696 Functional Relations

Durie M (1988). ‘The so-called passive of Acehnese.’ Lg 64, Kuno S (1975). ‘Three perspectives in the functional ap-
104–113. proach to syntax.’ In Grossman R E, San L J & Vance T J
Fillmore C J (1968). ‘The case for case.’ In Bach E & Harms (eds.) Papers from the parasession on functionalism.
R (eds.) Universals in linguistic theory. New York: Holt, Chicago, IL: Chicago Linguistic Society.
Rinehart and Winston. Lambrecht K (1986). ‘Topic, focus and the grammar
Firbas J (1966). ‘Non-thematic subjects in contemporary of spoken French.’ Unpublished Ph.D. diss. Berkeley:
English.’ Travaux linguistiques de Prague 2, 239–256. University of California.
Foley W A & Van Valin R D Jr (1984). Functional syntax Lambrecht K (1987). ‘Sentence focus, information struc-
and universal grammar. Cambridge: Cambridge Univer- ture, and the thetic-categorial distinction.’ BLS 13,
sity Press. 366–382.
Foley W A & Van Valin R D Jr (1985). ‘Information pack- Lambrecht K (1993). Information structure and sentence
aging in the clause.’ In Shopen T (ed.) Language typology form. Cambridge: Cambridge University Press.
and syntactic description I: Clause structure. Cambridge: LaPolla R (1990). ‘Grammatical relations in Chinese.’
Cambridge University Press. Unpublished Ph.D. diss. Berkeley, CA: University of
Gazdar G, Klein E, Pullum G & Sag I (1985). Generalized California.
phrase structure grammar. Oxford: Basil Blackwell. Li C N & Thompson S A (1976). ‘Subject and topic: A new
Grice H P (1975). ‘Logic and conversation.’ In Cole P & typology of language.’ In Li C (ed.) Subject and Topic.
Morgan J (eds.) Speech acts (Syntax and semantics 3). New York: Academic Press.
New York: Academic Press. Perlmutter D M (1982). ‘Syntactic representation,
Gruber J (1965). ‘Studies in lexical relations.’ Unpublished syntactic levels, and the notion of subject.’ In Jacobson
Ph.D. diss., Cambridge, MA: MIT. P & Pullum G K (eds.) The nature of syntactic represen-
Halliday M K (1967). ‘Notes on transitivity and theme in tation. Dordrecht: Reidel.
English.’ JL 3, 37–81, 199–244. Prince E (1981a). ‘Topicalization, focus-movement and
Halliday M K (1985). An introduction to functional gram- Yiddish-movement: A pragmatic differentiation.’ BLS 7,
mar. London: Edward Arnold. 249–264.
Horn L (1986). ‘Presupposition, theme and variations.’ Prince E (1981b). ‘Toward a taxonomy of given-new infor-
CLS 22(2), Papers from the parasession on pragmatics mation.’ In Cole P (ed.) Radical pragmatics. New York:
and grammatical theory. Chicago, IL: Chicago Linguistic Academic Press.
Society. Van Valin R D Jr (1977). ‘Aspects of Lakhota syntax.’ Unpub-
Jackendoff R S (1972). Semantic interpretation in genera- lished Ph.D. diss., Berkeley, CA: University of California.
tive grammar. Cambridge, MA: MIT Press. Van Valin R D Jr (1981). ‘Grammatical relations in ergative
Jackendoff R S (1987). ‘The status of thematic relations in languages.’ Studies in Language 5, 361–394.
linguistic theory.’ LIn 18, 369–411. Van Valin R D Jr (1990). ‘Semantic parameters of split
Kempson R (1975). Presupposition and the delimitation of intransitivity.’ Lg 66, 221–260.
semantics. Cambridge: Cambridge University Press. Van Valin R D Jr (1993). ‘A synopsis of role and reference
Kuno S (1972a). ‘Functional sentence perspective: grammar.’ In Van Valin R D Jr (ed.) Advances in role and
A case study from Japanese and English.’ LIn 3, reference grammar. Amsterdam: Benjamins.
269–320. Walton C (1986). Sama verbal semantics: Classification,
Kuno S (1972b). ‘Pronominalization, reflexivization, and derivation and inflection. Manila: Linguistic Society of
direct discourse.’ LIn 3, 161–195. the Philippines.

Functionalist Theories of Language

C S Butler, University of Wales, Swansea, UK this one conceals a much more complex reality. A
ß 2006 Elsevier Ltd. All rights reserved. more realistic way of looking at things would be to
posit a linear sequence of approaches, with strongly
formalist theories such as Chomsky’s at one end and
Functionalism within the Gamut of
the more radically functional approaches at the other.
