argument of a noun with –ee is the object of the verb.

• We need to incorporate these generalizations into our theory. But first…

4.3 ARGUMENT STRUCTURE II: NOUNS

• Do nouns really have external arguments?

They consider external argument [PREDICATE (to be) X ].or They made external argument [PREDICATE X ].

(20) a. They consider me [(to be) a fool].b. They consider this [(to be) a good book].c. They made me [prime minister].d. They made this [a good show].

(21) a. The metal is flat.b. The dog is biting me.c. The metal is steel.d. Bill is a fool.e. This is a good book.

(22) In a sense, nouns have external arguments / subjects, too. It’s somewhat hard to put one’sfinger on it, but the closest we can come is that the external argument of a noun is whatthe noun “refers” to (this term is borrowed from philosophy). Linguists therefore call thethematic role associated with this external argument the “R” role (for “referent”).

(23) Referent R The thing that the noun refers to.

• Some nouns have more complex argument structures (just as verbs do).

(24) a. picture <R, <(Theme)>>They consider this [(to be) a nice picture of Montréal]

b. president <R, <(Theme)>>They made Bill [president of the country.]

440A / 2002 / 46

4.4 ARGUMENT STRUCTURE AND AFFIXATION – NLC

(25) I consider the sky [(to be) dark]. dark, Adj <Th>I consider that [(to be) darkness]. darkness, N <R>

*I consider the sky [(to be) darkness]. *darkness, N <Th>

Noun <R>NLC

Adj, <Th>NLC

dark nessAdj, <Th> N, <R>

(Caution: we may revise this tree later)

(26) I consider that [(to be) rage]. rage, N <R>*I consider my mother [(to be) rage]. *rage, N <Th>I enraged *(my mother). enrage, V <Ag, Th>

I enraged my mother. í ≠ My mother is rage.≠ I am rage.

V, <Ag,Th>NLC

Noun <R>NLC

en rageV, <Ag,Th> N, <R>

4.5 ARGUMENT STRUCTURE AND AFFIXATION – PC

For now, it looks like we’re on the right track in treating Argument Structure as a feature.Argument structure information seems to be subject at least to the NLC. Is it subject to the PC?

(27) write <Ag,Th> rewrite <Ag,Th> I rewrote the handout twice.inspect <Ag,Th> reinspect <Ag,Th> They reinspected my car.

enter <Ag> reenter <Ag> They entered, left, and re-entered.marry <Ag> remarry <Ag> The couple remarried.awaken <Th?> reawaken <Th?> I reawakened shortly before ten.

440A / 2002 / 47

Verb <Ag>PC

Verb <Ag>NLC

re- enterØ V, <Ag>

Verb <Ag,Th>PC

Verb <Ag,Th>NLC

re- inspectØ V, <Ag,Th>

• If the prefix re- had its own Argument Structure, what would that predict for verbs that areprefixed with re-?

4.6 ARGUMENT STRUCTURE AND COMPOUNDING – HR

• As we should have expected by now, in simple compounds, the argument structure of theright hand member (in languages like English) determines the features of the whole thing:

(28) N+V = V with all the argument structure properties of the V

fry <Ag <Th>> I fried the steakpan-fry <Ag <Th>> I pan-fried the steak.

wash <Ag <Th>> I washed the sweater.hand-wash <Ag <Th>> I hand-washed the sweater.

feed <Ag <Th, Go>> I fed the corn to the baby.spoon-feed <Ag <Th, Go>> I spoon-fed the corn to the baby.

nap <Ag> I napped (all afternoon).cat-nap <Ag> I cat-napped (all afternoon).

walk <Ag> I walk on Thursdayspower-walk <Ag> I power-walk on Thursdays.

In each case, the compound takes the same argument structure (appears with the same kinds ofphrases in syntax) as its righthand member. This is not limited to verbs:

(29) V+N = N with the argument structure properties of the N

bridge <R <Loc/Go>> That is the bridge over the Lachine Canal..drawbridge <R <Loc/Go>> That is the drawbridge over the Lachine Canal.

cloth <R> I consider that [(to be) a nice cloth].washcloth <R> I consider that [(to be) a nice washcloth].

[wash <Ag <Th>>] * A washcloth by John of dishes.≠ A cloth for John’s washing of dishes.

440A / 2002 / 48

And, even when the two phrases take the same argument structure (as in many N+Ncompounds), it is clearly the argument structure of the right-hand member that is implied.

