|
|
|
|
| LOGICAL FORM AND RECONSTRUCTION |
The use of movement rules in the transition from d-structures to s-structures has been already discussed. What has not been fully explicated, however, are the logical forms (LF) and the concept of reconstruction. The best approach to these concepts is through the study of quantifiers. Consider the following sentences and their logical forms:
John saw Mary
John saw everyone
John saw every player
John saw someone
Everyone saw someone
A logical interpretation of the first sentence (John saw Mary) can be readily ascertained. The NPs (John and Mary) each pick out a referent from the universe of discourse. Next, the predicate (saw) establishes a relationship between these two entities.
Saw (John, Mary)
In logical form, these would be written as s (JM) where s is the predicate and J and M are the arguments. Now consider the next sentence (John saw everyone) which contains a quantifier (everyone). It would appear that this sentence could be readily interpreted as:
Saw (John, everyone)
The problem with interpretation, however, is that it is inadequate. Quantifiers differ from Reference expressions (R-expressions). They do not pick out a unique entity from the universe of discourse. A more realistic interpretation of this sentence would be:
Saw (John, x)
What John saw was not a unique person, but a variable. How that variable is to be interpreted depends on the domain of the quantifier. For this reason, it can be argued that the Referential NP (John) differs from the quantifier (x) in this sentence. R-expressions are constants; quantifiers are variables. This difference between constants and variables within the universe of discourse is not adequately captured by it s-structure representation or even by its restatement as a two place predicate: S (j, x). For this reason, logicians use special notations to represent these facts. S-structures must be reconstructed in this form:
x (Hx Sjx)
What this notation states is that for all x, it is the case that if x is human then John saw x. The argument x depends on the quantifier for its interpretation. This logical paraphrase of saw (John, x) contains a condition and this is represented by the arrow. The first part of the condition is to the left of the arrow and it restricts the range of x to humans. The second part of the condition depends on the quantifier for its interpretation. It is bound by the quantifier. Another way of stating this logical representation is:
x, x = H (Sjx)
For all x, such that x is human, John saw x
The letter x represents a variable. It is bound by a universal quantifier ( ) which appears in English as the lexical items "all" and "every." Quantifiers are also referred to as operators. An operator must bind each variable. Each operator must bind a variable. The interpretation of the variable depends on the operator that binds it. Now consider the sentence, John saw every player.
Two Place Predicate:: Saw (John, every player)
Logical Form: x, x = P (Sjx)
Interpretation: For all x, x is human and x is a player, John saw x.
In addition to the universal quantifier ( ), there are also existential quantifiers ( ). Consider the sentence: John saw someone. This meaning of this sentence cannot be captured by representing it as a two place argument: Saw (John, someone). Hence, it is represented in its logical form:
x, (x = H) & (Sjx)
There is an x such that x is human and John saw x.
What the existential quantifier ( ) means is that there exists at least one referent for the variable x. In its logical form, this existential quantifier binds its variable. The interpretation of this variable depends on the quantifier. In this case, an existential quantifier binds the variable (x). Now consider the following sentence in which contains both existential and universal quantifiers: Everyone saw someone. This sentence has two logical interpretations.
Ambiguous Sentence: Everyone saw someone
For every x there is some y such that it is the case that x saw y.
(Each person saw someone different)There is some y, such that for every x, it is the case that x saw y.
(There is one person who was seen by everyone)
How the ambiguous sentence is to be interpreted depends on the scope of the logical operators. The universal quantifier has a wide scope. The existential quantifier has a narrow scope. In the sentence in which every person saw someone different, the universal quantifier appears to the left of the existential quantifier in its logical form. In the sentence in which the same person was seen by everyone, the existential quantifier is not within the scope of the universal quantifier.
| QUANTIFIER MOVEMENT AND LOGICAL FORM |
PF Form: John saw everyone. (quantifier is in the object position)
D-Structure: [John -ed see everyone].