Linguistic Theories
Even this, though, is still something of an oversimpli-
Functionalist theories of language have most often fication: what we really need is a multidimensional
been contrasted with formalist theories, particularly model in which the whole space occupied by linguis-
those developed by Chomsky during the past half tic theories is populated by groups of theories that
century. This comparison is useful and will indeed form tighter or looser clusters according to their de-
form the starting point for the discussion in this arti- gree of similarity on a wide range of features. In what
cle. However, like many dichotomies in linguistics, follows, we examine a number of features that are
 Functionalist Theories of Language 697

characteristic of functionalist theories. Much more solving a problem or engaging in what has been called
detailed discussion of many of these points can be ‘phatic communion’ in order simply to oil the social
found in Butler (2003a: Chapter 1). wheels. For the functionalist, many such situations
A further complication that should be borne in would still be communicative, though what is com-
mind as we survey the range of functionalist ap- municated may be largely social or interpersonal in
proaches is the distinction between European and nature rather than purely representational. Further-
North American functionalism. Important theories more, even if some types of language use are non-
initially developed in Europe include Functional communicative, they are, the functionalist will claim,
Grammar (FG: Dik, 1997a, 1997b) and Systemic merely parasitic upon the central communicative use.
Functional Grammar (SFG: Halliday, 1994; Halliday If we are to study language as communication, then
and Matthiessen, 1999, 2004). In the United States, we will need to take into account the properties both
there are two rather distinct kinds of approach: on of human communicators and of the situations in
the one hand, Role and Reference Grammar (RRG: which linguistic communication occurs. Indeed, a
Van Valin, 2005; Van Valin and LaPolla, 1997) further important claim of functionalism is that lan-
belongs to the rather more formal subgroup of guage systems are not self-contained with respect to
functionalist theories; on the other hand, there is a such factors, and therefore autonomous from them,
set of approaches, more radically functionalist in na- but rather are shaped by them and so cannot be
ture, that have developed out of the ‘West Coast’ properly explained except by reference to them.
functionalism of scholars such as Givón, Haiman, Linguists who make this claim belong to the group
Bybee, Hopper, Thompson, DuBois, Noonan, and that Croft (1995) calls ‘integrative functionalists,’
Goldberg and that are often included, together with and they undoubtedly form the largest and most in-
Langacker’s cognitive grammar, under the label of fluential group of functional theorists. The main lan-
‘usage-based’ models. It is important to be aware guage-external motivating factors are of two kinds:
that often, in the writing of American functionalists, the biological endowment of human beings, including
the labels ‘functionalist’ and ‘functionalism’ are used cognition and the functioning of language processing
primarily in relation to this last group of models, mechanisms, and the sociocultural contexts in which
often with little reference to European ideas. communication is deeply embedded. We might also
expect that a functionalist approach would pay seri-
ous attention to the interaction between these factors
The Basic Tenets of Functionalism
and the ways in which languages change over time,
The three most basic principles of functionalist although in practice this varies considerably from one
approaches are that model to another.
. they regard communication as the primary func- The question of motivation for linguistic systems is,
of course, not a simple one. Much of the formalist
tion of language, which shapes the forms languages
criticism of functionalist positions has assumed a
take
. they attribute great importance to external rather naı̈ve view of functional motivation, in which
some linguistic phenomenon is explicable in terms of a
(cognitive and sociocultural) factors in explaining
single factor. Functionalists, however, have never seen
linguistic phenomena
. they reject the claim that syntax is autonomous things this way, but rather accept that there may be
competing motivations, pulling in different directions
from semantics and pragmatics.
and often leading to compromise solutions.
The most basic tenet of all, from which all others It is important to note that formalists are not nec-
derive to some degree, is that the primary purpose essarily totally opposed to the claim that language is
of language is human communication, and that this functionally motivated, although this is something
fact is crucial in explaining why languages take the that is not often discussed in the formalist literature.
form they do. This view contrasts somewhat starkly Newmeyer (2003a), for example, claims that even
with that of Chomsky, for whom language is essen- hard-line formalists would accept this claim, and
tially a vehicle for expressing thought, with inter- that Chomsky has never denied it. The question, for
human communication being just one of the uses to Newmeyer, is how much functional motivation there
which it can be put, and not to be prioritized over is, where it is located in the system, and to what
other possible functions. Chomsky (1980: 229–230) extent it should form part of the research program.
defends his position by reference to what he regards In particular, Newmeyer claims that function-
as noncommunicative uses of language, such as re- alists and practitioners of cognitive linguistics have
698 Functionalist Theories of Language

posited a tighter iconic form–function fit than is jus- . Recognition and modeling of the flexibility of
tifiable and is skeptical of the concept of competing language in use
motivations, on the grounds that anything can be . Attention to discourse as well as to the sentence
explained away by some combination of such antag- . A concern with typological variation in language
onistic forces (Newmeyer, 1998: 137–153, 2003b). . An approach to language acquisition in which the
The solution, of course, is to demonstrate such com- child constructs his or her language from the infor-
peting motivations and their antagonistic effects mation available in the linguistic environment,
clearly so that functional ‘explanation’ does not be- given inherent general cognitive capacities and
come vacuous, a problem that is discussed in Butler learning abilities
(2003a: 19–25).