(30) N+N = N where the <R> is the referent of the right-hand member

bull <R> <R> refers to bulls. That is a bull. =True iff “That” = “a bull”frog <R> <R> refers to frogs. That is a frog. =True iff “that” = “a frog”bullfrog <R> <R> refers to a type of frog, not a type of bull.

Indeed, we can say more generally that a sentence with a compound typically implies thecorresponding sentence with the right-hand member but NOT the corresponding sentence withthe left-hand member (in English):

(31) This is a bullfrog. IMPLIES This is a frog. NOT This is a bull.This is a drawbridge. IMPLIES This is a bridge. NOT This draws (bridges).I pan-fried the steak. IMPLIES I fried the steak. NOT. I am a pan of steak.etc…

These observations are all straightforwardly captured by the percolation of argument structurevia the (Right Hand) Head Rule.

(32)Noun <R>

HR

Verb Noun<Ag<Th>> <R>

NLC NLCwash cloth

V <Ag<Th>> Noun <R>

Verb <Ag<Th>>HR

Noun Verb<R> <Ag<Th>>

NLC NLCpan fry

Noun<R> V <Ag<Th>>

Noun <R2>HR

Noun Noun<R1> <R2>

NLC NLCbull frog

Noun,<R> Noun <R>

440A / 2002 / 49

4.7 BEYOND PERCOLATION

• We still haven’t accounted for the properties of affixation we started with. That is, sometimesthe argument structure of the whole word is not simply the features of one of its lexicalelements; sometimes it is a function of both.

• More examples: Derived Verbs.

(33) flatADJ í flattenVERB, softADJ í softenVERB, whiteADJ í whitenVERB …

Verb <Ag,Th>NLC

AdjNLC

flat enAdj Verb

<Th> <??> There is a relationship between the theme of the whole andthe theme of the adjective that is not expressed in our Tree.

(34) I flattened the metal. í The metal is flat.≠ I am flat.

(35) They consider the metal [(to be) flat].*They consider me [(to be) flat].

4.8 LINKING THEORY

(36) Japanese transcription: broadly phonemic

a. Kaiga iku. ‘Kai is leaving.’b. Taroga hataraku. ‘Taro is working.’c. Taroga hono kawu. ‘Taro is buying a book.’d. Kaiga sushio taberu. ‘Kai is eating sushi.’e. Kaiga sushio tabeta. ‘Kai ate sushi.’f. Kaiga Taroni hono ageru. ‘Kai is giving Taro a book.’

440A / 2002 / 50

(37) Nouns: Verbs: “Inflectional Suffixes:”

Kai Kai ik- ‘go’ -ga sUBJECT CASETaro Taro hatarak- ‘work’ -o OBJECT CASEhon book kaw- ‘buy’ -ni INDIRECT OBJECT CASEsushi sushi tabe- ‘eat’ -ta PAST

age- ‘give’ -(r)u PRESENT-ru / V___-u / C ___

(38) CRASH COURSE IN JAPANESE SYNTAX:

Basic Word Order: Subject (Indirect Obj) (Object) VerbCase-Marking: SUBJECT-ga (IND.OBJ-ni) (OBJECT-o)(FYI: traditional names:) Nominative Dative Accusative

(39) a. Hanako-ga Kai-o ik-ase-ru. ‘H. makes K. leave.’b. Hanako-ga Taro-o hatarak-ase-ru. ‘H. makes K. work.’

c. Hanako-ga Taro-ni hon-o kaw-ase-ru. ‘H. makes T. buy a book.’d. Hanako-ga Kai-ni sushi-o tabe-sase-ru. ‘H. makes K. eat sushi.’

e. Hanako-ga Kai-ni Taro-ni hon-o age-sase-ru. ‘H. makes K. give T. a book.’

(40)

-(s)ase --sase / V___-ase / C___

Ï Ì Ó

¸ ˝ ˛

"make s.o. Verb"[V __]Verb< ?? >

È

Î

Í Í Í Í Í Í Í Í Í

˘

˚

˙ ˙ ˙ ˙ ˙ ˙ ˙ ˙ ˙

CAUSATIVE

OBSERVATIONS:• Adding –(s)ase- to a 1-argument verb (intransitive) makes a 2-argument verb (transitive)• Adding –(s)ase- to a 2-argument verb (transitive) makes a 3-argument verb (ditransitive)• Adding –(s)ase- to a 3-argument verb (ditransitive) makes a 4-argument verb ( ?? )

Looking at the meanings of the sentences, what’s going on isn’t very hard to identify, nor, forthat matter, is it very different than what happens with English “make … verb”, except that it allends up as one verb-word in Japanese, while it’s two verbs in English, French, etc.

