Zgryźliwość kojarzy mi się z radością, która źle skończyła.
18. Grammaticalization : The Handbook of Historical Linguistics : Blackwell Reference Online
12/11/2007 03:40 PM
18. Grammaticalization
BERND HEINE
Subject
Linguistics
»
Historical Linguistics
Key-Topics
grammar
DOI:
10.1111/b.9781405127479.2004.00020.x
Grammaticalization theory is neither a theory of language nor of language change; its goal is to describe
grammaticalization,
1
that is, the way grammatical forms arise and develop through space and time, and to
explain why they are structured the way they are (see section 2). Grammaticalization is defined as a process
2
which is hypothesized to be essentially unidirectional
3
(see section 3).
Grammaticalization is frequently described as leading from lexical to grammatical (= functional) categories.
This view takes care of quite a number of linguistic phenomena, but it does not account for much of what
happens in the development of grammatical categories. It suffers in particular from two main shortcomings.
First, the process is not confined to the development of lexical forms; rather, grammatical forms themselves
can, and frequently do, give rise to even more grammatical forms. Second, since linguistic items require
specific contexts and constructions to undergo grammaticalization, grammaticalization theory is also
concerned with the pragmatic and morphosyntactic environment in which this process occurs. While
grammaticalization has both a synchronic and a diachronic dimension, its foundation is diachronic in nature.
In the following we will distinguish between grammaticalization, which relates to specific linguistic
phenomena, grammaticalization studies, which deal with the analysis of these phenomena, and
grammaticalization theory, which proposes a descriptive and explanatory account of these phenomena (see
section 2).
1 Earlier Work
In the history of grammaticalization studies, three main phases can be distinguished. The first phase is
associated with the work of eighteenth-century French and British philosophers. étienne Bonnot de Condillac
claims that grammatical complexity and abstract vocabulary derive historically from concrete lexemes.
Condillac (1746) argued that tense suffixes and other verbal inflections can be traced back to independent
words: the latter coalesce to give rise to verbal tense and aspect forms. Some notions of modern grammati-
calization theory are also contained in the work of John Horne Tooke (1857). In his work, first published in
1786 and 1805, he argues that language in its “original stage” is concrete, and abstract phenomena are
derived from concrete ones. Horne Tooke proposed “abbreviation” and “mutilation” as key notions: nouns
and verbs are called “necessary words” while other word classes, like adverbs, prepositions, and
conjunctions, are derived from “necessary words” via abbreviation and mutilation.
The second phase is associated mainly with German nineteenth-century linguists. The first main
representative was Franz Bopp (1816, 1833), who considered the change from lexical to grammatical forms
to be an essential component of his principles of comparative grammar. While various examples discussed
by Bopp, in the same way as those proposed by his predecessors, are etymologically of doubtful value, a
number of insights emerged in the course of his work. Bopp was but the first in a long series of nineteenth-
century linguists for whom grammaticalization became a key notion (although the term was introduced only
http://www.blackwellreference.com/subscriber/uid=532/tocnode?id=g9781405127479_chunk_g978140512747920
Page 1 of 20
18. Grammaticalization : The Handbook of Historical Linguistics : Blackwell Reference Online
12/11/2007 03:40 PM
century linguists for whom grammaticalization became a key notion (although the term was introduced only
much later; see below), other authors being August Wilhelm von Schlegel (1818), Wilhelm von Humboldt
(1825), Franz WÜllner (1831), William Dwight Whitney (1875), and, most of all, Georg von der Gabelentz
(1901). After the turn of the century, grammaticalization studies declined.
No major developments took place in the course of the twentieth century prior to 1970. The few authors
who made use of findings on grammaticalization, like Meillet (1912), who introduced the term (French:
grammaticalisation
), or Kurylowicz (1965), were Indo-Europeanists who used findings on grammaticalization
as part of their methodology in historical linguistics but did not contribute much beyond what had been
known already by the end of the nineteenth century.