In the following sections, these points are explored in
The great majority of functionalists also believe
more detail.
that within the nonautonomous linguistic system as
a whole, that part of it which deals with the forms of
language, and which we may call morphosyntax, is Inclusive Rather Than Core Grammars
also not autonomous but is intimately bound up with A truly functional approach, aiming to understand
the kinds of meaning that it serves to express, divided and account for language as communication, cannot
by many linguists into the semantic and the pragmat- restrict its scope to some kind of core grammar, as
ic. In other words, functionalists claim that the mor- with Chomskyan theories, but must, in principle, take
phosyntax of languages is motivated by the meanings on the full complexity of languages and our use of
that it conveys, although we shall see that the various them, provided, of course, that the phenomena under
theories differ with regard to the strength with which study are systematic.
they hold this position. Indeed, there are some schol-
ars, such as Prince and Kuno, whom Croft (1995) calls
‘autonomist functionalists’ precisely because they do The Use of Authentic Textual Data
accept the autonomy hypothesis, but it can be argued If, as functionalists claim, they are interested in study-
that these approaches are peripheral to the mainstream ing language as communication, then it is logical
of functionalism (Butler, 2003a: Chapter 1). Most func- to assume that one of the main sources of data for
tionalists would accept that there is some degree of their descriptions and theorizing will be samples of
(at least apparent) arbitrariness in the morphosyntax actual communication, in their contexts of use. In
of any language, but many would again attempt other words, we would expect functionalists to make
to link this to the concept of competing motivations. extensive use of authentic language data in their work,
The functionalist position again contrasts with the particularly those large bodies of material that have
formalist one, which maintains that the morphosyntax been collected in the form of computer-readable
forms a system that can be described and explained corpora.
independently of meaning.
Summing up the basic characteristics of function-
Flexibility of Meaning and Structure
alist theories, we may say that the core of the func-
tionalist position is that language systems and their If we are truly to study language in its communicative
components are so inextricably linked with the social, role, we must come to terms with the fact that what
cognitive, and historical contexts of language use, and we say, and the way we say it, responds flexibly to
with the meanings that language is used to convey, the demands made by the communicative context.
that it is futile to attempt to describe and explain There are various aspects to this flexibility: the
them except through reference to such factors. context-dependent meaning of words; the fact, clear
from corpus linguistics, that a good deal of what we
say is made up of reusable ‘chunks’ of language rather
Further Features than generated or analyzed afresh every time we need
to express a particular concept (see Formulaic Lan-
There are a number of further features that, although
guage); the fuzzy nature of linguistic categories,
ultimately derivable from the basic tenets, show
which has been modeled through recourse to the spe-
considerable variability across functionalist theories:
cification of prototypes; the fact that structures in
. The attempt to account for the full range of linguis- actual communication are often much less rigid than
tic phenomena rather than merely a ‘core grammar’ the textbook grammars would allow; the emergence
. The use of authentic linguistic productions as data of grammatical categories from the requirements of
 Functionalist Theories of Language 699

discourse; and the ‘leaky’ nature of the grammars we cognitive factors and learning capacities. This is not
write to describe languages (Givón, 1995: 9). to say that functionalists totally reject the possibility
of innate characteristics: what is at issue is rather
A Discourse Grammar, Not Just a Sentence the nature of this endowment. For the functionalist, it
Grammar is more likely that what is innate will be a set of gen-
Communication normally involves not just single, eral cognitive principles and predispositions to learn
isolated sentences but also larger stretches of multi- rather than some concrete set of linguistic rules and
propositional discourse. It follows that a truly func- principles.
tional theory should reflect this fact by containing not
just a sentence grammar but also a model of discourse Important Functional Theories
structure and of how the two interact.
This section examines briefly the European and
Typological Orientation American functionalist approaches mentioned at the
beginning of this article, with particular reference to
Functionalists are interested not only in individual
the features of functionalism discussed previously.
languages but also in ‘language’ as a whole. Since cru-
cial motivating factors such as the human biological Functional Grammar
endowment and the overall requirements of commu-
nication are universal, we may expect that they will Functional Grammar was developed by the Dutch
be reflected in linguistic universals, although it is also linguist Simon Dik and his colleagues from the late
important to realize that because competition among 1970s onwards. Dik made a strong commitment to
motivations can be resolved in many ways, and be- the centrality, in the formulation of a grammar, of the
cause there are considerable differences in the socio- concept of communicative function:
cultural conditions under which languages are used, The primary aim of natural languages is the establish-
there are also pressures leading to diversity among ment of interhuman communication; other aims are ei-
languages. These concerns are manifested in the inter- ther secondary or derived. (Dik, 1986: 21)
est shown by most functionalists in linguistic typol- In the functional paradigm, . . . a language is in the
ogy. As pointed out by Croft (2003: 2), the label first place conceptualized as an instrument for social
functional–typological approach is often given to an interaction among human beings, used with the inten-
approach that seeks to motivate similarities and dif- tion of establishing communicative relationships. Within
ferences between languages in functional terms rather this paradigm one attempts to reveal the instrumentality
than in terms of the ‘formal’ explanations sought by of language with respect to what people do and achieve
with it in social interaction. (Dik, 1997a: 3)
Chomskyan linguists. Nevertheless, as Croft (2003:
5–6) also observes, the functionalist and formalist Dik (1986) discussed in detail the kinds of external
approaches share several fundamental characteristics, motivation that we need to appeal to in explaining
although with different emphases: both begin by ana- why languages are as they are, also bringing in the
lyzing language structures (functionalists in terms of concept of competing motivations.
