Passive Agent Deletion.

Universal Grammar

The grammar that Chomsky (1957) developed consists of a complex set of trans-formation rules. The goal of developing a cognitive theory of language necessarily required that the rules apply to all languages, given that human cognition is the same regardless of language or culture. Chomsky (1995) therefore argued that the rules were linked to a universal grammar. We might be tempted immediately to conclude that universal grammar is identical to the linguistic universals associated with traditional grammar, but this would be a mistake. There is a connection, but a tenuous one. Linguistic universals refer to a relatively narrow range of shared features across languages, such as the fact that all languages have subjects and predicates, kinship terms, and a means of indicating when actions occur. Universal grammar is different. In Chomsky’s (1995) words:

The human brain provides an array of capacities that enter into the use and understanding of language (the language faculty); these seem to be in good part specialized for that function and a common human endowment over a very wide range of circumstances and conditions. One component of the language faculty is a generative procedure … that generates structural descriptions (SDs), each a complex of properties, including those commonly called “semantic” and “phonetic.” These SDs are the expressions of the language. The theory of a particular language is its grammar. The theory of languages and the expressions they gener-ate is Universal Grammar (UG); UG is a theory of the initial state … of the relevant component of the language faculty. (p. 167)

To be fair, we need to remember that phrase-structure grammar grew out of attempts in the late 19^th and early 20^th centuries to preserve American Indian tribal languages. The goal was to preserve the body of the languages as they were spoken—it was not to develop a theory of language or grammar. In fact, phrase-structure grammarians like Bloomfield (1933) were wary of universal-grammar claims because in the past they had resulted in distortions in the records of investigated languages. Nevertheless, Chomsky’s critique resonated strongly among scholars, in part because the alternative he proposed was elegant, powerful, and offered exciting new lines of research.

Today, almost 50 years later, the grammar Chomsky proposed to replace phrase-structure is still vibrant and, indeed, remains a significant factor in American language study. However, it does not have the same allure that it once had. One reason is that, over the years, Chomsky revised the grammar numerous times, which should have been viewed as perfectly reasonable and in keeping with scientific principles but which nevertheless has often been viewed as quirkiness. In addition, the revisions made the grammar more abstract and thus more difficult to understand for anyone without significant training in linguistics.

This chapter cannot provide an in-depth analysis of the grammar and all of its permutations; instead, it will offer an overview that traces some of the significant features of the grammar from the initial formulation in 1957, concluding with an examination of Chomsky’s latest version.

DEEP STRUCTURE AND SURFACE STRUCTURE

In Syntactic Structures, Chomsky (1957) hinted that grammatical operations related to language production work in the background. We do not really see them at work; we see only the consequences of their application on an underlying structure. In Aspects of the Theory of Syntax (1965), Chomsky developed this proposal by resuscitating the prestructuralism idea that there is something underneath language, some universal feature of the human mind, such as logic, that determines the substance of utterances. This argument effectively ad-dressed the problem presented by actives and passives. A passive sentence like Fritz was kissed by Macarena will have its corresponding active, Macarena kissed Fritz as an underlying structure. This structure is then transformed to the passive through a grammatical transformation rule.

Chomsky identified a basic grammatical structure in Syntactic Structures that he referred to as kernel sentences. Reflecting mentalese, or logical form, kernel sentences were where words and meaning first appeared in the complex cognitive process that resulted in an utterance. However, the overall focus in Syntactic Structures was syntax, not meaning. In fact, Chomsky indicated that meaning was largely irrelevant, as he illustrated in the sentence, “Colorless green ideas sleep furiously” (1957, p. 15).1 It means nothing but is nevertheless grammatical.

As impressive as Syntactic Structures was, the idea that any theory of language could ignore meaning was difficult to accept. Chomsky (1965) responded to the criticism in Aspects of the Theory of Syntax, in which he abandoned the notion of kernel sentences and identified the underlying constituents of sentences as deep structure. Deep structure was versatile: It contained the meaning of an utterance and provided the basis for transformation rules that turned deep structure into surface structure, which represented what we actually hear or read. Transformation rules, therefore, connected deep structure and surface structure, meaning and syntax.

Central to the idea that transformation rules serve as a bridge between deep structure and surface structure was the notion that transformations do not alter meaning. If they did, it would be difficult to justify the rules. Not only would they interfere with understanding, but they also would fail to realize Chomsky’s goal of developing a grammar that looks into the history of a sentence. Deep structure was a convenient means of countering an alternative and nagging argument: that meaning is in the surface structure, that the words we hear and read mean pretty much what the person who created them intended.

Understanding the consequences of this argument is important. If meaning is in the surface structure, there is no need for a mediating structure between mind and utterances. Transformation rules become irrelevant. However, it was clear that some transformations did change meaning. In the early version of the grammar, negatives are generated from an underlying affirmative through a transformation rule. That is, the negative transformation turns a positive statement into a negative one, as in these sentences:

4. Maria wanted to dance with Raul.

4a. Maria did not want to dance with Raul.

The deep structure of 4a is 4, and the meanings are clearly different. The question transformation results in a similar change, turning an assertion into a question. Sentences like these presented a big problem for T-G grammar. Just prior to the publication of Aspects, Lees (1962) and Klima (1964) proposed that such difficulties could be eliminated by specifying certain phrase-structure markers in the deep structure of sentences like 4a, which triggered transformation. These markers—governing, for example, negatives and questions—were hypothesized to reside in the deep structure of all utterances and were said to be activated by contextual cues. Once activated, they triggered the transformation. The result is that sentence 4a would not have sentence 4 as its deep structure but instead would have sentence 4b:

4b. neg Maria wanted to dance with Raul.