• The CAUSATIVE adds exactly one argument to the clause, and it is an Agent.

440A / 2002 / 51

(41) So why can’t we just say that –(s)ase is a V <Ag>?

‚ Verb <Ag>NLC

Verb <Ag>NLC

hatarak- -(s)aseVerb, <Ag> V, <Ag>

‚ Verb <Ag>NLC

Vb <Ag,<Th>>NLC

tabe- -(s)aseVb, <Ag,<Th>> V, <Ag>

• NLC applies, leading us to expect that the resulting verb will have an Agent, whichwill be the external argument. Looking at the examples in (39) we’re partially right:the verb tabe-sase- does take an Agent as its external argument, and moreover, theAgent is not the Agent associated with the inner verb tabe-. But it’s not quite right.The NLC leads us to believe that the argument structure of the result will be only theargument structure of the affix, which isn’t what happens.

(42) Try at home: Why won’t <Ø> or <(Ag)> work for –(s)ase ? (PC)Why won’t <(Ag), <(Th), (Go)>> [or anything else] work?

(43) make <Ag <…>>

Fido barked í I made [Fido bark].Fido bit the cat. í I made [Fido bite the cat].

The external argument of make is an Agent, the one who causes. We can in fact see thisby putting all of this into one of our external argument test frames:

(44) a. They made [me] [make Fido bite the cat.]b. They consider [me] [to be making Fido bite the cat.]

(45) make <Ag <Event>> where <Event> means whole embedded clause

• The internal argument of make is in some sense a whole embedded clause,complete with all its pieces (including its own internal argument), i.e., an Event,which, (like a sentence) is made up of an external argument + a predicate. It is in factthis property of make which allows us to use it as a test for external argument-hood.

(46) -(s)ase more or less means the same as make and is serving some similar function, but itdoesn’t attach to (embed) whole phrases. Rather, it takes the same phrases as the verb itattaches to plus the causer-Agent external argument.

What is happening here is that the argument structure of the non-head is replacing (orsubstituting for) the event argument of the affix. In this way, it can become a part of theargument structure of the whole word. The argument structures of the two morphemes aresaid to be linked.

440A / 2002 / 52

(47) SUBSTITUTION-LINKING

Verb <Ag, <Ag<Th>>>NLC

Verb <Ag,<Th>> -(s)ase(V), <Ag, <Ev>>

tabe-Verb, <Ag,<Th>> substitution

Verb <Ag, <Ag>>NLC

Verb <Ag> -(s)ase(V), <Ag, <Ev >>

hatak-Verb, <Ag > subst.

(48) In order to be able to have morphemes that add arguments we need the argumentstructure of the embedded verb to become the internal argument structure of the wholeword, but without losing the generalizations we have captured above with the NLC andthe PC.

Hint: Do not allow yourself to be intimidated by the Word Structure Trees. They compress alarge amount of information into a small diagram, but that is what diagrams are for.Think carefully and mechanically about what is going on here. If the argument structureof the embedded verb replaces the Event argument of –(s)ase, we explain two things:

• The whole word has the argument structure of the embedded verb plus a new externalargument.

• The whole word in Japanese does not take an Event argument phrase in the syntax,even though one is there in its meaning. This is because it has been replaced.

(49) To a very limited degree, this is recursive(this is not poductive, and c. sounds a little forced to most Japanese, but is productive in certain cases):

a. Inu-ga neko-o kam-ta [kam+ta = kanda]dog-NOM cat-ACC bite-PAST‘The dog bit the cat.’ kam- <Ag, <Th>>

b. Taro-ga inu-ni neko-o kam-ase-ta.Taro-NOMdog-DAT cat-ACC bite-(s)ase-PAST‘Taro made the dog bite the cat.’ kam-ase- <Ag, <Ag<Th>>>

c. Hanako-ga Taro-ni inu-ni neko-o kam-as-ase-ta.Hanako-NOM Taro-DAT dog-DAT cat-ACC bite-(s)as(e)-(s)ase-PAST‘Hanako made Taro make the dog bite the cat.’

kam-as-ase- <Ag,<Ag,<Ag<Th>>>>

Practice: Can you draw the Word-Structure Tree for (c.)?