The third phase of grammaticalization studies started in the 1970s and was initially connected with the
paradigm of localism (Anderson 1971, 1973). According to this school, spatial expressions are more basic
than other kinds of linguistic expressions and the former therefore serve as structural templates for the
latter. More importantly, however, developments in the early 1970s were connected with the work of Talmy
Givón, who argued that in order to understand language structure one must have a knowledge of its earlier
stages of development. With his slogan “Today's morphology is yesterday's syntax,”
4
which he considered to
be part of a more general cyclic evolution (see section 7), as sketched in (1), he opened a new perspective
for understanding grammar (Givón 1971: 12, 1979):
(1) Discourse > Syntax > Morphology > Morphophonemics > Zero
In the course of the 1970s and 1980s a number of studies appeared, many of them concerned with problems
of morphosyntactic change (see, e.g., the contributions in Li 1977), which were based on assumptions such
as the following:
i Language is a historical product and should therefore be accounted for first of all with reference to
the historical forces that are responsible for its present structure.
ii Accordingly, findings on grammaticalization offer more comprehensive explanations than findings
confined to synchronic analysis could offer.
iii As had already been claimed since Condillac's time, the development of grammatical categories is
unidirectional, leading from concrete/lexical to abstract/grammatical meanings. (Traugott 1980; Heine
and Reh 1982, 1984; Lehmann 1982; Bybee 1985)
While virtually all of the various authors adhering to that paradigm subscribe to the same general approach,
according to which grammaticalization is defined as the development from lexical to grammatical and from
grammatical to even more grammatical structures, a wide range of different opinions and theoretical
orientations arose. In some of the works (e.g., Traugott 1980), the main contribution of this field consists in
offering new ways of reconstructing semantic change. In other works, grammaticalization theory is viewed as
a means of describing and explaining the structure of grammatical categories across languages (Bybee 1985;
Bybee et al. 1991, 1994). Others again propose to treat grammaticalization as being synonymous, or nearly
synonymous, with grammar: for Hopper (1987), in particular, grammaticalization, or emergent grammar, has
to do with the recurrent strategies used for building discourses and involves a continual movement toward
structure. Finally, there are those who argue that grammar is the result of an interplay between
conceptualization and communication, and that grammaticalization theory provides a tool for reconstructing
some of the extralinguistic foundations of grammar (Heine et al. 1991; Heine 1997b).
The diversity of views that have been voiced on grammaticalization is also reflected in the terminology
employed: rather than “grammaticalization,” some authors prefer to call it “grammaticization,” or
“grammatization.” Furthermore, there are also major differences as to what subject matters should be
subsumed under the term. For some, the term “grammaticalization” is merely an equivalent to “grammatical
form”; such authors may say, for example, that language X has grammaticalized a dative case, which is
roughly equivalent to saying that in X there exists a grammatical form for this case function.
A wealth of books and articles is now available, either as monographic treatments (Traugott and Heine
1991a, 1991b; Heine et al. 1991; Hopper and Traugott 1993; Pagliuca 1994; Ramat and Hopper 1998), or as
applications of findings on grammaticalization to one particular language (e.g., Kilian-Hatz 1995 on Baka;
Sun 1996 on Chinese; Diewald 1997 on German).
http://www.blackwellreference.com/subscriber/uid=532/tocnode?id=g9781405127479_chunk_g978140512747920
Page 2 of 20
18. Grammaticalization : The Handbook of Historical Linguistics : Blackwell Reference Online
12/11/2007 03:40 PM
Sun 1996 on Chinese; Diewald 1997 on German).
Furthermore, a number of works are devoted to the grammaticalization of specific parts of grammar, such as
tense and aspect (Dahl 1985; Brinton 1988; Bybee and Dahl 1989; Fleischman 1982; Bybee et al. 1994;
Kuteva 1998), modality (Traugott 1989; Bybee and Fleischman 1995), evidentials (Willett 1988), auxiliaries
(Heine 1993; Kuteva 2001), copulas (Devitt 1990, 1994), serial verbs (Givón 1975; Lord 1993),
demonstratives (Diessel 1999), definite articles (Greenberg 1978; Himmelmann 1997), indefinite articles
(Givón 1981; Heine 1997b: 66–82), indefinite pronouns (Haspelmath 1997a), anaphora (Givón 1976;
Frajzyngier 1987), reflexives (Kemmer 1993; König and Siemund 2000; Schladt 2000; Heine 2000), passives
(Haspelmath 1990), temporal adverbs (Haspelmath 1997b), infinitives (Haspelmath 1990), quotatives (Ebert
1991; Saxena 1988, 1995; GÜldemann 2001), clause subordinators (Genetti 1986, 1991; Traugott 1985,
1986; König 1985, 1988; Hopper and Traugott 1993: 176ff), complementizers (Ransom 1988), relative
clauses (Lehmann 1984), numerals (Heine 1997b: 18–34), comparatives (Heine 1994, 1997b: 109–30),
spatial orientation (Svorou 1994; Heine 1997b: 35–65), or possession (Heine 1997a).