function and Chomskyans in terms of formal proper- Dik (1997a: 8) also rejects the thesis that syntax is
ties), both are concerned with the central question of autonomous from semantics and pragmatics:
what constitutes a possible human language and posit
Semantics is regarded as instrumental with respect to
universal constraints, both abstract patterns from
pragmatics, and syntax as instrumental with respect
the data (the functionalist across languages and the to semantics. In this view there is no room for something
formalist within languages), and both explain univer- like an ‘autonomous’ syntax. On the contrary, to the
sals in terms of universal human abilities (the func- extent that a clear division can be made between syntax
tional typologist in terms of general cognitive and and semantics at all, syntax is there for people to be able
sociocultural abilities and the formalist in terms of to form complex expressions for conveying complex
innate principles specific to language). meanings, and such meanings are there for people to be
able to communicate in subtle and differentiated ways.
A Constructivist Account of Acquisition
He does, however, accept that grammar is a structural
Functionalist theories, with their communication- system, governed by a set of rules and principles,
in-context view of language, understandably tend to which should nevertheless be explained, wherever
favor a constructivist view of language acquisition, possible, in terms of the functioning of language as
according to which enough information is available a communicative device. For example, Dik (1986:
in the linguistic environment for the child to be able 27–28, 43) points out that expression types used to
to construct a grammar, with the help of more general express politeness, deference, and social distance are
700 Functionalist Theories of Language

typically longer and more complex than less polite which aim to model discourse as well as the sentence
counterparts, and that this reflects two types of iconic grammar.
principle: that of diminution vs. augmentation (im- From its inception, FG has always had a strongly
portant ¼ high, big; unimportant ¼ low, small) and typological orientation. Dik (1997a: 14) states of FG
the avoidance of directness (see also Haiman, 1983). that it should
The previous quotation suggests that pragmatics,
be typologically adequate, i.e., that it should be capable
as the driver of semantics, which in turn is the driving
of providing grammars for languages of any type, while
force behind the syntax, should play an important at the same time accounting in a systematic way for the
part in FG, and Dik does indeed set up a criterion of similarities and differences between these languages.
pragmatic adequacy for his grammar. He also makes
a commitment to a criterion of psychological adequa- This orientation is very clearly reflected in the work
cy, according to which the grammar should at least of Dik and other FG linguists, and it persists in recent
be compatible with what we know of the storage models such as FDG and IFG.
and processing of language. It has been argued, On language acquisition, Dik (1997a: 7) states the
however, that Dik’s work failed to live up to both of following:
these standards of adequacy. A number of new From a functional point of view, . . . it is certainly more
models have been put forward under the general um- attractive to study the acquisition of language as it devel-
brella of FG, one of whose aims is to increase the level ops in communicative interaction between the maturing
of pragmatic and (particularly) psychological ade- child and its environment, and to attribute to genetic
quacy. Particularly important among these new devel- factors only those underlying principles which cannot
opments are the Functional Procedural Grammar be explained as acquired in this interaction.
(FPG) of Nuyts, the Functional Discourse Grammar However, only a limited amount of work on acquisi-
(FDG) of Hengeveld, and the Incremental Functional tion has been done so far.
Grammar (IFG) proposed by Mackenzie. Mention
should also be made of the Functional Lexematic Role and Reference Grammar
Model of Mairal Usón and colleagues, which is
RRG makes a strong commitment to the study of
derived from FG and Coseriu’s theory of lexematics.
language as communication, as shown by the follow-
This model pays considerable attention to cognitive
ing quotation from an early work in which the initial
aspects of lexical structuring. Brief introductions to
basis of the theory was developed (Foley & Van Valin,
all these models can be found in Butler (2003a,
1984: 7):
2003b), together with evaluations of the extent to
which they, as well as the parent theory of FG, attain The theme underlying the various functional approaches
the standards of adequacy they aspire to. is the belief that language must be studied in relation to
FG and its variants are committed to providing its role in human communication. Language is thus
an account of the full range of phenomena found viewed as a system of human communication, rather
in the world’s languages rather than of some re- than as an infinite set of structural descriptions of
sentences.