LF: [everyonei [John saw xj] ] (the quantifier is in a scope position)
The quantifier (everyone) must be moved to a left-peripheral position. This is a non-argument position and it is adjoined to the left of the IP. When the movement rule moves this quantifier to an initial adjunct position, it leaves behind a trace (x) which is coindexed (xj ). This trace is an empty category and it is represented as a variable xj. This variable is bound by the quantifier )everyone). The quantifier binds the variable x. This is because the variable is in the scope of the quantifier in its logical form. The movement rule that repositions the quantifier in its logical form is known as Quantifier Raising (QR).
The quantifier has a domain that it c-commands. By means of the rule of quantifier raising, it was moved to a scope position in the logical form of the sentence. The variable (xj) is trace-bound by this quantifier. It forms a syntactic chain: <everyone, xj >. This movement of QR differs from other movement rules in that it moves a quantifier into a non-argument position. One will recall that the rule of NP movement moved an NP from one argument position to another.
| WH-PHRASES AS OPERATORS |
John saw Mary
Who did John see?
Which player did John see?
The wh-constituents in these sentences do not have specific referents. They are not R-expressions. Hence, they must be restated in a LF that will provide for an adequate semantic interpretation.
Who did John see?
For which x, x is human, is it the case that John saw x?Which player did John see?
For which x, x is a player, is it the case that John saw x?
What is interesting about wh-constituents in English is that they already occupy a left-peripheral position (a scope position) in the s-structures. It is in a position to bind its trace, a sentence-internal empty category. It is similar in structure to its LF.
S-structure: [ whoi did [ John see ti] ] ?
LF: [ whoi did [ John see xi] ] ?S-structure: [ which player did [John see ti] ] ?
LF: [ whichi player did [ John see xi] ] ?
It is because of the repositioning of the wh-constituents in these sentences that they have acquired sentential scope. This wh-constituent is moved into [Spec, CP] and c-commands the clause which is its domain. Wh-movement, however, is not universal. In Japanese, for example, it does not occupy a scope position. It is now time to look at a language in which no wh-movement is necessary from the d-structure to s-structure level, but is necessary in GB Theory from the s-structure to LF level. This QR movement is necessary from the perspective of LF Theory. Quantifiers must be repositioned in Japanese so as to interpret their sentential scope.
| WH-RAISING IN JAPANESE |
Japanese is a verb final language. In this regard it is similar to Latin, Eskimo, and other languages with an SVO structure. Japanese differs from Latin, however, in that it has topic markers ( TM ), subject markers ( SM )and object markers ( OM ) rather than case endings.
Watakushi wa shimbun wo yonde imasu
[I TM newspaper OM read-ing be-polite ]
As for me, I am reading a newspaper.Sumisu-san wa hon wo yomimashita.
[Smith Mr. TM book OM read-polite-past tense]
Mr. Smith read the book.Atama ga itaimasu.
[head SM hurt-polite]
I have a headache.Anokata wa Tokyo kara kimashita.
[He TM Tokyo from come-polite-past]
He came from Tokyo.Kore wa hon desu.
[This TM book is]
This is a book.
When one asks questions in Japanese, no movement of components are necessary. A question marker (QM) or a question work is inserted into the sentence and the normal order is retained.
Atama ga itaimasu ka.
[head SM hurt-polite QM]
Do I have a headache?
Anokata wa Tokyo kara kimashita ka.
[He TM Tokyo from come-polite-past QM]
Does he came from Tokyo?Sumisu-san wa hon wo yomimashita ka.
[Smith Mr. TM book OM read-polite-past tense QM]
Did Mr. Smith read the book?Tokyo wa doko desu ka.
Tokyo TM where is QM]
As for Tokyo, where is it?(Anata wa) ikaga desu ka.
[(you TM how is QM]
(As for you,) How are you?Kore wa nan desu ka.
[this TM what is QM]
As for this, what is it?
Another fact about the Japanese language that one should know before discussing wh-raising is structure of relative clauses. Since Japanese is a VSO language, it follows the expected pattern of having relative clauses before NPs. This is the mirror image of VSO languages in which relative clauses follow NPs.
English: The porter (who carried our luggage)
Japanese: [the ] (our luggage carried) porter.
Hence, the follow relative clause sentences from Japanese all exhibit the predicted pattern. In relative clauses, however, the SM is no rather than ga.
Kare no kaita hon
[he SM write-past book]
The book that he wroteTakai ki no aru shima.