More recent work shows that grammaticalization studies are equally relevant to understanding language
change in situations of extreme language contact and unusual language transmission. There is now a wealth
of studies on grammaticalization in pidgins and creoles (see, e.g., Sankoff and Brown 1976; Arends 1986;
Plag 1992, 1993, forthcoming; Baker and Syea 1996, Bruyn 1995, 1996; Huber 1996; Mufwene 1996;
Poplack and Tagliamonte 1996; Romaine 1995, 1999), and these studies suggest that (with few exceptions;
cf. Bruyn 1996; Plag forthcoming), grammatical categories in these languages evolve along the same lines as
in languages with “natural” language transmission.
2 The Framework
As observed earlier, grammaticalization theory is a theory to the extent that it offers an explanatory account
of how and why grammatical categories arise and develop. It is based on the following assumption: the main
motivation underlying grammaticalization is to communicate successfully. To this end, one salient human
strategy consists in using linguistic forms for meanings that are concrete, easily accessible, and/or clearly
delineated to also express less concrete, less easily accessible, and less clearly delineated meaning contents.
To this end, lexical or less grammaticalized linguistic expressions are pressed into service for the expression
of more grammaticalized functions.
5
Accordingly, grammaticalization is a process whereby expressions for
concrete (= source) meanings are used in specific contexts for encoding grammatical (= target) meanings.
This process has a number of implications for the structure of the expressions concerned.
Technically, the grammaticalization of linguistic expressions involves four interrelated mechanisms:
6
i desemanticization (or “bleaching,” semantic reduction): loss in meaning content;
ii extension (or context generalization): use in new contexts;
iii decategorialization: loss in morphosyntactic properties characteristic of the source forms, including
the loss of independent word status (cliticization, affixation);
iv erosion (or “phonetic reduction”), that is, loss in phonetic substance.
Each of these mechanisms is concerned with a different aspect of language structure or language use: (i)
relates to semantics, (ii) to pragmatics, (iii) to morphosyntax, and (iv) to phonetics. While three of these
mechanisms involve a loss in properties, there are also gains: in the same way as linguistic items undergoing
grammaticalization lose in semantic, morphosyntactic, and phonetic substance, they also gain in properties
characteristic of their uses in new contexts (cf. (ii)), sometimes to the extent that their meaning may show
little resemblance to the original meaning. None of the mechanisms is confined to grammaticalization (see
Newmeyer 1998; Campbell 2001a); but to the extent that jointly they are responsible for grammaticalization
taking place, they can be said to constitute different components of one and the same general process.
Each of these mechanisms gives rise to an evolution which can be described in the form of a three-stage
model, called the overlap model (Heine 1993: 48–53). The stages concerned are as follows:
i There is a linguistic expression A that is recruited for grammaticalization.
http://www.blackwellreference.com/subscriber/uid=532/tocnode?id=g9781405127479_chunk_g978140512747920
Page 3 of 20
18. Grammaticalization : The Handbook of Historical Linguistics : Blackwell Reference Online
12/11/2007 03:40 PM
ii This expression acquires a second use pattern, B, with the effect that there is ambiguity between A
and B.
iii Finally, A is lost, that is, there is now only B.
The result of this process is that grammaticalization exhibits a chain-like structure (see section 5). Note that
not all instances of grammaticalization in fact proceed to stage (iii); it may happen that the process is
arrested at stage (ii); however, once stage (iii) is reached, B tends to be conventionalized, that is, it turns
into a new grammatical category.
Desemanticization results from the use of forms for concrete meanings which are reinterpreted in specific
contexts as more abstract grammatical meanings (see section 5 for a more detailed discussion). While this
term is commonly understood to refer to the loss of lexical content, an equally common type of
desemanticization concerns cases where a grammatical form having two (or more) grammatical functions
loses one (or all) of these functions. For example, the Old Swedish nominal inflections were typically
portmanteau (cumulative) morphemes simultaneously expressing gender, number, and case.
Desemanticization in this case had the effect that one of the three functions, namely case, was lost in the
development to modern Swedish (Norde 2001: 243).
The term “extension” is adopted from Harris and Campbell (1995; see also Campbell 2001a: 142–3).7 While
these authors emphasize the syntactic manifestations of this mechanism, we are confined here to one of its
pragmatic manifestations: we will assume that extension obtains when a linguistic item can be used in new
contexts where it could not be used previously.
