Functional Categories and Parametric Variation

Jamal Ouhalla

• Publisher: Routledge
• Year: 2003
• ISBN:
  9781134934744
• Edition: 1
• Topic: Filología
• Subtopic: Lingüística

This book explores the idea that functional categories are the flesh and blood of grammar'. From within the context of the Principles and Parameters framework put forward by Chomsky and others, Jamal Ouhalla develops the argument that much of what we understand by the term grammar and grammatical variation involves functional categories in a crucial way. His main thesis is that most, if not all, of the information which determines the major grammatical processes and relations (movement, agreement, case, etc.) and consequently parametric (or crosslinguistic) variation is associated with functional categories. By identifying parameters with a limited set of lexical properties associated with a well-defined group of functional categories, the book offers a new and highly constrained version of the theory of Lexical Parametrization. Dr Ouhalla begins by identifying a set of lexical properties which distinguish functional categories from substantives, arguing that each of them represents a parameter in its own right. He then goes on to argue on the basis of evidence drawn from a broad range of languages that functional categories, most of which are bound morphemes, behave in important respects like independent syntactic categories, and therefore should be assigned a full categorial status on a par with substantives. The remainder of the book contains detailed discussions of how this conclusion, together with the theory of Lexical Parametrization developed, account naturally for some major typological differences having to do mainly with word order in sentences and noun phrases. Although the various discussions it contains are conducted within the Chomskyan framework, Functional Categories and Parametric Variation is comprehensible to linguists of all theoretical persuasions. It is an original and important contribution to syntactic theory in general.

1
Introduction

1.1 GOALS AND OUTLINE

In the Principles and Parameters framework the idea that functional categories play a significant role in the grammar has for many years been one of the fundamental premises. Thus, C(omplementiser) and I(nflection) are assigned a crucial role in determining, among other things, movement processes, e.g. movement of wh-phrases to Spec of CP in wh-questions, V-movement to C in interrogative sentences, NP-movement to Spec of IP in raising and passive constructions etc. However, only recently has a serious attempt been made to investigate the role of functional categories in determining language variation (cf. Fukui and Speas (1986), Baker (1988), Chomsky (1988), Ouhalla (1988c) and Pollock (1989)). ¹ Among the consequences of this attempt is a growing awareness among linguists that language variation is probably to a large extent, if not exclusively, determined by functional categories. This awareness is best illustrated by the following quote from Chomsky (1988): ‘If substantive elements (verbs, nouns, etc.) are drawn from an invariant universal vocabulary, then only functional elements will be parametrised’ (See Page). In a theory where language variation is accounted for in terms of variation in the values of given parameters this statement surely has wide ranging implications.

The research programme carried out in this work is intended to shed light on the nature of the parameters responsible for certain major aspects of linguistic variation. It is argued with respect to the phenomena discussed that parametric variation reduces, to a large extent, to variation in the lexical properties of the functional categories involved. These properties are shown to interact, in a well defined manner, with the general principles of UG, giving rise to clusters of surface phenomena which may be instantiated inside the same language or across language types. The end result of this approach to parametric variation is a theory of typology which bases itself exclusively on the properties of functional categories. It is argued that linguistic differences involving substantive elements are determined by differences in the properties of functional categories. As we will see in the subsequent chapters, classifications based on substantives are both misguided and misleading. ²

The programme outlined above obviously presupposes a well-defined set of functional categories, given the crucial role they are assigned in determining grammatical processes as well as parametric variation. It also presupposes a well-defined set of lexical properties which characterise functional categories, and determine the range of possible variation. Below in this chapter a formal theory is developed which makes a clear-cut distinction between functional categories and substantives in terms of their inherent/lexical properties. This distinction will make it possible for us to identify functional categories on the basis of a set of distinctive lexical properties. The latter will be argued to be the dimensions or parameters which define the range of possible variation. The approach to parametric variation developed rests on two fundamental claims. First, parametric variation affects only (a limited set of) functional categories, to the exclusion of substantives, and second, it can only affect one property or a combination of properties drawn from a limited set of lexical properties. The product of these two claims is a theory of parametrisation which is highly constrained, in the sense that it severly restricts the range of possible variation. This theory is then embedded into a general theoretical framework, namely the Principles and Parameters framework as developed in Chomsky (1981), (1982), (1986a), (1986b), (1988), and works by various other researchers.

Chapter 2 is basically an attempt to demonstrate, on the basis of empirical evidence, that certain functional categories which generally appear as affixes attached to the verb are indeed syntactic categories in their own right. The evidence discussed consists of phenomena, drawn from a relatively broad range of genetically and typologically unrelated languages, which are shown to receive principled explanations only if the functional categories involved are assigned an autonomous syntactic status each, along with substantives; that is only if each of these elements is assumed to project its own Xbar structure. Among the functional categories discussed are NEG(ation), Tense (TNS), AGR(eement), ASP(ect) and the PASS(ive morpheme). Certain aspects of linguistic variation relating to the periphrastic versus morphological distinction in active and passives sentences are shown to reduce to variation in the lexical properties of ASP and PASS.

