Sie sind auf Seite 1von 29

1. 1.

Preliminaries Aims and scope of this study

This book presents a detailed corpus-based study of adverbial subordinate clauses in English within the framework of Functional Grammar (henceforth FG). Its aim is to demonstrate the relevance of FGs hierarchical model of the structure of the clause for the analysis of adverbial constructions by examining the systematic relationship between the semantic type of adverbial clauses and the way these are expressed in English. The method of analysis is largely based on the typological study of adverbial clauses carried out by Hengeveld (1998). The importance of this study not only lies in its contribution to the knowledge and typological classification of European languages, but also in its theoretical implications for FG. On the basis of an in-depth analysis of data obtained from the LOB-corpus, the present study shows that the expression of adverbial clauses in English runs parallel to the distribution of expression formats from a cross-linguistic perspective. The present work is organised as follows. Chapter 1 provides a description of the theoretical framework FG (1.2), of the object of analysis adverbial subordinate clauses (1.3), and of the data used for the analysis (1.4). The following two chapters deal with the classification of adverbial clauses. Firstly, the expression formats which are characteristic of these constructions in English are analysed (Chapter 2), and subsequently a semantic classification of adverbial clauses is provided using the theoretical framework of FG (Chapter 3). Chapter 4 studies the way in which the different types of adverbial subordinate clauses are expressed in English, in order to demonstrate, in Chapter 5, the existence of a systematic relationship between the semantic type of a subordinate clause on the one hand and the way in which it is expressed on the other. In the concluding Chapter 6 the results of this study are interpreted in terms of FG. It furthermore discusses the theoretical implications of the findings of this study for the FG model. 1.2 The Functional Grammar framework

This section provides a general overview of the basic principles of FG, with a focus on those aspects which are relevant to the study of adverbial subordinate clauses. The presentation of the FG model is based mainly on Dik (1997a/b), although reference will also be made to other works which, in one way or another, elaborate on or disagree with what is set out in that work. First the basic methodological principles of FG are presented (1.2.1), followed by a description of the general organisation of this model (1.2.2).

Mara Jess Prez Quintero

1.2.1 Methodological principles of Functional Grammar FG is a general model for the analysis of the organisation of natural languages which takes a functional view of language, defined as an instrument of social interaction among human beings, used with the intention of establishing communicative relationships (Dik 1997a: 3). Thus, FG adheres to the functional paradigm, as opposed to the formal paradigm, the main exponent of which is Chomskys formalist theory. In contrast to a formalist conception of language, in which language is considered an abstract object of study that must be analysed independently of its use, a functional approach to the study of language involves a different conception of the object of study, which is understood as an instrument of social interaction between natural language users (NLUs). While in the formal paradigm the psychological correlate of a language is competence, that is, the ability to construe and interpret linguistic expressions (Dik 1997a: 6) as opposed to performance, in the functional paradigm the psychological correlate of a language is communicative competence, his [NLUs] ability to carry on social interaction by means of language (Dik 1997a: 5). Hymes (1972), from whom Dik (1978) adopts the concept of communicative competence, claims that the attitude taken towards the conception of the object of study determines, to a great extent, the organisation of the different components of a linguistic theory. Regarding the formalist view of syntax as a module independent from semantics and pragmatics, Dik (1997a: 7-8) states that: in the functional paradigm the relation between the different components of linguistic organization is viewed in such a way that pragmatics is seen as the all-encompassing framework within which semantics and syntax must be studied. Semantics is regarded as instrumental with respect to pragmatics, and syntax as instrumental with respect to semantics. In this view there is no room for something like an autonomous syntax. On the contrary, to the extent that a clear division can be made between syntax and semantics at all, syntax is there for people to be able to form complex expressions for conveying complex meanings, and such meanings are there for people to be able to communicate in subtle and differentiated ways. Thus, it is clear that in the functional paradigm linguistic expressions must be described and explained in the context of verbal interaction, which in turn should be integrated within the more general framework of the NLUs cognitive capacity. Dik claims that such a model of analysis should aim at providing a functional explanation of linguistic phenomena and he adds: a functional explanation of grammatical phenomena will typically not be based on an assumption of simple form-function correlation, but will instead involve a network of interacting requirements and constraints, each of which may be understood in functional terms itself, but which interact in complex ways and in a certain sense

Preliminaries compete for recognition and expression in the final design of linguistic expressions. (Dik 1986: 7)

In Diks view, a functional grammar must analyse the properties of linguistic expressions in the context of their use and connect these properties with the rules and principles which govern verbal interaction. The final goal is to construct a Model of the NLU (M.NLU), capable of accounting for the abilities of speaker and addressee. These abilities form the competence of the NLU at three different levels: cognitive, pragmatic and grammatical. Accordingly, a functional Model of the NLU should take into consideration these three aspects, which constitute three theoretical components integrated into a modular system. FG, therefore, can be considered a modular theory in the sense suggested by Escribano (1992; 1993), who maintains that the concept of modularity has been used in a trivial way to refer to any theory that proposes different constituents of analysis. For this author a grammar is modular if and only if it features a many-to-many relationship between its systems (1993: 255-6). FG is then modular in the sense that a Grammatical Theory should be included within the framework of a Pragmatic Theory, which in turn is considered part of a Cognitive Theory. Gmez Solico (1995) points out that this modular system can be expressed in the following simplified manner:1

COGNITIVE THEORY Perceptual, logical and epistemological competence PRAGMATIC THEORY Communicative competence GRAMMATICAL THEORY Grammatical competence

Figure 1.1: Modularity in Functional Grammar (Gmez Solico 1995) Gmez Solico (1995: 203) argues that, according to this interpretation and contrary to what is postulated in Generative Grammar, the grammar of a language is not a restricted and autonomous system governed by independent principles which only partially interact with other human capacities.

Translated from the Spanish.

Mara Jess Prez Quintero

The criteria of explanatory adequacy postulated within FG are directly related to this functional conception of the organisation of natural languages. Dik (1997a: 12-5) proposed the following standards of adequacy: (i) (ii) (iii) Pragmatic adequacy: The theory should account for how people interact in verbal communication. Psychological adequacy: The theory must be compatible with models which explain the psycholinguistic processes of (de)codification. Typological adequacy: The theory should be valid for the analysis of any language, accounting both for the differences and for the similarities between different languages.

FG is not only considered functional because it is based on a functional view of the nature of language. Dik (1980) points out that this label is also due to the importance attributed, in the description of linguistic expressions, to functional or relational notions, as opposed to categorical ones. FG recognises functional relationships at three different levels: (i) Semantic functions: These specify the roles performed by the referents of the terms in the State of Affairs designated by the predication in which they appear. They include Agent, Positioner, Force, Processed and Zero (linked to the first argument), Goal (linked to the second argument) and Recipient, Location, Direction, Source and Reference (linked to the second or third argument). Syntactic functions: These specify the perspective from which the State of Affairs is presented in a linguistic expression. FG recognises only two syntactic functions, Subject and Object. Although Dik maintains the traditional terminology, these functions are interpreted communicatively to refer to the point of view from which the State of Affairs is presented. Thus the function of Subject is assigned to the argument from the perspective of which the State of Affairs is considered, while the function of Object would be related to a secondary perspective on that State of Affairs. Pragmatic functions: [These] specify the informational status of a constituent within the wider communicative setting in which it occurs (that is, in relation to the pragmatic information of S and A at the moment of use)2 (Dik 1997a: 26). FG distinguishes two types of pragmatic function, which are analysed in detail in Dik (1997b): (i) Extra-clausal functions (e.g. theme, initiatior), which are assigned to elements external to the predication, and (ii) Intra-clausal functions (topic and focus), which are assigned to constituents of the predication in terms of their informational status within the communicative setting in which they are used.