8
Once a form has acquired a new grammatical meaning, it tends to become increasingly divergent: it loses in
categorial properties characteristic of its source uses, hence it undergoes decategorialization, and it tends to
be used more frequently and in more contexts, to become more predictable in its text occurrence and,
consequently, it tends to lose in phonetic substance, hence to undergo erosion. Thus, to the extent that
extension, decategorialization, and erosion are components of a grammaticalization process, they
presuppose desemanticization
9
(cf. Haspelmath 1999: 1062). In the early stages of grammaticalization there
may be a shift from less to more grammatical meaning although there are as yet no noticeable pragmatic,
morphosyntactic, or phonetic changes associated with that shift (for examples, see below).
The following example from Swahili illustrates the effect of these mechanisms. Like many other languages
(see Bybee et al. 1991), Swahili has grammaticalized a verb of volition to a future tense marker. Example
(2a) illustrates the lexical use of the verb
-taka
‘want,’ while (2b) illustrates its use as a future tense marker
in relative clauses. In main clauses, the future marker was reduced to -ta-, cf. (2c). Desemanticization had
the effect that the lexical meaning of the verb was “bleached out.” Originally a lexical verb requiring typically
human subject referents, its use was extended to contexts involving inanimate subjects (extension). In
accordance with its use as a tense marker,
-taka
underwent decategorialization: it lost its status as an
independent word and most other verbal properties and became a prefix of the main verb. Finally,
-taka
underwent erosion, being phonologically reduced to -ta- in main clauses (but retaining its original full form
in relative clauses; see above):
(2) Swahili (Bantu, Niger-Congo):
http://www.blackwellreference.com/subscriber/uid=532/tocnode?id=g9781405127479_chunk_g978140512747920
Page 4 of 20
18. Grammaticalization : The Handbook of Historical Linguistics : Blackwell Reference Online
12/11/2007 03:40 PM
As noted above, there are gains deriving from the use of an item in new contexts that can offset losses of
properties it may undergo. Moreover, grammaticalization requires specific contexts to take place and it
therefore has been described as a product of pragmatic inferencing, pragmatic enrichment, strengthening, or
conversational implicatures (Hopper and Traugott 1993: 163–77), or, as we will say, context-induced
reinterpretation (see section 5).
As we will see below (sections 5 and 6), the framework described here has a number of implications for the
development and structure of grammatical categories, and a number of models and descriptive devices have
been proposed to deal with these implications.
3 Problems
Work on grammaticalization has been the subject of a number of critical discussions. Some authors
expressed dissatisfaction with the classical definition proposed by Kurylowicz (1965), proposing a more
extensive use of the term. For example, Traugott (this volume) proposes to define grammaticalization “as
the development of constructions […] via discourse practices into more grammatical material.” As we
observed in the introduction to this chapter, the development of grammatical items is shaped by the
constructions in which these items occur; nevertheless, many grammaticalization processes that have been
identified so far have been described largely without reference to constructions. Conversely there are no
convincing examples so far to suggest that instances of grammaticalization processes can be identified
exclusively in terms of constructions without referring to the form-meaning items involved in the process.
While there is now a wealth of publications on grammaticalization, extending from articles to books and
contributions to handbooks on language structure and language change, in more recent years there has
been massive criticism of grammaticalization theory (see especially Newmeyer 1998; Campbell 2001a;
Campbell and Janda 2001; Janda 2001; Norde 2001). In this work, a number of weaknesses and
inconsistencies found in previous analyses of grammaticalization are pointed out, and attention is drawn to
areas of research that have been neglected or ignored in earlier work. In the present section, the main points
of criticism are examined. It goes without saying that in a concise treatment like the present one it is not
possible to do justice to all the problems that have been raised and all the views that have been expressed.
We will therefore be confined to a few claims that challenge cornerstones of grammaticalization theory. Such
claims are:
i Not all instances of grammatical change are due to grammaticalization.
ii Grammaticalization is not unidirectional.
iii Grammaticalization is not a distinct process.
iv “Grammaticalization theory” is not a theory.