[high tree SM have island]
An island on which there are high trees.Kesa anata no mita Asahi.
[This morning you SM see-past Asahi (noted Japanese newspaper)]
The Asahi newspaper which you saw this morning.Boku no kaita tegami.
[I (male form) SM write-past letter]
The letter which I wrote
The Japanese language creates special problems for GB Theory. When it comes to creating a LF of the language, one finds that the question words, and quantifiers are not moved. They are not subject to a movement rule. Since the LF of a sentence must have a quantifier or a wh-constituent left-positioned as an adjunct on a sentence (IP), This means that it must be moved to this location even though the language has no movement rules. This is further exacerbated by the fact that the semantic interpretation of Japanese makes more sense if the scope position is right-peripheral rather than left-peripheral.
S-structure: John ga dare wo butta ka shiranai.
[John SM who OM hit-past QM know not]
I don't know who John hit.Japanese LF: [ [ John xi butta ] ka dare-woi ] shiranai
Required LF: [dare- woi [John ga xi butta ] ka] shiranai
This whole situation is ironic. Chomsky wants to simplify grammars. He has argued repeatedly that transformational rules were too complex. He wanted to create a schema I which these powerful rules could be constrained. This was done by restricting movement into empty categories and by creating base-generated pronouns and wh-constituents. These changes meant that many transformational rules were eliminated. Now, he has a situation is which to movement rule is needed in the transition from the original level of the d-structure to derived level s-structure in Japanese. No transformations are needed. When it comes to his LF component of the Universal Grammar, however, he is forced to introduce movement rules where none were previously needed. This seems to run counter to his quest to simplify grammars by removing, reducing, or constraining transformations.
| WH-RAISING IN CHINESE |
The Chinese language provides another example in which movement rules are not necessary when dealing with wh-constituents and quantifiers. Chinese is similar to Japanese in having topic markers, subject markers, and object markers. It differs in many aspects, however, in its word order. The following are examples from Mandarin:
Lào Wáng yè shì xuéshêng ma?
[old Wang also be student QM]
Is Lao Wang also a student?Tâ màì f ângzì le ma?
[3rd person buy house currently relevant state QM]
Did s/he buy a house?
The particle le signals a current relevant state (CRS). This means that the state of affairs has a special current relevance with respect to some particular situation.
Zhèi ge guâ hên tián
[this classifier melon very sweet]
This melon is very sweet (expected state of affairs)Zhèi ge guâ hên tián le
[this classifier melon very sweet CRS]
This melon is very sweet (unexpected state of affairs)wô zhîdào
[I know]
Yes, I knowwô zhîdào le
[I know CRS]
Now, I know.
Now, consider examples in which Chines has no overt wh-movement.
Wô xiang-zhîdào Lisi mai-le sheme.
Wô xiang-zhîdào [CP [ Lisi mai-le sheme] ].
[I wonder Lisi buy-CRS what]
I wonder what Lisi bought.Zhangsan wen shei mai-le shu.
[Zhangsan wen [ shei mai-le shu ] ]
[Zhangsan asked who buy-CRS book]
Zhangsan asked who bought books.Zhangsan ziangzin shei mai-le shu
[Zhangsan ziangzin [ shei mai-le shu] ]
[Zhangsan believe who buy CRS book]
Who does Zhangsan believe bought books?
In English the wh-constituents (who, what) must be moved into the (spec, CP) of the embedded sentence. In Chinese, however, no movement is necessary (CF. Charles Li and Sandra Thomson: Mandarin Chinese: A Functional Reference Grammar). According to the requirement of LF within Universal Grammar, however, movement is needed to reposition the wh-constituent into a left-peripheral position.
D-structure: [Wô xiang-zhîdào [CP [ Lisi mai-le sheme] ] ]
LF: Wô xiang-zhîdào [CP sheme [ Lisi mai-le xi]
I wonder [what [Lisi bought xi ] ]I wonder what Lisi bought.
D-Structure: Zhangsan ziangzin shei mai-le shu
LF: Zhangsan ziangzin [ [shei]i [ xi mai-le shu ] ] ]
Zhangsan asked [ [ who]i [xi bought books ] ] ].Zhangsan asked who bought books.