Chapter 3 discusses the nature of the parameters responsible for differences in the surface order of substantives, in particular the subject and the verb. It is argued that the difference in the surface order between VSO and SVO languages reduces to a minimal difference in the selectional properties of the AGR and TNS categories. This parametric difference is shown to account for a cluster of properties which characterise each of the two groups of languages. On the basis of these properties it turns out that the Celtic languages, which have traditionally been considered archetypal examples of genuine VSO languages, have more properties in common with SVO languages than with genuine VSO languages.

Another aspect of variation in the surface order of substantives discussed in this chapter concerns negative clauses. It is shown that languages tend to fall into at least two different typological groups depending on the position of the NEG category in the clause structure. This difference is argued to be determined by a difference in the selectional properties of the NEG category, in the sense that NEG selects a specific category in one group of languages and a different category in the other. Languages are also shown to differ with respect to whether the NEG element is affixal in nature or not, that is whether it is a bound or a free morpheme, a distinction which falls under the scope of the lexical property of morphological selection. It is argued that the affixal/non-affixal distinction plays a crucial role, as a result of an interaction with the general principles of UG, in determining the movement possibilities of the verb responsible for variation in the derived order.

Chapter 4 deals with variation relating to the structure and word order in noun phrases. It is argued that variation in the surface order of substantives in noun phrases, in particular the subject and the head noun, is also determined by variation in the lexical properties, most notably the selectional properties, of the functional categories involved, e.g. Det, AGR etc. The structures of noun phrases worked out on the basis of the selectional properties of functional categories, and the ways they tend to differ, turn out to parallel those of sentences discussed in the previous chapter, thus capturing the many respects pointed out in the literature in which noun phrases seem to resemble sentences. The final section of this chapter deals with the properties of Comp elements, in particular their status as functional categories and their role in the grammar. It is argued that Comp elements are basically nominalisers, that is nominal elements whose function is to nominalise argument clauses which are otherwise verbal due to the presence of a verbal TNS element. In this respect Comp elements parallel the nominalising elements used in gerundive noun phrases, hence the inclusion of this section in the chapter dealing with noun phrases.

1.2 PARAMETERS AND LANGUAGE VARIATION

Within the Principles and Parameters framework language variation is accounted for in terms of variation in the values of parameters of some sort or another. As far as this point is concerned there seems to be a fair degree of consensus among researchers. Differences of opinion arise when the nature of the elements with which parameters are associated is considered. In this respect one can discern at least two major opinions. One opinion is expressed in Chomsky (1986a) whereby parameters are associated with the principles of UG. Chomsky uses the ‘switch’ metaphor to explain the nature of these parameters, in the sense that they are conceived of as ‘switches’ with a finite number of open positions which are set on the basis of direct and positive experience. One such parameter frequently cited in the literature is the Head Parameter of X-bar theory which regulates the position of head categories in relation to their complements. This parameter is assumed to have two values, Head-first and Head-last, each of which accounts for the two patterns of word order, Head-Complement and Complement-Head, found in languages such as English and Turkish, respectively. In terms of this parameter the typological difference between these two types of languages is accounted for in a principled way.

The other opinion has been developed by Borer (1983) whereby parametric variation is argued to affect only the ‘inflectional system’ of languages. In other words, Borer’s is an attempt to ‘reduce all interlanguage variation to the properties of the inflectional system’, so that ‘Languages will differ in the availability of particular inflectional rules…’ (p. 27). Borer illustrates her approach to parameteric variation with the following examples from Lebanese Arabic and Hebrew, respectively:

(1)

1. a. hkit ma?-o la Karim.
   talked-I with-him to Karim
   ‘I talked with Karim.’
2. b. *dibarti ’im-a (le/s.el) Anna
   talked-I with-her to/of Anna
   ‘I talked with Anna.’

Both examples involve clitic-doubling of the object of the preposition. In Borer’s theory clitics absorb the case feature of the head they are attached to, so that in both sentences the object of the preposition is deprived of Case, giving rise to a potential violation of the Case Requirement on noun phrases. Borer argues that the two languages differ in that Lebanese Arabic makes use of the rule of preposition insertion as a ‘saving device’, while Hebrew does not. The inserted preposition assigns Case to the doubled noun phrase, thereby avoiding a violation of the Case Requirement. Thus, the availability or nonavailability of doubling in prepositional phrases results from ‘the presence versus absence of an inflectional rule…’, namely the rule which inserts the preposition.

One of the fundamental aspects of Borer’s approach to parametric variation is that it associates parameters with individual lexical items, as part of the information included in their lexical entries, rather than with the principles of UG. As Wexler and Manzini (1987) point out, among the interesting consequences of this approach is that a given language can be expected to instantiate more than one value of a given parameter in terms of different lexical items. That this is a desirable consequence is shown by the distribution of anaphoric elements. Wexler and Manzini show that binding domains (governing categories), that is the configurations which include an anphoric element and a possible antecedent, tend to differ from one lexical item to another not only across languages but inside the same language. In other words, assuming that binding domains are subject to parametric variation, there are empirical reasons to believe that two lexical items from the same language may select different values of the parameter in question.