With regard to (i), more recent research has demonstrated that grammatical change involves factors that are
not covered by grammaticalization theory (see especially Newmeyer 1998; Campbell 2001a; Janda 2001;
Joseph 2001a; Norde 2001), and future work will have to deal with these factors in more detail. Most of this
http://www.blackwellreference.com/subscriber/uid=532/tocnode?id=g9781405127479_chunk_g978140512747920
Page 5 of 20
zanotowane.pl doc.pisz.pl pdf.pisz.pl hannaeva.xlx.pl
12/11/2007 03:40 PM
18. Grammaticalization
BERND HEINE
Subject
Linguistics
»
Historical Linguistics
Key-Topics
grammar
DOI:
10.1111/b.9781405127479.2004.00020.x
Grammaticalization theory is neither a theory of language nor of language change; its goal is to describe
grammaticalization,
1
that is, the way grammatical forms arise and develop through space and time, and to
explain why they are structured the way they are (see section 2). Grammaticalization is defined as a process
2
which is hypothesized to be essentially unidirectional
3
(see section 3).
Grammaticalization is frequently described as leading from lexical to grammatical (= functional) categories.
This view takes care of quite a number of linguistic phenomena, but it does not account for much of what
happens in the development of grammatical categories. It suffers in particular from two main shortcomings.
First, the process is not confined to the development of lexical forms; rather, grammatical forms themselves
can, and frequently do, give rise to even more grammatical forms. Second, since linguistic items require
specific contexts and constructions to undergo grammaticalization, grammaticalization theory is also
concerned with the pragmatic and morphosyntactic environment in which this process occurs. While
grammaticalization has both a synchronic and a diachronic dimension, its foundation is diachronic in nature.
In the following we will distinguish between grammaticalization, which relates to specific linguistic
phenomena, grammaticalization studies, which deal with the analysis of these phenomena, and
grammaticalization theory, which proposes a descriptive and explanatory account of these phenomena (see
section 2).
1 Earlier Work
In the history of grammaticalization studies, three main phases can be distinguished. The first phase is
associated with the work of eighteenth-century French and British philosophers. étienne Bonnot de Condillac
claims that grammatical complexity and abstract vocabulary derive historically from concrete lexemes.
Condillac (1746) argued that tense suffixes and other verbal inflections can be traced back to independent
words: the latter coalesce to give rise to verbal tense and aspect forms. Some notions of modern grammati-
calization theory are also contained in the work of John Horne Tooke (1857). In his work, first published in
1786 and 1805, he argues that language in its “original stage” is concrete, and abstract phenomena are
derived from concrete ones. Horne Tooke proposed “abbreviation” and “mutilation” as key notions: nouns
and verbs are called “necessary words” while other word classes, like adverbs, prepositions, and
conjunctions, are derived from “necessary words” via abbreviation and mutilation.
The second phase is associated mainly with German nineteenth-century linguists. The first main
representative was Franz Bopp (1816, 1833), who considered the change from lexical to grammatical forms
to be an essential component of his principles of comparative grammar. While various examples discussed
by Bopp, in the same way as those proposed by his predecessors, are etymologically of doubtful value, a
number of insights emerged in the course of his work. Bopp was but the first in a long series of nineteenth-
century linguists for whom grammaticalization became a key notion (although the term was introduced only
http://www.blackwellreference.com/subscriber/uid=532/tocnode?id=g9781405127479_chunk_g978140512747920
Page 1 of 20
18. Grammaticalization : The Handbook of Historical Linguistics : Blackwell Reference Online
12/11/2007 03:40 PM
century linguists for whom grammaticalization became a key notion (although the term was introduced only
much later; see below), other authors being August Wilhelm von Schlegel (1818), Wilhelm von Humboldt
(1825), Franz WÜllner (1831), William Dwight Whitney (1875), and, most of all, Georg von der Gabelentz
(1901). After the turn of the century, grammaticalization studies declined.
No major developments took place in the course of the twentieth century prior to 1970. The few authors
who made use of findings on grammaticalization, like Meillet (1912), who introduced the term (French:
grammaticalisation
), or Kurylowicz (1965), were Indo-Europeanists who used findings on grammaticalization
as part of their methodology in historical linguistics but did not contribute much beyond what had been
known already by the end of the nineteenth century.