There are two kinds of movement rules. One of them occurs between d-structures and s-structures. The other occurs between s-structures and LFs. To distinguish between these two kinds of movements, Chomsky called the first type of syntactic movement Overt and the second type of logical movement as Covert. In Japanese and Chinese there is no overt movement, but both are supposed to undergo covert movement.
| MULTIPLE WH-MOVEMENT |
What did John give to whom?
When did John say what?
In these sentences there are two wh-constituents. One of these is moved to [Spec, CP] and the other remains in situ (in place). The wh-constituent which is moved is in scope position and the other which has not moved remains in situ. The reason why the other wh-constituent does not move is because English does not allow multiple syntactic movements. At the level of logical form, however, both wh-constituents are moved into a scope position. Both function as operators which bind variables.
Which of these two wh-constituents will function as the operator (Spec, CP)? This question is answered by the claim that the lower Spec (what) percolates up into the higher (Spec) in order for it to function as a sentential operator. This process is called wh-absorption. French differs from Japanese and English in a very interesting way with regard to overt syntactic movement. It is required in embedded sentences, but it is optional in matrix sentences.
FRENCH: REQUIRED IN EMBEDDED SENTENCES
Je me demande qui tu as vu
[I myself ask who you have seen]
I wonder who you have seen?* Je me demande Sylvia a vu qui?
FRENCH: OPTIONAL IN MATRIX SENTENCES
Qui as tu vu?
[who have you seen?]
Who did you see?Tu as vu qui?
[you have seen who?]
Who did you see?
Now it is time to return to the question about multiple overt syntactic movement. It is interesting to note that in a language such as Polish, multiple syntactic movement of wh-constituents do occur.
[who what does?]
Who does what?
Why do some languages such as Japanese and Chinese delay movement of quantifiers and wh-constituents until the LF Component where they under covert movement? Why do other languages such as Polish move their quantifiers and wh-constituents before they are subjected to a logical form? The answer to these questions were stated in terms of principles within Universal Grammar. Pesetsky (1989) believes that it is natural for early movement to occur and called this tendency the Earliness Principle. Chomsky (1991), on the other hand, felt that it was natural for languages to delay movement and called this his Principle of Procrastination.
| RECONSTRUCTION AND THE MINIMALIST PROGRAM |
Overt syntactic movement is more costly and less efficient with the Minimalist Program and as a consequence it must be delayed as long as possible. This means that in some languages syntactic movement must occur earlier than in other languages. It occurs earlier in French, for example, as verb movement because of agreement needs which have to be reflected as inflectional endings.
The French verb (embrass-) is moved to T (embrass-ai-) and then to AGR (embrass-ai-t) where it gives rise to the following structure:
FRENCH:
[ Violetta [ [ [ embrass]ai]t] [V-trace [NP-trace V-trace Alfredo] ] ]
In English, the lexical verb does not leave the VP. What happens instead is the inflectional endings AGR and T are lowered onto the V. This results in the following structure:
ENGLISH:
[ Violetta [ Agr-trace [T-trace [NP-trace kissed Alfredo]
Notice that sentences (IP) is now AGRP. INFL has been modified into a tense phrase (TP) and AGR category for the marking of person and number. The subject of the sentence in this new model is part of the VP. This change is interesting because it follows the suggestion made by Burzio (1986) for dealing with unaccusatives in Italian.
| BURZIO'S TREATMENT OF VERB CLASSES |
Burzio arrived at this classification from his study of Ne-Cliticization in Italian. Clitics are attached to a verb head and the conditions under which this can occur is limited in Italian to NPs that are complements of Vs.
ITALIAN: Giacomo ha insultato due studenti.
Giacomo has insulted two students.Giacomo ne ha insultati due.
[Giacomo of-them has insulted two]
Giacomo has insulted two of them.Giacomo passa tre settimane a Milano.
Giacomo ne passa tre a Milano.
Giacomo spends three weeks in Milan.Giacomo resta tre settimane a Milano.
*Giacomo ne resta tre settimane a Milano.
Giacomo remains in Milan for three weeks.
Only NPs that are complements of verbs can be made into clitics. Burzio noted that this same condition also pertains to passive sentences.