The theory of parametrisation developed in the present work is basically an attempt to develop further Borer’s approach to parametric variation. It assumes that parameters are associated with individual lexical items as part of the information specified in their lexical entries, and that the set of lexical items with which parameters are associated is restricted to the class of inflectional/functional categories. The first assumption is needed to explain the fact discussed above in relation to Wexler and Manzini that a given language may instantiate more than one value of a given parameter in terms of different lexical items. In the subsequent chapters we will discuss a number of parametric differences which are instantiated inside the same language, with the purpose of conveying the message that the phenomenon is more pervasive than is otherwise believed. On the assumption that this is indeed the case, it should be clear that any adequate theory of parametrisation should accommodate this apparently fundamental property of human languages.

The second assumption is intended to give further substance to the claim which underlies the present work, namely that functional categories represent the flesh and blood of grammar, in the sense that they are the locus of grammatical information which determines the structural representation of given constructions, as well as the various grammatical processes they may undergo. The theory of parametrisation outlined above strengthens this claim by assuming that functional categories are also the locus of information which determines parametric variation, an assumption which we believe should be intuitively true, at least in spirit. Substantives are known to have uniform properties across languages. Thus, in all languages the verb give, for example, selects two arguments as complements, a fact which follows from its conceptual/semantic structure. On the other hand, functional categories are known to have idiosyncratic properties, that is properties which differ from one language to another (cf. Emonds (1985)). For example, we will see in the subsequent chapters that the selectional properties of functional categories differ from one language to another, along typological lines. That is, a given functional category may select a specific category in one language and a different one in another, thus giving rise to a difference in the arrangements of these categories in the structure. It is precisely differences of this type, we will argue, which are responsible for some major aspects of linguistic variation.

Although the theory of parametrisation adopted here borrows heavily from Borer’s approach it differs from it in two major respects. The first difference concerns the notion ‘inflectional system’. As we saw above, Borer’s understanding of this notion involves, among other things, a set of ‘inflectional rules’ of the type illustrated above with respect to examples (1a&b). In the present context, the notion ‘inflectional system’ is understood simply as a set, a list, of inflectional or functional categories, with a number of lexical properties associated with them. For reasons which will become clear later this system crucially excludes any type of rules, be they inflectional or otherwise.

The second difference concerns the nature of the lexical information which determines parametric variation. In the present context this information is considered to be nothing other than the usual type of information relating to selection and grammatical features, that is information which is independently needed for familiar reasons. In other words, it is not information which is available over and above the familiar type of lexical properties; rather, these properties themselves determine parametric variation. To illustrate with a concrete example, variation in the selectional properties of a given functional category is considered in the present context as representing a parametric variation involving the (lexical) parameter of selection. Likewise, variation in the categorial features of a given category constitutes a parametric difference involving the parameter of categorial features, and so on. This approach amounts, in the long run, to an attempt to identify parameters with (a restricted set of) lexical properties. This set of lexical properties is isolated and defined in the next section. ³

By appealing to lexical properties which are independently needed, the present approach to parametric variation is basically an attempt to contribute to the ultimate goal of developing an overall model of grammar which is optimally simple. However, this is by no means the only respect from which it seeks to derive its legitimacy, important though it is. Equally importantly, its legitimacy also derives from what is so far a putative ability to account for certain aspects of language variation which otherwise remain obscure or have received non-explanatory or unsatisfactory treatment in the literature. In other words, the present work tries to justify itself on both theoretical and empirical grounds. Whether it is successful in doing so depends, obviously, on its ability to provide principled accounts for the phenomena investigated.

1.3 TOWARDS A THEORY OF FUNCTIONAL CATEGORIES

1.3.1 Functional categories versus substantives

Traditionally, categories are divided into two major classes, ‘open’ and ‘closed’. The open class contains the major lexical categories V, N and A, that is the class of categories which we have been referring to in this work as substantives. The rest of the categories, e.g. Comp, Infl, Aux, Det etc., that is the categories which we have been referring to as functional categories, belong to the closed class. ⁴ A number of properties have been pointed out in the literature which are thought to distinguish between these two types of categories. The following two are widely recognised as the most important of them:

(2) Emonds (1985)

1. (i) Only open categories have indefinitely many members in the dictionary of a language—say several hundred at least. Closed categories have twenty to thirty members at most.
2. (...

Table of contents

• Cover Page
• Title Page
• Copyright Page
• Preface
• 1: Introduction
• 2: The X-bar status of the inflectional elements and the clause structure
• 3: On the nature of some word order parameters
• 4: The structure of noun phrases and word order variation
• Notes
• Bibliography