440A / 2002 / 53

4.8.1 Substitution-Linking in English

The suffix –ing makes Nouns out of Verbs. Consider the effect that it has on the argumentstructure of the verb:

(50) a. John washed the dishes.i. John’s washing of the dishes (took all day).ii. The washing of the dishes by John (took all day).iii. I will consider THAT [John’s (fastest) washing of the dishes.]

b. The dog chews books.i. The dog’s chewing of the books (upsets me).ii. The chewing of books by the dog.iii. I consider this the worst chewing of books by the dog that I will tolerate.

c. Fido barks.i. Fido’s barking annoys the neighbours.ii. The (constant) barking by Fido drives me crazy.iii. I consider that noise [to be the loudest barking by Fido] (I’ll tolerate).iv. [That noise] was Fido’s barking.

• The external argument of the verb is an internal argument of the noun and is thus expressedinside the noun phrase (either as a possessive, or with the preposition by). The noun alsotakes a new external argument, which is an <R> (as always, with nouns). [I.e., “Fido’sbarking” refers to the noise in (iv)].

• These facts can be explained just like the Japanese facts. English –ing has a normal argumentstructure for Nouns, (i.e., an external <R> argument, and takes an <Event> internal argumentwhich gets replaced (via substitution-linking) by the argument structure of the verb.

It is a property of English nouns that all of their arguments are typicallyoptional. The argument structure reflects the ones that can be expressed (asopposed to those that need be).

(51)

-ing"the action of Ving"

attaches to verbsNoun

< R < Ev >>

È

Î

Í Í Í Í

˘

˚

˙ ˙ ˙ ˙

440A / 2002 / 54

Noun <R, <Ag, Th>>NLC

Verb <Ag,<Th>> -ingN <R <Ev>>

washVerb, <Ag,<Th>> substitution

(52) I consider this John’s fastest washing of the dishes.R Agent Theme

4.8.2 Compounds with –er and –ing Nouns

(53) Some transitive verbs <Ag <Th>>:

These lions eat meat.This machine washes dishes.Our Prime Minister used to drive trucks.

(54) Consider the corresponding –er and –ing nouns (nominalizations). We can see that thatsubject (Agent) and object (Theme) of the verb become arguments of–ing and –er nouns.

• We saw in the last handout how this works (via substitution-linking) for –ingnominals. As these examples show, -ing nouns inherit the argument structure of theverb they attach to and, effectively, add an external argument, which is an <R>.

• Things work slightly differently for the –er nouns. The object of the verb <Th> isan object of the –er noun but the external argument of the verb <Ag> is also theexternal argument of the noun <R>. We’ll see how to get this shortly. For now, wewill just note this fact by writing <R=Ag>.

a. I consider our biggest problem [(to be) the eating of meat by lions].I consider our biggest advantage [(to be) the washing of dishes by machines].I consider a major hazard [(to be) the driving of trucks by politicians].

<R> <Th> <Ag>

b. I consider these lions [(to be) eaters of meat].I consider this machine [(to be) a good washer of dishes].I consider J.C. [(to be) a driver of trucks].

<R=Ag> <Th>

• The important thing to note about these examples is that the phrases introduced by of arefulfilling the same role (thematically/semantically) as they do in (53). Now consider thesecompounds:

440A / 2002 / 55

(55) a. This lion is a meat-eater.This machine is a dish-washer.The P.M. is a truck-driver.<R = Ag> <Th>

b. One problem is meat-eating (by lions).Another problem is truck-driving (by politicians).<R> <Th> <Ag>

OBSERVATION: These are N-N compounds, but the first Noun is clearly being interpretedas the <Theme> of the second noun. I.e., it is fulfilling the same role as the of-phrases in(54) and the objects of the verbs in (53).

(56) ARGUMENT IDENTIFICATION (=IDENTIFICATION-LINKING):Write a superscript letter (e.g. i ) to the right of the label of the node which is identifiedwith an argument of its sister, and a matching superscript to the right of the ThematicRole it satisfies. (Like SUBSTITUTION-LINKING in (47)), this is a process that applies tosister nodes, i.e., nodes which share a unique dominating node.)

(57) A simplified tree, many details omitted.Note the position and identity of sub- and super-scripts.

Noun …

Noun <R> Noun<R <Ag, Th >>

truck drivingNoun, <R> [internally complex]

ii

What happens to the argument structure of the whole compound? Consider the following data:

(58) an eater of fish a meat-eater * a meat-eater of fishthe washing of dishes a dish washer * the dish-washing of platesa driver of semis a truck-driver * a truck-driver of semis

440A / 2002 / 56

OBSERVATION: The argument of –er or –ing nouns that is identification-linked is “usedup” by this identification-linking. We call this “saturation” and have:

(59) The SATURATION PRINCIPLE

If an argument of the head is identified with (= linked to) the non-head, then thatargument is not passed on to the larger word.