Giacomo passa tre settimane a Milano
Giacomo spent three weeks in Milan.Giacomo ne passa tre a Milano
Giacomo spent three of them in Milan.Tre settimane sono state passate a Milano.
Three weeks were spent in Milan.
Sentences in which verbs are not followed by NP complements cannot be made into passives.
*Tre settimane sono state rimaste a Milano.
*Three weeks have remained in Milan.
The story of unaccusatives does not end here. They are also able to explain two different kinds of Subject NPs in Italian. It has been assumed that there is free inversion between subject NPs which occur before verbs and those which occur after. Burzio disagrees. He claims that they have different underlying structures.
Preverbal NP Subject in Transitive Sentences
La ragazaa l'ha comprato.
[the girl it has bought]
The girl has bought it.Post Verbal Subjects NP in Transitive Sentences
L'hanno comprato tre ragazze.
[it have bought three girls]
Three girls have bought it.Preverbal and Post Verbal NP Subjects in Passives
Molti studenti furono arrestati.
[many students were arrested]
Many students were arrested.Furono arrestati molti studenti
[were arrested many students]
Many students were arrested.Ne-Cliticization with Post Verbal Subjects in Passives
Ne furono arrestati molti
[of them were arrested many]
Many of them were arrested.
The clue to solving this problem can be found in the clitic formation (ne). These forms can only occur under special conditions. The NP which is extracted and made into a clitic must occur as the complement of the verb. This means that the structure underlying Ne furono arrestati molti consists of a verb and its complement.
The significance of this discovery will soon become obvious. It is known that post-verbal subjects are derived from one-argument verbs. The question is which one? Burzio noted that one class of one-argument verb is the intransitive and the other was the unaccusative. Which of these will explain preverbal and post verbal NP subjects in Italian? The answer can be found in the following patterns.
Molti studenti telefonano (Preverbal Subject NP)
Telefonano molti studenti (Post Verbal Subject NP)
Many students are callingMolti studenti arrivano (Preverbal Subject NP)
Arrivano molti studenti (Post Verbal Subject NP)
Ne arrivano molti (Ne Cliticization of Verb Complement)
Many students arrive. .
What these sentences demonstrate is that the NPs in sentences with the verb telefonare function as external arguments. The subject NP in sentences with the verb arrivare function as internal arguments. They are complements of the V the in VP.
If the NP is moved into the empty category, it appears at the s-structure level as a preverbal NP subject. If it does not move, it remains at the s-structure level as a post verbal NP subject. Both arrivare and telefonare are one place predicates, but they differ in their theta grids. The subject NP associated with the verb telefonare has an external argument. The subject NP associated with the verb arrivare has an internal argument.
The preverbal subject in telefonare are filled by the external argument. The subject position in arrivare, on the other hand, is not filled. It consists of an empty category. Although both verbs are one place predicates and although both verbs are appear to be intransitives, they differ in their underlying structure. Telefonare is an intransitive verb and arrivare is an unaccusative verb.
[molti studenti [ INFL [telefonare] ]
[ e [ INFL [ arrivare [molti studenti ] ] ]
Now comes the surprising ending. Burzio claims that there are two lexical entries for telefonare. In one of them the subject NP contains an external argument; in the other it belongs to an internal argument. The first shows up as a preverbal Subject NP and the latter shows up as a Post Verbal Subject NP if no movement is involved. If movement of the NP is involved to fill the empty category slot, the result is a preverbal Subject NP.
Telefonare, (NP ____ ), <Agent>
telefonare, ( ____NP ), <Theme>
Hence, the Subject NPs in constructions with the one place predicate telefonare may originate from either of the following d-structures.
[ e [ INFL [ telefonare [molti studenti ] ] ]
It is interesting to note that Chomsky originally rejected this analysis of unaccusatives. He was the exception. Many of his students accepted the Burzio reformulation. It is interesting to note that this analysis now underlies the VP construction within the Minimalist Program. The difference, however, is that Burzio wanted to limited passive verbs and raising verbs to the class of unaccusatives. Chomsky appears to treat all subject NPs as being derived from a VP in the new framework.
| REFERENCES |