(ii)

(iii)

S represents Speaker and A Addressee.

Preliminaries

Dik (1997a) states that while semantic and pragmatic functions are universally relevant to all languages, syntactic functions are not universal in this sense, since there are languages which do not include subject and object assignment in their grammatical organisation. Nevertheless, these functions are universal in the sense that if a language makes use of them, this use follows some basic principles which are language independent. 1.2.2 General organisation of Functional Grammar Within FG clauses are described in terms of underlying structures, from which, through the application of expression rules, the corresponding linguistic expressions are obtained. This process may be represented in the following way (Dik 1997a: 49): (1) UNDERLYING CLAUSE STRUCTURE EXPRESSION RULES LINGUISTIC EXPRESSIONS The underlying clause structure constitutes a complex abstract structure in which, as will be shown below, different levels of organisation can be distinguished. Likewise, expression rules, which determine the final form of linguistic expressions, constitute a complex component of FG. The complexity of linguistic expressions is described from the most basic to the most complex elements in a bottom-up model of language production. The organisation of this model is represented in Figure 1.2, taken from Dik (1997a: 60). In this model, three basic components can be distinguished: the fund (1.2.2.1), the hierarchical structure of the clause (1.2.2.2) and the expression rules (1.2.2.3). 1.2.2.1 The fund The fund is the lexical component of FG which contains all the predicates and terms necessary for the construction of predications. In FG all lexical elements of a language are analysed as predicates, expressions which designate properties or relations between entities. Dik (1997a: 59) distinguishes three different categories of predicate verbal, nominal and adjectival on the basis of their formal and functional properties, to which Hengeveld (1992) adds the adverbial category. The predicates of a language can be basic, if they must be learned as such, and derived, if they can be formed through the application of a synchronically productive rule of predicate formation. Basic predicates constitute, along with basic terms, the lexicon of a language. However, in FG predicates are considered to be structures rather than isolated elements. These structures, called predicate frames, provide the information necessary for determining the type of predication in which a certain predicate can be used. This information consists of:

Mara Jess Prez Quintero


FUND

predicate formation

LEXICON derived basic predicates predicates PREDICATE FRAMES

basic terms

derived terms TERMS

term formation

nuclear predication

core predication
2 2

syntactic functions extended predication


3 3

proposition
4 4

pragmatic functions clause structure EXPRESSION RULES form order prosody LINGUISTIC EXPRESSION

Figure 1.2: General organisation of Functional Grammar (Dik 1997a)

Preliminaries (i) (ii) (iii) (iv) (v) The form of the predicate The category of the predicate (verbal, nominal, adjectival or adverbial) The number of arguments The semantic functions assigned to the arguments The selection restrictions imposed on the arguments

By way of illustration, Dik (1997a: 59) gives the following example of a predicate frame: (2) give [V] (x1:<anim>(x1))Ag (x2)Go (x3:<anim>(x3))Rec

Here the verbal predicate give (v) takes three arguments (x1, x2, x3) which carry the semantic functions of Agent (Ag), Goal (Go) and Recipient (Rec). The first and the third argument show the selection restriction <animate>. In the lexicon basic predicates are semantically related to one another through meaning postulates, which characterise the meaning of a predicate in terms of a unidirectional relation (e.g. bachelor not married Siewierska 1991), or through meaning definitions, if the relationship which is established is bidirectional (e.g. bachelor unmarried man Siewierska 1991). Apart from predicates, the fund in the FG model contains terms: linguistic expressions with referential potential. Prototypically, terms designate entities which exist in space, called first order entities. However, terms can also refer to entities of different orders, as is the case with: (i) properties or relations (e.g. intelligence), (ii) states of affairs (e.g. match), (iii) propositional contents (e.g. fact) or (iv) speech acts (e.g. question). In these cases of non-prototypical reference, terms can present a derived or complex structure, containing embedded predications, propositions or speech acts. A distinction should also be made between basic terms, contained in the lexicon and limited in number, and derived, which constitute the large majority of term structures and which are formed by means of term formation rules. The general format of derived terms can be represented as follows (Dik 1997a: 61): (3) ( xi:
1(x i): 2(x i):...: n(x i))

where represents one or more term operators (e.g. number, quantification, . . .), xi represents the referent, and each (xi) constitutes an open predication in x i (a predicate frame in which all positions, except that occupied by x i, have been occupied by terms), which restricts the possible values of x i. Summarising, the lexical component of FG contains two types of elements, predicates, with their corresponding predicate frames, and terms. Through the insertion of the appropriate terms in the positions of the argument slots of predicate frames, a nuclear predication is obtained. The structure of this predication, as well as the other expansions within the structure of the clause, is analysed below.

Mara Jess Prez Quintero

1.2.2.2 The hierarchical structure of the clause In her analysis of FG Siewierska points out that Inherent in the functional approach to language is the recognition of several layers of the structural organization of the clause corresponding to the multiple functions that the clause fulfils in the act of communication (1991: 36). She also mentions the fact that in the first version of the FG model (Dik 1978), the study of the clause was limited to its representational function, which was identified with the predication. However, in the second version of FG, Dik, inspired by the idea of the existence of different layers postulated by Foley and Van Valin (1984), recognises that the underlying clause structure is made up of a complex abstract structure in which several levels or layers of formal and semantic organization have to be distinguished (1997a: 50). Therefore, according to this model, originally introduced by Hengeveld (1989) and further developed by this author (Hengeveld 1990; 1992; 1997), any utterance can be analysed in terms of an underlying structure composed of the two levels shown in (4) (Hengeveld 1990): (4) (E 1: [ILL (S) (A) (X1: [ ] (X1))] (E 1)) (e1: [Predb (x1: Predn (x1)). . .(xn)] (e1)) The upper level, the interpersonal level, represents a speech act (E 1), which is structured on the basis of an abstract illocutionary frame (ILL) 3 with three arguments: (i) a speaker (S) who transmits (ii) a propositional content (X 1) to (iii) an addressee (A). This propositional content makes reference to a State of Affairs (e1), which constitutes the lower level, the representational level. This level is structured on the basis of a predicate frame (Pred b)4 with one or more argument positions filled by terms ((x 1). . .(xn)). 5 The hierarchical structure of the utterance given in (4) was later modified by Hengeveld (1992), who proposed the incorporation of a variable for every layer of the hierarchy, in order to represent the different levels in a uniform format. The modified representation of the hierarchical structure of the clause is shown in (5):

The units considered to be abstract illocutionary frames in Hengeveld (1990) are analysed as illocutionary operators in Dik (1997a). 4 In Dik (1997a/b) the symbol b , which represents the type of predicate, has been substituted by [T]. 5 The distinction between these two functions of language, interpersonal and representational (or ideational in Hallidays 1994 terminology) comes, as Butler (1996) points out, from Systemic Functional Grammar. In Role and Reference Grammar (Foley and Van Valin 1984; Van Valin 1993) such a distinction is not made explicit.