(If you’ve taken Syntax before, you will recognize the Theta-Criterion at work here.)

• How do we know the HR is still applying? The rest of the argument structure is passed up:

(60) a. Meat-eating [<Ag> by vegetarians] has to be stopped.b. Truck-driving [<Ag> by politicians] should be encouraged.c. Gift-giving [<Ag> by the rich] [<Go> to the needy] is commendable.

(61)

Noun <R <Ag>>

HRNoun <R> Noun

<R <Ag, Th >>

truck drivingNoun, <R> [internally complex]

ii

440A / 2002 / 57

(62) A Word-Structure Tree for truck-driving, fully labelled for category and argumentstructure features.

Noun <R <Ag>>HR (+Sat. Princ.)

Noun <R> Noun<R <Ag, Th >>

NLC NLC

Verb<Ag <Th>>

substitutionNLC

truck driv(e) -ingNoun <R> Verb <Ag <Th>> N <R <Ev>>

ii

Extra: Read this box only if it helps you to understand things better:Why do <Ev> (Event) arguments disappear in substitution-linking?One could think of the verb as identification-linking to the Event argument. (This makesthe parallel between Japanese causatives and English much closer.) We still needsubstitution-linking (to get the argument structure of the verb to pass up the tree), but wecould now see things such that the Event argument’s failure to appear on the higher nodeas a result of the Saturation Principle.

440A / 2002 / 58

4.9 –ER NOUNS AND ARGUMENT BINDING

• We began the argument structure section (§4.1) with the observation that the referent of –ernouns corresponds to the Agent of the verb that –er attaches to. In the previous section, wewrote <R=Ag> under the nouns (“arguments”) in question [e.g., in (54)]. Theintuition/observation is that the R argument of a complex word [verb + er] corresponds to theAgent of the verb. How can we implement this?

• We saw in the previous section that the non-head of a compound can be identified with (“cansatisfy”) a Theta-Role of the head of the compound. We captured this formally byARGUMENT-IDENTIFICATION which we indicated by writing a corresponding superscripts onthe argument and the Theta-Role it satisfied. The Saturation Principle blocked the theta-rolewhich had been thus satisfied from being passed up to the next node by H.R.

(63) ARGUMENT BINDING (different from wording in text)

A nominalizing morpheme may be lexically specified to bind one of the internalarguments it inherits via substitution.

We indicate this with the same notation as Argument Identification (i.e., co-superscripting), but binding happens within the argument structure of a single node.

We use the same notation since the processes are very similar. In fact, it is theoretically possibleto reduce the two to a single process, but we will not pursue that in this course. Feel free to try towork it out for your own edification.

Note that Argument Binding, like Argument Identification, triggers the Saturation Principle.

Additional Question:

Why does only the internal argument get blocked from percolating by the Sat. Princ?

(64) EXTERNAL ARGUMENT CONDITION

The non-head of a word cannot be linked to the external argument of the head.

440A / 2002 / 59

(65) Consider applying the same ideas to –er nouns:

Noun<R < Th >>

NLC

Verb<Ag <Th>>

substitutionNLC

driv(e) -(e)rVerb <Ag <Th>> N <R <Ev>> =

N <R <Ag Th>>ii

ii• Step 1: Substitution-Linking applies, replacing the <Ev> argument of –er with the

argument structure of drive.• Step 2: Identification-Linking applies, linking the R argument of –er to the <Ag>

argument.• Step 3: The NLC applies, but the Saturation Principle prevents the (now identified)

<Ag> argument from passing up the tree. The other arguments (<R <Th>>) dopass up the tree.

We see that this derives the correct argument structure for the resulting noun driver:

(66) I consider J.C. [(to be) a driver of trucks].

440A / 2002 / 60

4.9.1 –er Nouns in Compounds: Putting it all together.

• We know that –er nouns can occur in compounds, and that the non-head may satisfy the Thargument of the head. We have all the tools (now) to correctly analyse:

truck-driver <R> meaning: one who drives trucksa truck-driver (*by politicians) (*of semi-s)

Noun < R >HR (+Sat. Princ.)

Noun <R> Noun<R < Th >>

NLC NLC

Verb<Ag <Th>>

substitutionNLC

truck driv(e) -erNoun <R> Verb <Ag <Th>> N <R <Ev>>

N <R <Ag Th>>

k

i i

k