stricted core grammar. In Dik’s work, most examples
used to illustrate the grammar were constructed, but The last sentence is clearly intended to contrast the
proponents of FG and its variants are beginning perspective taken in RRG with that of Chomskyan
to take more seriously the need to work with authen- linguistics. With this orientation goes a concern
tic productions, and the use of computer-assisted for the functional motivation of the grammar: ‘‘RRG
corpus analysis is increasing. FG has not been takes language to be a system of communicative
among the theories that have foregrounded the flexi- social action, and accordingly, analyzing the commu-
bility of language, although there is some work on nicative functions of grammatical structures plays
prototypes. a vital role in grammatical description and theory
Dik’s early model was purely a sentence grammar, from this perspective’’ (Van Valin, 1993: 2). In prac-
although he did present a programmatic model of tice, RRG prioritizes cognitive over sociocultural
discourse structure in his later work (Dik, 1997b: explanation, adopting what Van Valin and LaPolla
Chapter 18). However, there have always been (1997: 15) refer to as a ‘communication-and-
groups of classicists and anglicists in The Netherlands cognition perspective’ and leaving out of account, for
interested in the textual dimension of FG, and this now, the ways in which language is used in different
interest has recently increased, with a proliferation social situations (Van Valin and LaPolla, 1997: 3, 15).
of discourse models beginning in the 1990s. Particu- The autonomy of syntax is firmly rejected (Van
larly important in this connection are FDG and IFG, Valin, 1993: 2):
 Functionalist Theories of Language 701

Syntax is not autonomous. In terms of the abstract para- There have been a number of studies using RRG as a
digmatic and syntagmatic relations that define a struc- basis for accounting for observed patterns in the ac-
tural system, RRG is concerned not only with relations of quisition of morphosyntax in a variety of languages
co-occurrence and combination in strictly formal terms (for a summary, together with a discussion of the
but also with semantic and pragmatic co-occurrence and
relationship between acquisition studies and the
combinatory relations.
linguistic theory, see Butler, 2003b: 402–413).
In other words, RRG postulates that semantic, prag-
matic, and syntactic aspects of language, and their Systemic Functional Grammar
combinations, are all crucial in explaining the ob-
served properties of language. Syntax, however, is SFG, as developed by Halliday, has its roots in the
only relatively motivated by semantics, pragmatics, work of the British linguist J. R. Firth, with influence
and cognitive concerns; that is, syntactic phenomena also from Hjelmslev, the anthropologist Malinowski,
cannot just be reduced to matters of semantics and and the Prague School. Halliday makes it clear that
pragmatics, and there remains some arbitrariness in his aim is to give an account of linguistic communica-
the formal system. tion, or ‘language in use,’ and that languages are as
RRG aims, in principle, to provide an account of all they are because of the functions they have evolved to
the grammatical phenomena attested in any language. serve:
To date, it has not made extensive use of authentic Language has evolved to satisfy human needs; and the
data. Like FG, it does not prioritize the concept of way it is organized is functional with respect to these
flexibility in its account of languages. It does, howev- needs – it is not arbitrary. A functional grammar is essen-
er, make use of the concept of prototype, especially in tially a ‘natural’ grammar, in the sense that everything in
its use of hierarchies of argument types to predict it can be explained, ultimately, by reference to how
more and less typical choices in two areas: (1) the language is used. (Halliday, 1994: xiii)
‘macroroles’ of Actor and Undergoer in the clause, Both the general kinds of grammatical pattern that
have evolved in language, and the specific manifesta-
which are generalizations over more specific thematic
tions of each kind, bear a natural relation to the mean-
relations, and (2) syntactic relations, which RRG ings they have evolved to express. (Halliday, 1994: xviii)
conceptualizes not in terms of the usual functions of
Subject and Object but, rather, in terms of the notion Note that Halliday takes a very strong stand on func-
of the ‘privileged syntactic argument’ of a construc- tional motivation: in principle, everything in the
tion. Although RRG does offer a detailed account of grammar can be motivated through function – a po-
a number of discourse pragmatic areas, such as in- sition that some other functionalists would regard as
formation distribution (in terms of topic and focus) rather extreme. As might be expected from the ante-
and switch reference, it has not developed its own cedents of SFG in the work of Firth and Malinowski,
model of discourse structure and is still essentially a Halliday has always been concerned more with social
sentence grammar. and cultural motivations than with those related to
Like FG, RRG is very much concerned with typo- psychological/cognitive processes: indeed, the social
logical matters: ‘‘The theory is greatly concerned with dimension is more strongly developed in SFG than in
typological issues. In particular, it seeks to uncover any other functional theory. This is particularly ap-
those facets of clause structure which are found in all parent in the theory of social context developed by
human languages’’ (Van Valin, 1993: 4). Indeed, Van Halliday and colleagues in the 1960s and since devel-
Valin says that the theory arose out of the question oped not only by Halliday but also by Hasan, Martin,
of what linguistic theory might look like if, rather and others. SFG distinguishes between dialectal and
than being centered on English, it were based on diatypic (or ‘register’) variation in language, and
languages such as Dyirbal, Lakhota, or Tagalog. within the latter between the variables of field (con-
RRG takes a constructivist line on language acqui- cerned with the nature of the social action), tenor
sition (Van Valin, 1993: 2): (involving the relationships between interactants in
communication), and mode (the medium of commu-
The RRG approach to language acquisition . . . rejects nication and its role in the linguistic event). SFG has
the position that grammar is radically arbitrary and
also developed accounts of genres, seen as ways of
hence unlearnable, and maintains that it is relatively
motivated (in Saussure’s sense) semantically and prag- getting things done linguistically.