Preliminaries (5) (E1: [(F1: ILL (F 1)) (S) (A) (X1: [ ] (X1))] (E1)) (e1: [(f1: Predb (f1)) (x1: (f2: Predn (f2)) (x1))] (e1))

In this hierarchical structure different layers are distinguished, each associated with a specific designation and a variable which represents it, as can be seen in the following table, adapted from Hengeveld (1996: 120): Table 1.1: Units of the hierarchical structure of the clause (Hengeveld 1996) LAYER Clause Illocutionary Frame Proposition Predication Predicate Term VARIABLE E1 F1 X1 e1 f1 x1 DESIGNATION Speech Act Illocution Propositional Content State of Affairs Relation or Property Individual

Some of the layers which constitute this hierarchical structure correspond to the types of entity recognised by Lyons (1977). Thus, it can be asserted that a term designates a first order entity,6 a predication a second order entity and a proposition a third order entity in the classification proposed by Lyons, while the distinction of zero order (predicate) and fourth order (speech act) entities constitute an innovation of FG. Within this layered model of the clause, an individual is a first order entity which can be situated in space and evaluated in terms of its existence. A State of Affairs is a second order entity which can be situated in space and time and evaluated in terms of its reality. A propositional content is a third order entity which cannot be situated either in space or time, but can be evaluated in terms of its truth. A speech act is a fourth order entity which situates itself in space and time and which can be evaluated in terms of its felicity. The two lower layers in this model, predicate and term, were analysed previously when discussing the lexical component of FG, the fund, to which these two elements belong. When the appropriate term structures have been inserted in the argument slots of the predicate frame, the nuclear predication is obtained. This is represented as follows (Dik 1997a: 291): (6) nuclear predication = [pred [type] (args)]

In the context of clause structure, term is used in a restricted sense to refer to the prototypical use of this linguistic unit (the designation of a first order entity). As has already been mentioned, in Dik (1997a/b) term is also used, in a wider sense, to refer to every linguistic unit that can function as argument or satellite.

10

Mara Jess Prez Quintero

A nuclear predication designates a State of Affairs (hereafter SoA), an expression taken by Dik to mean the conception of something which can be the case in some world (1997a: 105). Within FG a classification of SoAs is offered which is based on the interaction of various semantic parameters (dynamism / control / telicity). Thus the following classification of SoAs is established (Dik 1997a: 114): Table 1.2: Typology of States of Affairs (Dik 1997a) General term [dyn] Situation [+dyn] Event [telic] Event [+telic] Event [+control] Position Action Activity Accomplishment [control] State Process Dynamism Change

The type of SoA designated by a nuclear predication correlates to a great extent with the type of semantic function assigned to the first argument within the predicate frame on which the predication is based. Once a nuclear predication has been formed, each layer can be modified by grammatical elements, operators, and lexical elements, satellites, which provide additional information. Operators and satellites may realise the same functions. The difference between them is that operators are expressed by grammatical means and satellites by lexical means. Five different types of operator can be distinguished: term operators ( ), predicate operators ( 1), predication operators ( 2), proposition operators ( 3) and illocutionary operators ( 4). Similarly, there are five types of satellite: predicate satellites ( 1), predication satellites ( 2), proposition satellites ( 3), illocutionary satellites ( 4) and clause satellites ( 5). The functions of the different types of operators and satellites (excluding term operators, which function within the fund) are described below. 1.2.2.2.1 Predicate operators and satellites ( 1 and 1) First layer operators and satellites specify additional properties of the internal structure of the SoA designated by the nuclear predication. Predicate operators, called Qualifying operators, constitute the grammatical elements which specify additional features of the nature or quality of the SoA (Dik 1997a: 219). The additional properties related to the internal organisation of the SoA belong to: (i) the domain of verbal aspect, such as the Perfective / Imperfective distinction, and (ii) the domain of modality inherent in the SoA. First layer satellites belong mainly to three groups: (i) those which designate additional participants in the SoA (Beneficiary, Company); (ii) those which specify the way in which the SoA is attained (Instrument, Manner, Speed, Quality) and (iii) those which express spatial orientations of the SoA (Direction, Source, Path).

Preliminaries

11

The result of the modification of the nuclear predication through first layer satellites and operators is the core predication, a structure which can be represented as follows (Dik 1997a: 64): (7) core predication = [
1 [nuclear

predication]

1]

A variable representing the SoA can be added to the core predication and this results in structures of the type: (8) ei: [core predication]7

1.2.2.2.2 Predication operators and satellites ( 2 and 2) The core predication can be modified by second layer operators and satellites which locate the SoA with respect to the temporal, spatial and cognitive setting, without affecting the properties of this SoA. Regarding predication operators, a distinction can be established between Quantifying and Localizing operators. Quantification of the predication is carried out through aspectual distinctions which concern the frequency of the SoA (Semelfactive, Iterative, Frequentative and Distributive Aspect). The operators that contribute to the location of the SoA are tense, perspectival aspect, objective modality and polarity.8 The satellites at this layer locate the SoA in the spatial dimension (Location), the temporal dimension (Time, Circumstance) and the cognitive dimension (Result, Purpose, Reason and Cause). The result of applying grammatical or lexical modifications to the core predication is the extended predication, which Dik (1997a: 291) represents in the following way: (9) extended predication = [
2

ei: [core predication] ( 2)]

It is at this layer of the underlying structure of the clause that the assignment of syntactic functions (Subject and Object) takes place, as can be seen in Figure 1.2. The extended predication can be modified by third layer satellites and operators, thus giving rise to a proposition, the next layer in the hierarchical structure, which is represented by the variable (X i), which designates a propositional content.

Given this representation, in the previous edition of Dik (1997a [1989]) it is noted that this predication can be called an embedded predication, since it is embedded under the influence of the SoA variable. 8 Although polarity is generally considered a second layer operator, Dik (1997b), inspired by Lyons (1977), analyses different types of negation which correspond to the different layers of the hierarchical structure of the clause (illocutionary, propositional, predicational and predicate negation).