This approach solved the problem in a clever way, and Chomsky adopted it. But the solution was highly artificial and not very satisfactory. In fact, it created more problems than it solved. Markers for questions and negatives seem straightforward, but we have no way of determining what kind of markers would govern such sentences as the following, which also undergo a change in meaning as a result of transformation:

• To solve the crossword is difficult.

• The crossword is difficult to solve.

T-G grammar specifies that the second sentence is derived from the first through what is called the object-raising transformation. (The crossword functions as the subject in the second sentence but as an object in the first.) In the first sentence, the focus is on the process of solving the crossword, whereas in the second it is not. Thus, the meaning of the first sentence can be generic; in the second, it cannot.

Or consider the following:

• Fritz gave the flowers to Macarena.

• Fritz gave Macarena the flowers.

• Fred cleared the table for his mother.

• *Fred cleared his mother the table.

How would markers account for the fact that the transformation that de-rived the grammatical Fritz gave Macarena the flowers from Fritz gave the flowers to Macarena also produces the ungrammatical Fred cleared his mother the table?

Equally problematic is that psychological research on language processing could find no evidence of markers of any type in language. It also failed to find any evidence that meaning resides anywhere other than in the surface structure. The rationalist response has been that such evidence counts for very little, but there also is no intuitive basis for specifying such markers in the deep structure. Thus, these problems remained unsolved.

THE BASICS OF TRANSFORMATION RULES

For the time being, let’s set aside the issue of meaning in a theory of language and grammar and turn to the transformation rules themselves. Transformation rules have undergone significant change over the years. Necessarily, this section serves merely as an introduction to some of the rules in Chomsky’s early work. Later in the chapter, we consider the current approach to transformations. Thus, the goal here is to provide some understanding of the general prin-ciples of T-G grammar rather than an in-depth analysis.

In Syntactic Structures and Aspects, Chomsky (1957, 1965) proposed a variety of transformation rules, some obligatory and others optional. The rules themselves specify their status. Rather than examining all possible transformation rules, only a few are presented, those that govern some common constructions in English. Be-fore turning to these rules, however, it is important to note that transformations are governed by certain conventions. Two of the more important are the ordering convention and the cycle convention. When a sentence has several transformations, they must be applied in keeping with the order of the rules. In addition, when a sentence has embedded clauses, we must begin applying the transformations in the clause at the lowest level and work our way up. This is the cycle convention. Failure to abide by these conventions when analyzing structure with T-G grammar may result in ungrammatical sentences. What we see in T-G grammar, therefore, is a formalistic model of language production that employs a set of rigid rules that must operate in an equally rigid sequence to produce grammatical sentences.

The Passive Transformation

The relation between actives and passives was an important part of Chomsky’s (1957) critique of phrase-structure grammar, so it is fitting that we examine the rule that governs passives first. Only sentences with transitive verbs can be passivized, and we always have the option of keeping them in the active form, which means that the passive transformation is an optional rule.

Consider sentence 5:

5. Fred bought a ring.

If we change this sentence to the passive form, it becomes:

5a. A ring was bought by Fred.

In keeping with the early version of T-G grammar, sentence 5 represents the deep structure of 5a. The process of the transformation is as follows: First, the object NP (a ring) shifted to the subject position. Second, the preposition by appeared, and the deep-structure subject (Fred) became the object of the preposition. Third, be and the past participle suffix appeared in the auxiliary, turning the deep structure verb buy into a passive verb form.

The grammar rule represents these changes symbolically. In this rule, the symbol fi means “is transformed into”:

Passive Transformation Rule.

NP1 Aux V NP2 (Fred bought a ring)

NP2 Aux + be -ed/en V by + NP1 (A ring was bought by Fred)

With respect to sentence 5:

NP1 = Fred

NP2 = a ring

V = bought

T-G grammar is predicated on examining the history of a given sentence, and the most effective way of doing so is through tree diagrams, which allow us to examine the deep structure and its corresponding surface structure. The process, however, is different from phrase-structure analysis because it requires a minimum of two trees, one for the deep structure and one for the surface structure. For more complicated sentences, there are more trees, each one reflecting a different transformation and a different stage in the history of the sentence. A convenient guideline is that the number of trees in a T-G analysis will consist of the number of transformations plus one.

Passive Agent Deletion.

In many instances, we delete the agent in pas-sive sentences, as in sentence 6: 6. The cake was eaten. When the subject agent is not identified, we use an indefinite pronoun to fill the slot where it would appear in the deep structure, as in 6a:

6a. [Someone] ate the cake.

This deep structure, however, would result in the surface structure of sentence 6b:

NEGATIVES

Although there are many ways to say no in English even when appearing to say yes, grammatically we form the negative using no, not, and never. Technically, these words are adverbials, but phrase-structure grammar analyzes them as negation markers in the Aux, as the following phrase-structure rule shows:

Aux tense (neg) (DO) (M) (prog) (perf)

neg no

not

never

An interesting feature of the negative is that it triggers Do Support in the verb phrase of simple active sentences. Consider these examples:

71. Fred kissed Macarena.

71a. Fred did not kiss Macarena.

Sentence 4.

68: The triangle in which they were embroiled defied logic.

Strangely enough, negation does not have this effect on progressive or per-fect verb forms, as the following sentences illustrate:

72. Buggsy is inviting Michael Star to his next party.

72a. Buggsy is not inviting Michael Star to his next party.

73. Buggsy had left the waiter a huge tip.

73a. Buggsy had not left the waiter a huge tip.

Дата добавления: 0000-00-00