The third phase of grammaticalization studies started in the 1970s and was initially connected with the
paradigm of localism (Anderson 1971, 1973). According to this school, spatial expressions are more basic
than other kinds of linguistic expressions and the former therefore serve as structural templates for the
latter. More importantly, however, developments in the early 1970s were connected with the work of Talmy
Givón, who argued that in order to understand language structure one must have a knowledge of its earlier
stages of development. With his slogan “Today's morphology is yesterday's syntax,”
4
which he considered to
be part of a more general cyclic evolution (see section 7), as sketched in (1), he opened a new perspective
for understanding grammar (Givón 1971: 12, 1979):
(1) Discourse > Syntax > Morphology > Morphophonemics > Zero
In the course of the 1970s and 1980s a number of studies appeared, many of them concerned with problems
of morphosyntactic change (see, e.g., the contributions in Li 1977), which were based on assumptions such
as the following:
i Language is a historical product and should therefore be accounted for first of all with reference to
the historical forces that are responsible for its present structure.
ii Accordingly, findings on grammaticalization offer more comprehensive explanations than findings
confined to synchronic analysis could offer.
iii As had already been claimed since Condillac's time, the development of grammatical categories is
unidirectional, leading from concrete/lexical to abstract/grammatical meanings. (Traugott 1980; Heine
and Reh 1982, 1984; Lehmann 1982; Bybee 1985)
While virtually all of the various authors adhering to that paradigm subscribe to the same general approach,
according to which grammaticalization is defined as the development from lexical to grammatical and from
grammatical to even more grammatical structures, a wide range of different opinions and theoretical
orientations arose. In some of the works (e.g., Traugott 1980), the main contribution of this field consists in
offering new ways of reconstructing semantic change. In other works, grammaticalization theory is viewed as
a means of describing and explaining the structure of grammatical categories across languages (Bybee 1985;
Bybee et al. 1991, 1994). Others again propose to treat grammaticalization as being synonymous, or nearly
synonymous, with grammar: for Hopper (1987), in particular, grammaticalization, or emergent grammar, has
to do with the recurrent strategies used for building discourses and involves a continual movement toward
structure. Finally, there are those who argue that grammar is the result of an interplay between
conceptualization and communication, and that grammaticalization theory provides a tool for reconstructing
some of the extralinguistic foundations of grammar (Heine et al. 1991; Heine 1997b).
The diversity of views that have been voiced on grammaticalization is also reflected in the terminology
employed: rather than “grammaticalization,” some authors prefer to call it “grammaticization,” or
“grammatization.” Furthermore, there are also major differences as to what subject matters should be
subsumed under the term. For some, the term “grammaticalization” is merely an equivalent to “grammatical
form”; such authors may say, for example, that language X has grammaticalized a dative case, which is
roughly equivalent to saying that in X there exists a grammatical form for this case function.
A wealth of books and articles is now available, either as monographic treatments (Traugott and Heine
1991a, 1991b; Heine et al. 1991; Hopper and Traugott 1993; Pagliuca 1994; Ramat and Hopper 1998), or as
applications of findings on grammaticalization to one particular language (e.g., Kilian-Hatz 1995 on Baka;
Sun 1996 on Chinese; Diewald 1997 on German).
http://www.blackwellreference.com/subscriber/uid=532/tocnode?id=g9781405127479_chunk_g978140512747920
Page 2 of 20
18. Grammaticalization : The Handbook of Historical Linguistics : Blackwell Reference Online
12/11/2007 03:40 PM
Sun 1996 on Chinese; Diewald 1997 on German).
Furthermore, a number of works are devoted to the grammaticalization of specific parts of grammar, such as
tense and aspect (Dahl 1985; Brinton 1988; Bybee and Dahl 1989; Fleischman 1982; Bybee et al. 1994;
Kuteva 1998), modality (Traugott 1989; Bybee and Fleischman 1995), evidentials (Willett 1988), auxiliaries
(Heine 1993; Kuteva 2001), copulas (Devitt 1990, 1994), serial verbs (Givón 1975; Lord 1993),
demonstratives (Diessel 1999), definite articles (Greenberg 1978; Himmelmann 1997), indefinite articles
(Givón 1981; Heine 1997b: 66–82), indefinite pronouns (Haspelmath 1997a), anaphora (Givón 1976;
Frajzyngier 1987), reflexives (Kemmer 1993; König and Siemund 2000; Schladt 2000; Heine 2000), passives
(Haspelmath 1990), temporal adverbs (Haspelmath 1997b), infinitives (Haspelmath 1990), quotatives (Ebert
1991; Saxena 1988, 1995; GÜldemann 2001), clause subordinators (Genetti 1986, 1991; Traugott 1985,
1986; König 1985, 1988; Hopper and Traugott 1993: 176ff), complementizers (Ransom 1988), relative
clauses (Lehmann 1984), numerals (Heine 1997b: 18–34), comparatives (Heine 1994, 1997b: 109–30),
spatial orientation (Svorou 1994; Heine 1997b: 35–65), or possession (Heine 1997a).