12

Mara Jess Prez Quintero

1.2.2.2.3 Proposition operators and satellites ( 3 and 3) Third layer operators and satellites express the speakers evaluation or attitude towards the content expressed by the proposition. The modifiers at this layer, therefore, relate the propositional content to the subjective world of the speaker, expressing: (i) aspects of subjective modality, related to the speakers personal opinion or attitude, and (ii) specifications about the manner and circumstances in which the speaker obtained the information contained in the proposition, by means of evidential modality or attitudinal satellites. The result of applying third layer modifiers is the proposition, represented as follows (Dik 1997a: 291): (10) proposition = [
3

Xi: [extended predication] ( 3)]

The proposition can be expanded into the clause, which designates the speech act to which the linguistic expression refers and which is represented by the variable (E i).9 1.2.2.2.4 Illocutionary operators and satellites ( 4 and 4) The operators and satellites at this layer are related to the speakers communicative ability. In Diks model (1997a) illocutionary operators designate the basic illocutionary force (DECL(arative), INT(errogative), IMP(erative). . .) of the utterance.10 This illocutionary force, codified in some way in the linguistic expression [IllE], does not necessarily have to correspond to the intention of the speaker [IllS] or the interpretation of the addressee [Ill A].11 The basic illocutionary force can undergo a process of (pragmatic, lexical or grammatical) conversion and thus give way to another type of illocutionary force. Illocutionary satellites specify the way in which the speaker wishes the addressee to interpret the speech act and, therefore, include lexical elements which express Manner, Reason, Condition or Purpose, not regarding the SoA, but regarding the speech act.
9

Although it is the propositional content and not the SoA that can be modified by illocutionary operators and satellites, imperative operators are an exception, since they operate directly on the predication, without an intermediate propositional level (Dik 1997a: 53). 10 Following the model proposed by Hengeveld (1990), however, this illocutionary force would, as has been mentioned earlier, constitute an abstract illocutionary frame with the speaker, the addressee and the propositional content as its arguments. Therefore, the operators at this level would modify the basic illocutionary force of the utterance, whether the speaker wants to mitigate the force of the speech act (mitigating mode) or wants to reinforce it (reinforcing mode). 11 Regarding fourth layer operators, Moutaouakil proposes a distinction between sentence type operator (Tp) and illocutionary operator (Ill), since he believes that the type of utterance (Decl., Int.,...) and the illocutionary force (statement, question,...) are quite distinct features although they interplay in determining the formal (i.e. morphosyntactic and prosodic) properties of linguistic expressions as well as their interpretation (1996: 224).

Preliminaries

13

The result of modifying the basic structure of the clause by illocutionary operators and satellites is the extended structure of the clause which Dik (1997a: 292) represents in the following way: (11) clause = [
4

Ei: [proposition] ( 4)]12

Dik (1997a) points out that once the different layers of the hierarchical structure of the clause are modified through the insertion of operators and satellites, the upper layer of the organisation of the clause is reached, leading to the assignment of the third type of functions, pragmatic functions, as can be seen in Figure 1.2. The complete structure of the clause is represented in the following way (Dik 1997a: 68): [ 1 [pred [T] args] 1] 2] 3] 4] [.....nucleus.....] [...core predication...] [........extended predication........] [........................proposition........................] [.....................................clause.....................................] Figure 1.3: Hierarchical structure of the clause (Dik 1997a) However, Hengeveld (1990) recognises a fifth type of satellite which he calls clause satellites ( 5) which are not included in the FG model given by Dik (1997a). The lexical modifiers at this layer specify aspects relative to the setting of the utterance. This type of satellite has been considered by Wanders (1993) in her analysis of adverbs in Spanish. This author proposes the addition of a layer which represents units larger than the clause, in order to be able to account for the scope of textual adverbs. She calls this new layer discourse episode and represents it with the variable (D). Wanders proposes to call clause satellites, represented as 5, rhetoric satellites, since this term reflects more clearly the function that these satellites realise. Figure 1.4 shows the representation of operators and satellites in the hierarchical model proposed by Wanders (1993): [
4 Ei:

3 Xi:

2 ei:

12

However, in Hengevelds model, this level is structured by an abstract illocutionary frame (ILL), made up of three arguments (a speaker, an addressee and a propositional content), which leads to a different representation (Hengeveld 1990: 6): Ei: [ 4 ILL: 4 (S) (A) (extended proposition)] (Ei)

14

Mara Jess Prez Quintero D1: [ (E 1: [ 4F1: ILL(F1): (S) (A) ( 3X1: [ . . . (E n)] (D1) ] (X1):
3(X1)) 4(F1)]

(E1):

5(E1))

( 2e1: [ 1f1: Predb (f1): (Wx1: Predn(x1))...(xn)

1(f1)]

(e1):

2(e 1))

Figure 1.4: Operators and satellites in the hierarchical structure Recently, there has been great interest in developing FG from a sentence grammar into a grammar of discourse,13 although this area of study constitutes an important aspect still to be developed within the FG model. However, it is the organisation of the clause, and not that of discourse, which is the part of FG theory most relevant to the present study. Dik argues that, although the semantic relations expressed by satellites at clause level seem to have a projection towards the level of discourse, he does not go along with the idea that clausal, propositional and predicational satellites should not be treated as subordinate or embedded at the clause level (1997b: 432). Therefore, it is the hierarchical structure of the clause, consisting of different levels of increasing complexity, which is adopted here as the model of analysis for adverbial subordinate clauses. 1.2.2.3 Expression rules According to the FG model, once the complete representation of the underlying structure of the clause is obtained, the expression rules, which translate underlying representations into actual linguistic expressions, are applied. This final component of FG takes care of the following characteristics of linguistic expression: (i) The form of its constituents In FG the underlying structure of the clause only contains lexical elements. All the grammatical elements of a linguistic expression (e.g. morphological affixes, adpositions and grammatical particles) are provided by the expression rules which are language specific. Expression rules are the result of the application of (semantic, syntactic and pragmatic) functions and of operators, both specified in the underlying structure and called morphosyntactic operators. The rules of formal expression have the following structure (Dik 1997a: 69): (12) operator(s) [input form] = output form

13

One of the first contributions in this direction was Hengevelds (1997) work, who proposed the addition of a higher level, the rhetorical level, to the hierarchical structure. For other works relating to discourse and pragmatic aspects within the framework of FG, see Gmez Solico (1996), Dik (1997b), Connolly et al. (1997) and Hannay and Bolkestein (1998).

Preliminaries

15

(ii) The order of its constituents The order in which the different elements in the underlying structure of the clause are presented does not reflect the order in which they appear in the linguistic expression, but they are intended to be a reflection of the semantic relations which hold between them. It is through the application of one type of expression rule, called placement rules, that the different constituents are linearly ordered in the linguistic expression. (iii) Prosodic contour Finally, prosodic features characteristic of each language, such as tone, accent and intonation, must also be assigned to the linguistic expression derived from the underlying structure. 1.3 Adverbial subordination in Functional Grammar

1.3.1 Concept When dealing with complex constructions in FG, Dik (1997b) establishes a distinction between coordination and embedding. The term coordination includes the relations that exist between elements that are equivalent from a functional point of view, whereas the term embedding covers the relations in which one element realises a function within another element. Other linguists ascribed to the FG model prefer to use the term subordination instead of embeddedness. Thus, in FG the traditional dichotomy coordination / subordination is maintained,14 the main difference lying in the functional perspective adopted for the analysis of complex constructions. In order to present a functional classification of subordinate clauses, Hengeveld (1995) takes as a point of departure the classification of the parts of speech, summarised in the following table:15 Table 1.3: Classification of the parts of speech (Hengeveld 1995) Noun Phrase Verb Phrase
14

Head Noun Verb

Restrictor Adjective Adverb

Many authors have tried to offer alternative solutions to the problem that the delimitation of the concept of subordination presents. These solutions can be essentially divided into three groups: (i) those who propose a tripartite classification (Van Valin 1984; 1993; Foley and Van Valin 1984; Van Valin and LaPolla 1997); (ii) those who consider that sentence relations have to be analysed, not as a dichotomy, but as a continuum (Givn 1990; Lehmann 1988) and (iii) those who suggest that subordination should not be considered a onedimensional but a multi-dimensional phenomenon (Haiman and Thompson 1984). 15 Translated from the Spanish.