matically. Accordingly, there is sufficient information SFG totally rejects the concept of an autonomous
available to the child in the speech to which it is exposed syntax. The theory is firmly centered on meaning,
to enable it to construct a grammar, and therefore the with syntactic patterns being seen as one mechanism
kinds of autonomous linguistic structures posited by for realizing meanings, a means to an end. Indeed,
Chomsky are unnecessary. Halliday prefers to avoid the term ‘syntax’ altogether
702 Functionalist Theories of Language

in order to distance himself from the view that syn- parameters that shape language and grammar: cogni-
tax can be opposed to semantics and pragmatics. tion and communication, the brain and language
Instead, he uses the term ‘lexicogrammar,’ chosen to processing, social interaction and culture, change and
emphasize the claim that grammar and lexis are not variation, acquisition and evolution.’’ It is, however,
totally different kinds of patterning but are part of possible for competing motivations to lead to a situa-
the same level, different in degree rather than in tion in which grammar ‘‘assumes its own reality’’
kind. Both act, together with intonation, to give shape (Givón, 1995: 11). The central part played by seman-
to the meanings we wish to convey. Furthermore, tics, pragmatics, and discourse within Givón’s ap-
the division between lexicogrammar and semantics proach to language is abundantly clear in all his
is itself not a clear one: ‘‘There is no clear line between work. Furthermore, relationships between linguistic
semantics and grammar, and a functional grammar usage, synchronic variation, and language change
is one that is pushed in the direction of the semantics’’ are discussed in terms of the grammaticalization of
(Halliday, 1994: xix). SFG has not, historically, originally lexical elements.
given high priority to typological matters, and this Givón rejects Chomsky’s sanitization of the data
is reflected in the fact that the most comprehen- for linguistic theory building, through the prioritiza-
sive accounts of this theory (Halliday, 1994, 2004; tion of competence over performance. He thus takes
Halliday and Matthiessen, 1999) are centered very an inclusive stance on the range of phenomena for
firmly on English. However, currently there are signs linguistic study, and makes frequent use of attested
of greater interest in typological matters (Caffarel linguistic data, especially in his studies of the frequen-
et al., 2004). The approach to typology in SFG cy distribution of features in different languages and
differs sharply from that in FG and RRG in that the their varieties. Although Givón does not offer a model
emphasis is on the detailed description of particular of discourse structure as such, his work is very much
languages, followed by generalization where possible, involved with the relationship between grammar and
rather than on using data from a wide range of lan- discourse, and context is a key concept in his work. The
guages in order to formulate the theory in the first flexibility of language is an important theme in Givón’s
place. writing, and the concept of prototype is central.
SFG takes a strongly constructivist line on language Givón’s work is strongly typological in nature,
acquisition, and there have been longitudinal studies striking a balance between what is common across
of three individual English-speaking children (for languages and their individual characteristics. He
summaries, discussion, and references, see Butler, (1995: 17) comments that ‘‘[w]hile recognizing
2003b: 413–424). cross-language typological diversity, most functional-
We should note here the existence of a second ists remain committed to the existence of language
version of SFG, developed by Fawcett and colleagues. universals, presumably both of meaning/function and
This model takes Halliday’s work as its starting point grammatical structure.’’
but modifies it in a number of important ways, one Finally, Givón takes an explicitly constructivist
of which is to propose clearly separate levels of position on language acquisition, claiming that what
semantics and syntax (Fawcett, 2000). However, the child initially acquires is not syntax, in the
this version still treats the semantics as the underlying Chomskyan sense, but rather a more rudimentary
driving force at the center of the model, with syntax communicative system, which only later becomes
as one means (together with lexical items and modified toward syntax. He castigates Chomsky for
intonation) for the realization of meaning. his rejection of communicative function, general
cognitive capacities, and sociocultural effects as im-
portant factors in acquisition.