More recent work shows that grammaticalization studies are equally relevant to understanding language
change in situations of extreme language contact and unusual language transmission. There is now a wealth
of studies on grammaticalization in pidgins and creoles (see, e.g., Sankoff and Brown 1976; Arends 1986;
Plag 1992, 1993, forthcoming; Baker and Syea 1996, Bruyn 1995, 1996; Huber 1996; Mufwene 1996;
Poplack and Tagliamonte 1996; Romaine 1995, 1999), and these studies suggest that (with few exceptions;
cf. Bruyn 1996; Plag forthcoming), grammatical categories in these languages evolve along the same lines as
in languages with “natural” language transmission.
2 The Framework
As observed earlier, grammaticalization theory is a theory to the extent that it offers an explanatory account
of how and why grammatical categories arise and develop. It is based on the following assumption: the main
motivation underlying grammaticalization is to communicate successfully. To this end, one salient human
strategy consists in using linguistic forms for meanings that are concrete, easily accessible, and/or clearly
delineated to also express less concrete, less easily accessible, and less clearly delineated meaning contents.
To this end, lexical or less grammaticalized linguistic expressions are pressed into service for the expression
of more grammaticalized functions.
5
Accordingly, grammaticalization is a process whereby expressions for
concrete (= source) meanings are used in specific contexts for encoding grammatical (= target) meanings.
This process has a number of implications for the structure of the expressions concerned.
Technically, the grammaticalization of linguistic expressions involves four interrelated mechanisms:
6
i desemanticization (or “bleaching,” semantic reduction): loss in meaning content;
ii extension (or context generalization): use in new contexts;
iii decategorialization: loss in morphosyntactic properties characteristic of the source forms, including
the loss of independent word status (cliticization, affixation);
iv erosion (or “phonetic reduction”), that is, loss in phonetic substance.
Each of these mechanisms is concerned with a different aspect of language structure or language use: (i)
relates to semantics, (ii) to pragmatics, (iii) to morphosyntax, and (iv) to phonetics. While three of these
mechanisms involve a loss in properties, there are also gains: in the same way as linguistic items undergoing
grammaticalization lose in semantic, morphosyntactic, and phonetic substance, they also gain in properties
characteristic of their uses in new contexts (cf. (ii)), sometimes to the extent that their meaning may show
little resemblance to the original meaning. None of the mechanisms is confined to grammaticalization (see
Newmeyer 1998; Campbell 2001a); but to the extent that jointly they are responsible for grammaticalization
taking place, they can be said to constitute different components of one and the same general process.
Each of these mechanisms gives rise to an evolution which can be described in the form of a three-stage
model, called the overlap model (Heine 1993: 48–53). The stages concerned are as follows:
i There is a linguistic expression A that is recruited for grammaticalization.
http://www.blackwellreference.com/subscriber/uid=532/tocnode?id=g9781405127479_chunk_g978140512747920
Page 3 of 20
18. Grammaticalization : The Handbook of Historical Linguistics : Blackwell Reference Online
12/11/2007 03:40 PM
ii This expression acquires a second use pattern, B, with the effect that there is ambiguity between A
and B.
iii Finally, A is lost, that is, there is now only B.
The result of this process is that grammaticalization exhibits a chain-like structure (see section 5). Note that
not all instances of grammaticalization in fact proceed to stage (iii); it may happen that the process is
arrested at stage (ii); however, once stage (iii) is reached, B tends to be conventionalized, that is, it turns
into a new grammatical category.
Desemanticization results from the use of forms for concrete meanings which are reinterpreted in specific
contexts as more abstract grammatical meanings (see section 5 for a more detailed discussion). While this
term is commonly understood to refer to the loss of lexical content, an equally common type of
desemanticization concerns cases where a grammatical form having two (or more) grammatical functions
loses one (or all) of these functions. For example, the Old Swedish nominal inflections were typically
portmanteau (cumulative) morphemes simultaneously expressing gender, number, and case.
Desemanticization in this case had the effect that one of the three functions, namely case, was lost in the
development to modern Swedish (Norde 2001: 243).
The term “extension” is adopted from Harris and Campbell (1995; see also Campbell 2001a: 142–3).7 While
these authors emphasize the syntactic manifestations of this mechanism, we are confined here to one of its
pragmatic manifestations: we will assume that extension obtains when a linguistic item can be used in new
contexts where it could not be used previously.