16

Mara Jess Prez Quintero

Following this classification, the distribution of subordinate clauses can be established as in Table 1.4: Table 1.4: Distribution of subordinate clauses (Hengeveld 1995) Noun Phrase Verb Phrase Head Restrictor Head Restrictor Complement clause Relative clause Predicate clause Adverbial clause

Adverbial clauses are comparable to adverbs, since both function as restrictors of a verb phrase. Therefore, it is possible to establish a semantic classification of adverbial clauses, parallel to that proposed for adverbs or satellites, paying attention to the layer of the hierarchical structure which they modify. The layered structure of the clause has been used as a descriptive framework for carrying out studies on specific types of subordinate clauses, such as complement clauses (Dik and Hengeveld 1991) and adverbial clauses (Hengeveld 1993; 1996; 1998; Hengeveld and Wanders 1997).16 In his study of the internal structure of adverbial clauses, Hengeveld (1996: 121) puts forward the following classification of subordinate clauses: Table 1.5: Classification of subordinate clauses (Hengeveld 1996) Superordinate Subordinate Open Closed Main clause Relative clause Predicate clause Complement clause Adverbial clause

Governing Governed

Obligatory Optional

Hengeveld argues that the study of adverbial clauses is in many respects parallel to that of complement clauses since they have one characteristic in common, namely, that both are governed clauses, in the sense that their underlying structure is determined by elements belonging to the main clause. Using this classification as a basis, Hengeveld (1998), in his typological study of adverbial clauses, defines a subordinate clause as one whose existence depends on another, inasmuch as it satisfies the requirements of the predicate frame of the matrix predicate, and an adverbial clause as one which can be
16

Other studies of subordinate clauses within the framework of FG can be mentioned such as those of Zimmermann (1985), Bolkestein (1986), Rijksbaron (1986), Wakker (1987; 1992; 1996), Harder (1989; 1996), Dik (1990), Vester (1990), Cuvalay (1996), Genee (1998). Although these are more specific in the sense that they concentrate on the analysis of a single language or a specific type of construction.

Preliminaries

17

removed without affecting the grammaticality of the main clause, since, as is the case with the category of adverbs, it functions as a satellite. Apart from Hengevelds proposal, adverbial subordination has not received much attention in the framework of FG. In Dik (1997b), a work dedicated to the study of derived and complex constructions, it is pointed out that adverbial clauses are subject to the same analysis as other types of subordination, such as relative clauses17 and complement clauses, since they are considered to be embedded constructions or complex terms.18 Regarding this, Dik (1997b: 95) claims: Adverbial clauses will not be discussed in the present context. There is no reason to assume, however, that they cannot be dealt with in terms of the parameters that will be presented in this and the next two chapters in the description of complex terms occupying argument positions. The clauses presented as complex terms occupying argument positions which Dik analyses in detail are complement clauses. This analysis takes into consideration three important aspects of subordinate clauses: semantic, functional and formal aspects. From a semantic point of view, Dik classifies embedded constructions according to the layer of the hierarchical structure of the clause (clause, proposition, predication) which they designate, while from a functional point of view, he considers the function (semantic, syntactic and pragmatic) fulfilled by these clauses. Finally, from a formal perspective Dik analyses the general features which characterise the expression of embedded constructions. These formal features include: the position of the embedded clause in the matrix domain, the presence of subordination markers, the internal order of the constituents of the embedded construction, the presence of finite or non-finite verbal forms and verbal mood. In this book, the study of adverbial subordination is carried out using the same model of analysis which was applied by Dik to complement clauses. However, this study concentrates on the analysis of formal aspects (Chapter 2) and semantic aspects (Chapter 3), given that from a functional point of view, adverbial clauses do not offer the same variability as complement clauses. Firstly, adverbial clauses cannot be analysed from the point of view of their syntactic functions, since it is not possible to assign them any of the functions (Subject/Object) recognised within FG. From the point of view of their semantic functions, adverbial clauses function as satellites, optional elements to which a function derived from their semantic value is assigned. Finally, regarding the assignation of pragmatic functions, Dik (1997b: 124) points out:

17

Relative clauses are analysed as embedded constructions, since they function as restrictors within a term. 18 Term is understood in its widest sense, as used by Dik (1997a/b).

18

Mara Jess Prez Quintero When the embedded construction functions as a satellite, it is even the rule rather than the exception that the satellite present the salient, focal, foregrounded information. In general, satellites have a certain degree of intrinsic focality because, if the satellite information were not important, there would be no point in adding it at all.

However, the pragmatic functions associated with adverbial clauses are not analysed here, since such an analysis should contemplate the domain of discourse and, therefore, is beyond the scope of this work. Summarising, the term subordination is used here, in opposition to the term coordination, to refer to a clause which is in a hypotactic relation to another clause, and which realises a function within it (whether it be subject, object, complement or adverbial). And, according to the functional model adopted, a clause is considered adverbial if it fulfils the function of satellite within the main clause and therefore can be omitted without affecting the grammaticality of the main clause. 1.3.2 Delimitation of the object of study Taking into account the definition of adverbial clause given in the previous section, a number of constructions, often considered adverbial, have been excluded. 1.3.2.1 Embedded clauses The first type of clause that has been excluded consists of those clauses which fulfil a function within a non-verbal phrase, whether nominal or adjectival. Even though the phrase in which they are embedded realises a semantic function similar to that of an adverbial clause, these clauses do not themselves belong to the domain of the matrix clause and, therefore, they do not have the status of satellites. (13) (14) The time we first met he hardly spoke to me at all (Halliday 1994: 247) (function of post-modifier in a noun phrase) Angry at being suspected, he could hardly speak (function of postmodifier in an adjective phrase)

However, it is not always easy to determine whether a clause is post-modifying the head of a noun phrase which functions as an adverbial, or whether it is a construction with a nominal conjunction in structures which have been grammaticalised. Quirk et al. (1991) consider that structures such as the moment (that), every time (that), during the period when, until such time as, the fact that . . . are examples of free syntactic constructions and not complex subordinate conjunctions. However, they do not explain the difference between such constructions and others with a similar structure which they do include in their classification of subordinate conjunctions (e.g. in case, in the event that, on condition (that). . . ). Halliday (1994: 238) includes both types of structure in his