West Coast Functionalism
So-called West Coast functional grammar was never a
Usage-Based Functionalist–Cognitive Models
single, unified model but, rather, a fairly loose collec-
tion of individual approaches. We cannot, therefore, The label usage-based is used for a group of models
survey the whole area here but will concentrate on that claim that ‘‘the speaker’s linguistic system is fun-
one important exemplar, the work of Givón. damentally grounded in ‘usage events’: instances of a
Givón subscribes fully to the central tenets of func- speaker’s producing and understanding language’’
tionalism discussed previously. He (1995: xv) holds (Kemmer and Barlow, 2000: viii). This label brings
that ‘‘language (and grammar) can be neither de- together models derived from earlier work in West
scribed nor explained adequately as an autonomous Coast functionalism with approaches such as Lang-
system,’’ but that in order to understand why language acker’s cognitive grammar and some variants of con-
is as it is, we must make reference to ‘‘the natural struction grammar, such as Goldberg’s model. Here,
 Functionalist Theories of Language 703

we concentrate on those approaches that derive from (Bybee and Hopper, 2001). Typological data are
the earlier work of West Coast functionalists. also crucial and are taken to support the paramount
One important strand of usage-based functional importance given to discourse motivations for gram-
linguistics is Hopper’s Emergent Grammar, which matical structuring. Finally, a clearly constructivist
takes a very radically functionalist stand in claiming line is taken with respect to language acquisition
that ‘‘[g]rammar . . . is not the source of understand- (Tomasello, 2003).
ing and communication, but a by-product of it. Gram-
mar is, in other words, epiphenomenal’’ (Hopper, Conclusions
1998: 156). Thus, a grammar is seen not as a unified,
relatively stable system but, rather, as ‘‘an open-ended Functionalism, like other terms for ‘schools’ of lin-
collection of forms that are constantly being restruc- guistics, is a convenient label for a complex, varied set
tured and resemanticized during actual use’’ (Hopper, of approaches to linguistic theory and description. We
1998: 159). This model is thus firmly rooted in the have seen, however, that these approaches are united
communicative functions of language and in the cog- by rejection of the claim that the linguistic system
nitive and social contexts of language use. It also pays should be studied independently of the cognitive,
great attention to relationships between usage, syn- sociocultural, and temporal factors that at least par-
chronic variation, and diachronic change, through tially motivate it and also by rejection of the claim
the study of grammaticalization. that syntax is autonomous from semantics and prag-
Closely related to emergentism is work by Thompson, matics. Functional models regard language as primar-
Hopper, Bybee, and others, which prioritizes the study ily a means for human communication in context and
of authentic conversational data, and in which the attempt to explain as much as possible in terms of
concept of broad syntactic templates is replaced by functional motivations, which may compete to give
local, lexically bound schemas or collocations, often the appearance of arbitrariness in the system. Many
formulaic in nature, which are specific to the language also exhibit a number of derived characteristics, al-
concerned. These reusable fragments, often with slots though these are much more variable across models:
for variable items, are seen as empirically verifiable They are not confined to a core grammar but, rather,
instantiations of the concept of construction, seen attempt to take on board the full range of systematic
as a pairing of form and meaning/function, which linguistic phenomena; they often use authentic lin-
is central to construction grammar models. For in- guistic productions as part of their database; they
stance, Thompson (2002) presents evidence that attempt to model the flexibility and ‘fuzziness’ of
in conversational English what are traditionally ana- language; they are concerned with discourse as well
lyzed as complement-taking predicates (e.g., think þ as the sentence grammar; they study typological vari-
complement clause) are better seen in terms of the ation in language; and they adopt a constructivist
combination of (1) a fragment, often formulaic in approach to the acquisition of language by the child.
nature, with epistemic, evidential, or evaluative Finally, it is important to note that there are impor-
function, such as I think/I don’t think/I thought/ tant similarities between functional models, as defined
I didn’t think/I guess/I remember, etc., and (ii) a de- here, and explicitly cognitively oriented approaches
clarative or interrogative clause that carries the main such as Cognitive Grammar and the various models of
representational content. Construction Grammar. This is reflected, for example,
Clearly, these usage-based approaches adopt a very in the fact that recent usage-based models embrace
inclusive stance on the range of phenomena that are both functional and cognitive orientations (see also
to be accounted for, since the basic aim is to provide Horie and Comrie, 2000).
an explanatory treatment of the rich diversity to be
found in authentic interaction. These models are very See also: Cognitive Grammar; Construction Grammar;
much concerned with the flexible responses of lan- Formulaic Language; Functional Discourse Grammar;
guage to the contexts in which it is used: they are Functional Grammar: Martinet; Minimalism; Principles
rooted in the discourse and the ways in which it and Parameters Framework of Generative Grammar;
conditions the choice of what is said and how it is Role and Reference Grammar; Systemic Theory.
expressed. Considerable weight is given to the
frequency of particular items and combinations in
language use, on the grounds that frequency demon-
Bibliography
strably affects a range of phenomena, including pho- Butler C S (2003a). Structure and function: a guide to three
nological reduction, constituent structure, and the major structural–functional theories. Part 1: approaches
retaining of irregular or even archaic patterns to the simplex clause. Amsterdam: Benjamins.
