Cognitive Processes in Grammaticalization - Bybee

8/20/2019 Cognitive Processes in Grammaticalization - Bybee

http://slidepdf.com/reader/full/cognitive-processes-in-grammaticalization-bybee 1/21

Cognitive processes in grammaticalization

Joan Bybee

University of New Mexico

All of linguistic theory is concerned with the enterprise of elucidating the natureof the grammar of human languages. But along with asking the uestion !what is the

nature of grammar"#$ we can also ask !how do languages acuire grammar"#. %n the last

twenty years$ researchers interested in the latter uestion have elaborated a theory ofgrammaticali&ation$ the process by which grammar is created$ and in doing so have also

come up with some interesting new perspectives on the former uestion. ' (our main

findings of this research are listed here)

*rammar is not a static$ closed or self+contained system$ but is highly susceptible

to change and highly affected by language use. ,he loss of grammar is generally

acknowledged and often lamented by prescriptive grammarians$ who mourn the loss of

the distinction between who and whom but fail to re-oice in the creation of new grammar$such as the new future tense signaled by gonna. %n fact$ the creation of new grammatical

morphemes and structures is as common as the loss of old ones.

nce underway$ the course of grammaticali&ation is unidirectional and thus in

principle predictable. Nouns and verbs lose their categorical status and become prepositions$ auxiliaries and other grammatical forms. (ree elements become more

restricted and fuse with other elements. /oosely con-oined main clauses fuse to become a

main plus subordinate clause. ,he reverse directions are rarely attested.

Both the general trends in grammaticali&ation and many of the very specific

developments are not restricted to individual languages$ but are common acrossgenetically and geographically unrelated languages. ,his widespread distribution$ whichwill be illustrated in section 0$ provides a new view of language universals. 1ince patterns

of change cannot in themselves exist in speakers# minds$ the more basic universals must

be the mechanisms that create the changes that are so similar across languages.

Many of the very basic mechanisms that constitute the process of

grammaticali&ation are cognitive processes that are not necessarily restricted to language.

By better understanding these cognitive processes and how they function incommunicative situations$ we will eventually learn the answers to the most fundamental

uestions that linguists ask.

1. Grammaticalization

*rammaticali&ation is usually defined as the process by which a lexical item or aseuence of items becomes a grammatical morpheme$ changing its distribution and

function in the process 2Meillet '3'0$ /ehmann '340$ 5eine and 6eh '347$ 5eine$

8laudi and 59nnemeyer '33'$ 5opper and ,raugott '33:;. ,hus <nglish going to

' ,he terms grammaticali&ation and grammatici&ation are used interchangeably.

'



becomes the intention=future marker gonna. 5owever$ more recently it has been observed

that it is important to add that grammaticali&ation of lexical items takes place within

>A6,%8U/A6 8N1,6U8,%N1 and further that grammaticali&ation is the creation of newconstructions 2Bybee$ to appear$ ,raugott$ to appear;. ,hus going to does not

grammaticali&e in the construction exemplified by I’m going to the store but only in the

construction in which a verb follows to$ as in I’m going to help you. %fgrammaticali&ation is the creation of new constructions 2and their further development;$

then it also can include cases of change that do not involve specific morphemes$ such as

the creation of word order patterns.

(or illustration let us consider the canonical type of grammaticali&ation$ that in

which a lexical item becomes a grammatical morpheme within a particular construction.

<xamining the details will help us understand what cognitive mechanisms are involved inthe process. 1ome characteristics of the grammaticali&ation process are the following)

2'; ?ords and phrases undergoing grammaticali&ation are phonetically reduced$ with

reductions$ assimilations and deletions of consonants and vowels producing seuencesthat reuire less muscular effort 2Browman and *oldstein '330$ Mowrey and >agliuca

'33@;. (or example$ going to becomes gonnaand even reduces further in

some contexts to as in I'm (g)onna

20; 1pecific$ concrete meanings entering into the process become generali&ed and moreabstract$ and as a result$ become appropriate in a growing range of contexts$ as for

example$ the uses of be going to in sentences 2'; through 2:;. ,he literal meaning in 2';

was the only possible interpretation in 1hakespeares <nglish$ but now uses such as those

shown in 20; and 2:; are common.