Preliminaries

19

classification of complex conjunctions, stating that they are structures which have been formed from prepositional phrases which appeared with an embedded clause, but what they actually do is introduce hypotactic clauses like any other conjunction. The examples he gives are in case, in the event that, to the extent that and the combination of the definite article with various nouns which indicate time and manner, such as the day, the moment, the way. However, when dealing with cases of embedded clauses with adverbial meaning (embedded enhancing), Halliday points out that these are structures which are formed by a noun postmodified by an embedded clause and that the relation between the two members of this construction has a circumstantial meaning of time, place, manner, cause or condition. He believes that there are two different types of embedded structures: (i) those in which the adverbial meaning is expressed by a nominal head (e.g. the time when/that, the place where, reason why/that . . . ) and (ii) those in which it is the embedded clause that carries the circumstantial meaning (e.g. the house where/in which she lived). However, Halliday does not explain the difference that exists between these embedded constructions, which are not examples of adverbial subordination, and the complex prepositions consisting of a nominal head which express time and manner, which he mentions when dealing with the different types of adverbial conjunction. Given the aim of this book, it is fundamental to establish a clear distinction between nominal conjunctions, constructions which have suffered a process of grammaticalisation and which, therefore, introduce subordinate clauses, and the structures formed by a noun phrase whose head is post-modified by an embedded clause and which, therefore, are excluded. As a valid criterion for establishing this differentiation, it should be established whether the nominal part of a complex conjunction has lost its nominal features and is thus part of a fixed construction which functions as a subordinating conjunction. This criterion is adopted by Huddleston (1985), when establishing a distinction between complex prepositions and structures of the type preposition + noun + preposition. He points out: we may think of complex prepositions as arising historically through the lexicalisation the fusion into a single lexical item of the first words of some productive construction (1985: 342). The need to establish a distinction between simple and complex prepositions is also found in Quirk et al. (1991: 671): In the strictest definition, a complex preposition is a sequence that is indivisible both in terms of syntax and in terms of meaning. However, there is no absolute distinction between complex prepositions and constructions which can be varied, abbreviated, and extended according to the nominal rules of syntax. Rather, there is a scale of cohesiveness running from a sequence which behaves in every way like a simple preposition, e.g.: in spite of (the weather), to one which behaves in every way like a set of grammatically separate units, e.g.: on the shelf of (the door).

20

Mara Jess Prez Quintero

The tests which Huddleston sets out in order to identify complex prepositions are applied below in order to distinguish nominal conjunctions. By way of illustration, Huddleston uses the constructions by dint of hard work (complex preposition + noun phrase) and after years of hard work (preposition + noun phrase + preposition + noun phrase): (i) A complex preposition cannot realise the syntactic functions typically associated with a noun phrase, such as subject or complement. Thus, dint of hard work cannot realise the function of subject or complement in a clause, but years of hard work (e.g. Years of hard work had taken their toll (subject); He had wasted years of hard work (complement)) can fulfil these functions. Similarly, if in case (complex conjunction) is compared with the time (that) (noun phrase), it can be seen that the former structure cannot function as subject but the latter can (e.g. The time to be ready is four oclock). Complex prepositions do not allow the same variety of determiners as the head of a noun phrase does. The indefinite construction years of hard work contrasts with the definite construction the years of hard work, allowing also the use of other kinds of determiner such as these, a few, several . . . However, this variation is not possible with dint of hard work. Huddleston points out that the possibility of variation has to be determined bearing in mind that the original meaning must be maintained. Thus, for example, the conjunction in case does not allow a great variety of modifiers while preserving its conditional value. It can be quantified by most of, in which case a nominal construction is obtained which has a different meaning (e.g. in most of the cases temporal sense). The head of the nominal construction the time can be quantified as in most of the time, while at the same time allowing a greater variety of modifiers, such as every time (that), such time as, the very first time. In all these cases the basic meaning of the construction is maintained. The nominal part of a complex preposition does not allow variation in number. While the plural form years of hard work can be substituted by a singular a year of hard work, the singular form of dint of hard work does not have a plural alternative. It can also be seen that in the examples of complex conjunctions in case cannot be substituted by in cases, while the noun phrase the day has alternative plural expressions, such as since the days (when).

(ii)

(iii)

To Huddlestons criteria, others can be added from among those presented by Quirk et al. (1991: 671-2), who use the complex preposition in spite of (the weather) and the nominal construction on the shelf of (the door) as examples: (iv) The preposition which introduces a nominal construction allows variation. Thus, for example, the change of the preposition on in the structure on the shelf of for under (e.g. under the shelf by (the door)) is possible, but it is not possible in the case of the complex preposition (e.g. in spite of cannot

Preliminaries

21

(v)

(vi)

become *for spite of). In the case of conjunctions, it can be observed that the noun phrases which form part of grammaticalised structures do not allow variation regarding the preposition which introduces them, since they are fixed expressions (e.g. in case, on condition, but not *on case, *in condition). However, non-grammaticalised nominal expressions show a greater diversity of introductory elements (e.g. every time, until such time, by the time, since the last time, before the time, throughout the time (that), from the very first time). In the case of a combination of single prepositions, the prepositional complement and the second prepositional phrase can be substituted by a pronoun (e.g. on the surface of the table on its surface), but this is not a possibility in the case of a complex preposition (e.g. in spite of the result *in its spite). Similarly, Huddleston (1985) points out the possibility that a phrase introduced by the preposition of presents an alternative possessive construction (e.g. after the arrival of the prime minister / after the prime ministers arrival). However, this criterion will not be taken into account in the present work because it is considered irrelevant to the case of nominal conjunctions, and also of little use, as Huddleston and Quirk et al. point out, when dealing with some examples of complex prepositions which show the following variation (e.g. on behalf of my father / on my fathers behalf). The noun which forms part of a nominal construction can be pre-modified by adjectives (e.g. on the low shelf by (the door)), but this is not the case with one which forms part of a grammaticalised structure (e.g. *in evident spite of). In the same way, a complex conjunction does not allow the incorporation of adjectives which function as modifiers of the nominal term (e.g. *in good case), while the presence of an adjective which modifies the head is frequently part of a noun phrase (e.g. the last time).

1.3.2.2 Complement clauses Complement clauses are examples of embedding and, therefore, do not fit the definition of adverbial subordinate clauses which is proposed here. (15) I dont believe that Mary is pregnant

Regarding those cases of complement clauses in which a verbal predicate presents an argument realised by a clause, a similar problem to the one mentioned in the previous section might be noted that is, those constructions formed by a past or present participle which no longer functions as a verbal predicate, but that has acquired the status of a conjunction (e.g. assuming (that), given (that), provided (that), providing (that), supposing (that)) or a preposition (e.g. regarding, concerning). Quirk et al. state that those participles which have been grammaticalised as conjunctions have lost their verbal features, as can be seen by the fact that although they do not contain an explicit subject, they do not require identification with the subject of the main clause. Thus, for example, in Covered with confusion, they apologized abjectly, the participle ending in -ed is a verbal