704 Functionalist Theories of Language

Butler C S (2003b). Structure and function: a guide to three Halliday M A K & Matthiessen C M I M (1999). Constru-
major structural–functional theories. Part 2: from clause ing experience through meaning: a language-based
to discourse and beyond. Amsterdam: Benjamins. approach to cognition. London: Cassell.
Bybee J & Hopper P J (2001). Frequency and the emergence Hopper P J (1998). ‘Emergent grammar.’ In Tomasello M
of linguistic structure. Typological studies in language (ed.) The new psychology of language: cognitive and
No. 45. Amsterdam: Benjamins. functional approaches to language structure. Mahwah,
Caffarel A, Martin J R & Matthiessen C M I M (eds.) NJ: Erlbaum. 155–175.
(2004). Language typology: a functional perspective. Horie K & Comrie B (2000). ‘Introduction.’ In Horie K
Amsterdam: Benjamins. (ed.) Complementation (cognitive and functional perspec-
Chomsky N (1980). Rules and representations. New York: tives). Amsterdam: Benjamins. 1–10.
Columbia University Press. Kemmer S & Barlow M (2000). ‘Introduction: a usage-based
Croft W (1995). ‘Autonomy and functionalist linguistics.’ conception of language.’ In Barlow M & Kemmer S (eds.)
Language 71, 490–532. Usage-based models of language. Stanford, CA: CSLI.
Croft W (2003). Typology and universals. Cambridge vii–xxviii.
textbooks in linguistics (2nd edn.). Cambridge, UK: Newmeyer F J (1998). Language form and function.
Cambridge University Press. Cambridge, MA: MIT Press.
Dik S C (1986). ‘On the notion ‘‘functional explanation.’’’ Newmeyer F J (2003a). ‘Grammar is grammar and usage is
Belgian Journal of Linguistics 1, 11–52. usage.’ Language 79(4), 682–707.
Dik S C (1997a). The theory of Functional Grammar, part 1: Newmeyer F J (2003b). ‘Meaning, function, and the auton-
the structure of the clause (2nd edn.). Hengeveld K (ed.) omy of syntax.’ In Moore J & Polinsky M (eds.) The
Functional Grammar series No. 20. Berlin: de Gruyter. nature of explanation in linguistic theory. Stanford, CA:
Dik S C (1997b). The theory of Functional Grammar, part CSLI. 263–280.
2: complex and derived constructions. Hengeveld K (ed.) Thompson S A (2002). ‘‘‘Object complements’’ and conver-
Functional Grammar series No. 21. Berlin: de Gruyter. sation: towards a realistic account.’ Studies in Language
Fawcett R P (2000). A theory of syntax for Systemic Func- 26, 125–163.
tional Linguistics. Current issues in linguistic theory Tomasello M (2003). Constructing a language: a usage-
No. 206. Amsterdam: Benjamins. based theory of language acquisition. Cambridge, MA:
Foley W A & VanValin R D Jr (1984). Functional syntax Harvard University Press.
and universal grammar. Cambridge, UK: Cambridge VanValin R D Jr (1993). ‘A synopsis of Role and Reference
University Press. Grammar.’ In Van Valin R D Jr (ed.) Advances in Role
Givón T (1995). Functionalism and grammar. Amsterdam: and Reference Grammar. Amsterdam: Benjamins.
Benjamins. VanValin R D Jr (2005). Exploring the syntax–
Haiman J (1983). ‘Iconic and economic motivation.’ semantics interface. Cambridge, UK: Cambridge Univer-
Language 59, 781–819. sity Press.
Halliday M A K (1994). An introduction to functional VanValin R D Jr & LaPolla R J (1997). Syntax: structure,
grammar (2nd edn.). London: Arnold. meaning and function. Cambridge, UK: Cambridge
Halliday M A K (2004). An introduction to Functional University Press.
Grammar (3rd edn., revised by C M I M Matthiessen).
London: Arnold.

Future Tense and Future Time Reference

Ö Dahl, Stockholm University, Stockholm, Sweden can be the subject of our hopes, plans, conjectures,
ß 2006 Elsevier Ltd. All rights reserved. and predictions. Philosophers such as Aristotle have
claimed that the future has a special nature not only
epistemologically but also ontologically: statements
It is tempting to think of time simply as a line extend- about the future do not yet have a determinate truth
ing in both directions from the point at which we value. In a possible worlds framework, the ‘branching
happen to be located. However, in constructing a futures model’ can be seen as an expression of a
theory of temporal semantics, we have to acknowl- similar idea: time looks like a tree rather than a line,
edge that what is ahead of us – the future – is episte- and at any point in the tree there is only one way back
mologically radically different from both what is into the past, but many branches lead into the future.
behind us – the past – and what is taking place at Against this background, it is perhaps not so
this moment – the present. Future states of affairs strange that there tend to be asymmetries in the
cannot be perceived or remembered, although they ways in which temporal reference is structured in