8
Once a form has acquired a new grammatical meaning, it tends to become increasingly divergent: it loses in
categorial properties characteristic of its source uses, hence it undergoes decategorialization, and it tends to
be used more frequently and in more contexts, to become more predictable in its text occurrence and,
consequently, it tends to lose in phonetic substance, hence to undergo erosion. Thus, to the extent that
extension, decategorialization, and erosion are components of a grammaticalization process, they
presuppose desemanticization
9
(cf. Haspelmath 1999: 1062). In the early stages of grammaticalization there
may be a shift from less to more grammatical meaning although there are as yet no noticeable pragmatic,
morphosyntactic, or phonetic changes associated with that shift (for examples, see below).
The following example from Swahili illustrates the effect of these mechanisms. Like many other languages
(see Bybee et al. 1991), Swahili has grammaticalized a verb of volition to a future tense marker. Example
(2a) illustrates the lexical use of the verb
-taka
‘want,’ while (2b) illustrates its use as a future tense marker
in relative clauses. In main clauses, the future marker was reduced to -ta-, cf. (2c). Desemanticization had
the effect that the lexical meaning of the verb was “bleached out.” Originally a lexical verb requiring typically
human subject referents, its use was extended to contexts involving inanimate subjects (extension). In
accordance with its use as a tense marker,
-taka
underwent decategorialization: it lost its status as an
independent word and most other verbal properties and became a prefix of the main verb. Finally,
-taka
underwent erosion, being phonologically reduced to -ta- in main clauses (but retaining its original full form
in relative clauses; see above):
(2) Swahili (Bantu, Niger-Congo):
http://www.blackwellreference.com/subscriber/uid=532/tocnode?id=g9781405127479_chunk_g978140512747920
Page 4 of 20
18. Grammaticalization : The Handbook of Historical Linguistics : Blackwell Reference Online
12/11/2007 03:40 PM
As noted above, there are gains deriving from the use of an item in new contexts that can offset losses of
properties it may undergo. Moreover, grammaticalization requires specific contexts to take place and it
therefore has been described as a product of pragmatic inferencing, pragmatic enrichment, strengthening, or
conversational implicatures (Hopper and Traugott 1993: 163–77), or, as we will say, context-induced
reinterpretation (see section 5).
As we will see below (sections 5 and 6), the framework described here has a number of implications for the
development and structure of grammatical categories, and a number of models and descriptive devices have
been proposed to deal with these implications.
3 Problems
Work on grammaticalization has been the subject of a number of critical discussions. Some authors
expressed dissatisfaction with the classical definition proposed by Kurylowicz (1965), proposing a more
extensive use of the term. For example, Traugott (this volume) proposes to define grammaticalization “as
the development of constructions […] via discourse practices into more grammatical material.” As we
observed in the introduction to this chapter, the development of grammatical items is shaped by the
constructions in which these items occur; nevertheless, many grammaticalization processes that have been
identified so far have been described largely without reference to constructions. Conversely there are no
convincing examples so far to suggest that instances of grammaticalization processes can be identified
exclusively in terms of constructions without referring to the form-meaning items involved in the process.
While there is now a wealth of publications on grammaticalization, extending from articles to books and
contributions to handbooks on language structure and language change, in more recent years there has
been massive criticism of grammaticalization theory (see especially Newmeyer 1998; Campbell 2001a;
Campbell and Janda 2001; Janda 2001; Norde 2001). In this work, a number of weaknesses and
inconsistencies found in previous analyses of grammaticalization are pointed out, and attention is drawn to
areas of research that have been neglected or ignored in earlier work. In the present section, the main points
of criticism are examined. It goes without saying that in a concise treatment like the present one it is not
possible to do justice to all the problems that have been raised and all the views that have been expressed.
We will therefore be confined to a few claims that challenge cornerstones of grammaticalization theory. Such
claims are:
i Not all instances of grammatical change are due to grammaticalization.
ii Grammaticalization is not unidirectional.
iii Grammaticalization is not a distinct process.
iv “Grammaticalization theory” is not a theory.
With regard to (i), more recent research has demonstrated that grammatical change involves factors that are
not covered by grammaticalization theory (see especially Newmeyer 1998; Campbell 2001a; Janda 2001;
Joseph 2001a; Norde 2001), and future work will have to deal with these factors in more detail. Most of this
http://www.blackwellreference.com/subscriber/uid=532/tocnode?id=g9781405127479_chunk_g978140512747920
Page 5 of 20