22

Mara Jess Prez Quintero

form which requires identification of the subject with that of the main clause, while in Provided that the film entertains, few people care about its merits, identification of a subject is not necessary. Similarly, Knig and Kortmann (1991) point out some features which indicate that a verb has been re-analysed as a preposition: change in the word order, change of grammatical relations and phonological and morphological criteria. These features, as well as some additional ones, are analysed by Kortmann and Knig (1992: 686), who state: Many of the changes leading to a recategorization of verbs as prepositions can be seen as a loss of certain properties: a loss of semantic, phonological, and morphological substance, a loss of the ability to inflect for case, number, and gender, a loss for agreement with a subject, a loss of the ability to be marked for tense and aspect. Nevertheless, Kortmann (1991) points out that not all participles which are used as conjunctions show the same degree of lexicalisation, and mentions two parameters which prove this. Firstly, in some cases the identification of an implicit subject is established with an indefinite pronoun or with the speaker (e.g. . . . as if I couldnt figure out for myself that things had better be just so, considering whos coming Kortmann 1991: 51), and secondly, the participle can appear in contexts in which it still functions as a verbal form (e.g. The new airship . . . could keep station above the fleet wherever the US chose to go, providing early warning of aircraft or missile attack Kortmann 1991: 52). Kortmann and Knig (1992: 683) comment: Deverbal prepositions are not only marginal members in their lexical class, they are also an extremely heterogeneous group as a result of the fact that the various changes discussed in section 1 have not affected each individual item in the same way and to the same degree. This characteristic of subordinating particles derived from a participle makes it difficult to distinguish between a complement clause, whose main verb is in a non-finite form, and an adverbial clause, introduced by a non-finite verbal form which has been re-analysed as a preposition. 1.3.2.3 Verbless clauses Of the three types of structure which Quirk et al. (1991) point out as possible realisations of a subordinate clause, those constructions which belong to the third group verbless clauses are not included, since the main aim of this work is to study the systematic relations between the semantic type of adverbial subordinate clause and the way in which these clauses are expressed in English, paying special attention to the verbal forms contained in these constructions. Therefore, constructions of the type Although always helpful, he was not much liked, in which the subordinate clause lacks a verbal form, will be excluded.

Preliminaries

23

1.3.2.4 Copy-cleft sentences Talmy (1978a) analyses a series of constructions which have in common the fact that, from a semantic point of view, they are equivalent to subordinate clauses. He calls these copy-cleft sentences. Talmy argues that the complex clause Mays provided some excitement for the viewers by batting in three runs can have the following possibilities as alternative constructions: (16) (17) (18) (19) (20) Mays batted in three runs; he provided some excitement for the viewers thereby (without any type of linking element) Mays batted in three runs, and provided some excitement for the viewers thereby (with a coordinating conjunction) Mays batted in three runs, providing some excitement for the viewers thereby (non-finite construction -ing form) Mays batted in three runs, to provide some excitement for the viewers thereby (non-finite construction infinitive) Mays batted in three runs, whereby he provided some excitement to the viewers thereby (relative construction)

Of these five types, only the non-finite constructions (18) and (19) are considered adverbial subordinate clauses. Examples (16) and (17) are examples of paratactic constructions, juxtaposition and coordination, respectively, which do not form part of the set of structures under study. Example (20) represents an example of a relative clause in which the presence of an anaphoric element can be seen. Relative constructions constitute a type of subordinate clause different from adverbial clauses. 1.3.2.5 Independent relative clauses Geis (1970), when presenting types of constructions which he considers to be adverbial clauses, mentions one type which he calls independent relative clause, to which clauses such as John lives where Harry said he did or I found him where he said he would be, belong. These types of structures have been excluded because they are examples of relative clauses which form part of a group of subordinate clauses that are not adverbial. Moreover, in the first example, the temporal expression introduced by the relative pronoun does not function as a satellite, an optional element added to the predication, but as an argument, an obligatory constituent required by the predicate. Other excluded constructions are those which Quirk et al. (1991) call sentential relative clauses, that is, clauses which do not have a noun phrase as an antecedent, but refer to the predicate (e.g. They say he plays truant, which he doesnt), to the predication (e.g. He walks for an hour each morning, which would bore me), to a complete clause (e.g. Things then improved, which surprises me) or even to a series of clauses (e.g. Colin married my sister and I married his brother, which makes Colin and me double in-laws).

24

Mara Jess Prez Quintero

1.3.2.6 Clauses of Comparison and Comparative clauses A distinction should be established, as Quirk et al. (1991) propose, between adverbial clauses of Comparison (e.g. He bent down as if tightening his shoe laces) and Comparative clauses, in which a relation of comparison between a proposition from the main clause and one from the subordinate clause is established, (e.g. Marilyn was too polite to say anything about my clothes). Clauses belonging to this latter type are also excluded, since they are examples of clauses embedded in an adjectival group. 1.3.2.7 Clauses introduced by two conjunctions It can sometimes happen that a clause is introduced by two conjunctions. These clauses are classified only according to the meaning of the second conjunction, since in these examples the value of the first conjunction has been cancelled by that of the second. (21) Whether Handel planned it as he began the movement or whether it occurred to him as when improvising, this way of integrating the movement was exactly right in this place, and sensible people may call it a symphonic way. (LOB G42 153)

There are also cases in which the meaning of the first conjunction is not lost, but there is only one verbal form that is introduced by the second conjunction. This type of clause is also classified according to the meaning expressed by the second conjunction. (22) (23) Joyce did, of course, starve; Proust did not, except when the waiters at the Ritz were inattentive. (LOB G41 32) That will wasnt made until after Id gone away! (LOB L22 163)

1.3.2.8 Grammaticalised constructions Haspelmath and Knig (1998), in their typological study of ConcessiveConditional clauses, mention the existence of a type of construction, slightly grammaticalised, which expresses this adverbial meaning. They claim that expressions such as let it rain, it may rain, let it be that it rains, it may be that it rains express the idea of even if it rains. (24) She may be the worlds leading Etruscologist, but I doubt that she knows what concessive conditionals looked like in Etruscan.

These constructions are not taken into account here, since there seems to be no direct relation between this type of expression and a concessive-conditional meaning. Only finite clauses introduced by a conjunction are analysed, with the exception of Conditional clauses expressed through the inversion of the auxiliary (e.g. Had I known that she was here I wouldnt have come), because in this case there is a systematic relation between these constructions and the meaning which they express.

Preliminaries 1.4 Functional Grammar and Corpus Linguistics

25

1.4.1 Corpus analysis in Functional Grammar In his analysis of linguistics as a science, Crystal (1971) mentions three basic conditions that any scientific study should fulfil: explicitness, systematicness and objectivity. Thus, this author points out: I think every linguist would grant explicitness and systematicness a fairly fundamental role in any linguistic approach, though their ways of defining and evaluating them may differ somewhat. But the status of empirical evidence has been called into question by a sufficiently large number of influential people to make it necessary to approach this topic with particular care. (Crystal 1971: 99) Therefore, although any scientific study ought to be characterised by its ability to offer empirical data and observable results, in the field of linguistics two different approaches are found the corpus-based approach and the intuition-based approach. The choice of one or the other approach is conditioned by the conception of the object of study. A linguistic study carried out within the framework of FG, a theory whose ultimate goal is to be integrated into a general theory of verbal interaction, cannot disregard the analysis of corpora. In this respect, Mackenzie (1992: 10) posits the following questions: Can FG increase its compatibility with a wider theory of interaction without including in its formalisms the pragmatic principles underlying speakers actual choices of language structure? How can a bridge be built between the process-oriented claims of discourse analysis and the product-oriented contentions of grammar? However, although in theory the analysis of a corpus is considered relevant, in practice many studies carried out within FG have generally been based on intuition or on the analysis of data taken from grammars, dictionaries, etc., and not from corpora. It is only in the field of classical languages that the use of corpora has played an important role since the origin of FG. The general practice has been based on what Givn (1995: 20) considers to be the most common functional approach, called The Pull-em-out-of-the-text functionalist methodology, which he describes in the following way: a. b. c. d. Make a hypothesis that grammatical Form A has the communicative Function X. Look for some real text (communication) Identify (one, some, or many) instances in the text where Form A is paired with Function X. Declare your hypothesis proven.

26

Mara Jess Prez Quintero

Givn (1995: 20) warns that this method of analysis raises a series of important questions: First, how many instances of Form A in the text were not paired with Function X, but rather with Functions Y, Z, Q? Second, how many instances of Function X in the text were not paired with Form A, but rather with Forms B, C, D? Third, given the percent of Form A that indeed correlates with Function X, is it statistically significant, in the view of (i) the size of the total population; (ii) the size of the sample; and (iii) the amount of variation within the sample? Without answering these questions, we perpetuate the bad habit of testing hypothesis by attempting to verify them. Whereas what we should be doing is attempting and hopefully failing to falsify them. Within the framework of FG this practice is inadequate if the standards of adequacy proposed within the model are taken into account: (i) FG aims to be a pragmatically adequate grammar, that is, a grammar taking account of the interaction between speaker and addressee in verbal communication. In order to achieve this aim FG must focus on the analysis of authentic uses of language and not on isolated data extracted from different sources. Moreover, it is important to point out the importance of using oral language corpora, since these probably constitute the most natural form of verbal interaction. In this respect, Butler (1999) questions the extent to which FG reflects what happens in real language. FG aims to be a psychologically adequate theory, that is, it aims to be compatible with the processes of (de)codification which take place in the human mind in the course of communication. Butler (1999) also questions the extent to which FG reflects the choice of linguistic elements in processing language. It is only through experimenting, observation and the exhaustive analysis of the authentic use of language that conclusions about the cognitive processes that are involved in communication can be reached. FG has amongst its aims that of achieving typological adequacy. In order to become a theory which is capable of representing different languages and explaining the differences and similarities between them, it is essential that it starts from (qualitative and quantitative) descriptive studies of particular languages which can only be conveniently analysed using a corpus. Finally, a grammatical model which is based on the use of language, and not on the speakers abstract knowledge, ought to be a dynamic model capable of taking account of linguistic variation. Such variation only becomes apparent through the observation of authentic data. In this respect, FG should develop a model which is capable of giving proper account of this variation.

(ii)

(iii)

(iv)

Preliminaries

27

Today, within the framework of FG, a growing interest in the analysis of corpora can be observed, an interest that must undoubtedly be assessed positively: One effect of this development has been to provide a new manner of testing the empirical claims of FG; another has been to throw up novel questions and often bewildering challenges which are bound to become dominant issues in the FG of the nineties. (Mackenzie 1992: 10) This book contributes to enriching the FG model with the analysis of authentic uses of language. Thus it aims to make a contribution to the analysis of corpora suggested by corpus linguistics, without rejecting introspection, which is a necessary tool for interpreting empirical data. As Johansson (1991: 313) states when discussing the use of linguistic corpora: In spite of the great changes in the less than three decades since the first computer corpus, there is one way in which the role of the corpus in linguistic research has not changed. The corpus remains one of the linguists tools, to be used together with introspection and elicitation techniques. Wise linguists, like experienced craftsmen, sharpen their tools and recognize their appropriate uses. 1.4.2 Information about the corpus used The corpus used is the Lancaster Oslo/Bergen (LOB) Corpus.19 This corpus comes in different versions: horizontal/vertical and tagged/untagged. The present study was carried out using the horizontal tagged version. In this version the corpus comes in the form of a text in which each word is accompanied by a wordtag which has been assigned to it through the use of automatic programmes and manual edition tasks.20 A selection of adverbial clauses in their contexts was carried out using 25% of the total corpus. Butler (1985: 2) puts forward the advantages of using a sample of data in the following terms:

19

Unfortunately, at the moment the research for this book was done, it was impossible to use the British National Corpus (BNC), mainly for technical reasons. Nevertheless, this circumstance may be expected not to affect the findings. Firstly, because subordination is much more common in written than in spoken language. Secondly, given that we are dealing with written data, the fact that the data contained in the LOB corpus is older than that of the BNC is only slightly relevant. 20 For information on the sources and selection of the texts that conform the LOB corpus, see Johansson et al. (1978). For detailed information on the tagged version see Johansson et al. (1986).

28

Mara Jess Prez Quintero with a finite population21 which is not too large, we may be able to investigate the whole population. But if our population is potentially infinitive, or if it is finite but very large, we shall have to be content with samples drawn from the population concerned. The use of samples, even for the study of finite populations, cuts down the labour and cost involved in obtaining results, and minimises the errors which can easily be made during the processing of large amounts of statistical data.

In order to guarantee the representativity of a sample of data methods of probabilistic sampling are used. As Bisquerra Alzina (1987: 7) explains, these methods are based on the principle of equiprobability, which states that all individuals of a population have the same probability of becoming part of the sample. The probabilistic method used here is the one known as random selection. According to Bizquerra Alzina (1987), this method of sampling consists of: (i) (ii) Dividing the population into different strata. The LOB corpus (population) is divided into different categories (strata) which constitute the different types of text. Selecting a sample from each stratum. To select a sample within each textual category a simple random process has been used. It has not been necessary to assign a number to each text, since texts are already numbered within each category. Using the SPSS programme (Statistical Package for Social Sciences), texts have been selected randomly using these numbers. Deciding on the number of individuals in each stratum. In this study the number of texts to be included in each category is derived by proportional affixation, that is, taking into account the proportion of texts which make up each category.

(iii)

The results of selecting 25% of the texts for each textual category is given in Appendix I. After randomly selecting the sample of texts from the corpus, the adverbial clauses were identified using the Tact programme (Text Analysis Computing Tools, version 2.1.4, June 1995). The selection of adverbial clauses with an independent verbal form was carried out by searching the tags CS (Subordinating Conjunction) and WRB (Wh-adverb). On the other hand, adverbial clauses with a dependent verbal form were selected through: (i) The tags BEN, HVN and VBN, for past participle forms; (ii) the tags BE, DO, HV and VB, for infinitive forms; and (iii) the order .*ing, for -ing forms. The exclusion of the non-relevant constructions (such as complement clauses introduced by a conjunction, restrictive relative clauses realised by a past
21

The term population is used in a broad sense to refer to any collection of entities, of whatever kind, that is the object of investigation (Butler 1985: 1).

Preliminaries

29

participle or an -ing form and by finite verbal forms tagged BE, DO, HV and VB) was carried out manually.

Das könnte Ihnen auch gefallen