2'; movement)We are going to Windsor to see the King 20; intention) We are going to get married in June.2:; future) These trees are going to lose their leaves.

2:; A grammaticali&ing construction#s freuency of use increases dramatically as it

develops. ne source of the increased freuency is an increase in the types of contexts in

which the new construction is possible. ,hus when be going to had only its literalmeaning 2as in 'C;$ it could only be used in contexts where movement was to take place$

with sub-ects that were volitional and mobile. Now it can be used even in 2:;$ where no

movement in space on the part of the sub-ect is implied$ or indeed possible. As the gonna

construction becomes appropriate with more types of sub-ects and verbs$ it occurs morefreuently in texts.

27; 8hanges in grammaticali&ation take place very gradually and are accompanied bymuch variation in both form and function. % have already illustrated the variation in form

with be going to and gonna. Dariation in function can be seen in the three examples

above$ of movement$ intention and future$ all of which are still possible uses inModern <nglish.

0



,he mechanisms underlying these changes will be the main focus of this chapter$ but

before examining these in greater detail$ it is important to document the fact thatgrammaticali&ation occurs spontaneously and in the same form at all documented time

periods and in all languages.

2. General patterns of grammaticalization

/et us first list some changes that have occurred in the <nglish language over thelast millenium. 1ince <nglish began to appear in writing some '0EE years ago$ we can

document the development of the definite article$ the, out of the demonstrative$ that $ and

the development of the indefinite article aan out of the numeral one. ,he function of

articles such as the and a is to distinguish between nouns that the hearer can identify asalready known in the discourse or conversational context and those that are being

introduced for the first time. 2(or example$ I met a man at the ban!... where this is the

first mention of a man vs. The man I met at the ban!... which refers back to some

previous mention.; ld <nglish 2as documented in manuscripts from about 4EE+''EE AF;used no articles at all$ but rather could change the position of nouns to show which were

new and which were previously mentioned.

1imilarly$ the <nglish modal auxiliaries$ which express grammatical distinctions

within the categories of tense 2future will ; and modalities such as possibility 2"an and

may;$ all developed from verbs. Will $ which now indicates future tense$ developed from a

verb$ willan$ which meant to wantG "an came from a verb$ "unnan$ meaning to be

acuainted with or to know how toG may came from a verb$ magan$ meaning to be able

to$ to have the power. #ould and might developed from the past tense forms of "unnan and magan respectively. ?e have already mentioned the more complex phrase be going

to$ which in 1hakespeares <nglish still described actual movement in space$ fuses into

gonna and comes to be used for future time reference.

<ven affixes derive from full words. (or instance$ the <nglish suffix $ly derived

from a noun$ which in ld <nglish was li%$ meaning body. ,he compound mann$li% originally meant having the body or appearance of a man whence it generali&ed to

having the characteristics of a man$ the modern sense of manly.

,hese facts of <nglish are interesting enough as isolated facts about one language$ but they develop a profound importance with the discovery that all around the world$ in

languages that are not related genetically or geographically$ we find analogous examples)

definite articles developing from demonstratives$ indefinite articles from the numeralone$ future tenses from verbs meaning want or go to and auxiliaries indicating

possibility and permission from verbs meaning know and be able.

(or instance$ in many <uropean languages$ an indefinite article has developed out

of the numeral one) <nglish aan$ *erman ein$ (rench unune, 1panish ununa$ and

Modern *reek ena. ?hile these are all %ndo+<uropean languages$ in each case this

development occurred after these languages had differentiated from one another and

:



speakers were no longer in contact. (urthermore$ the numeral !one# is used as an

indefinite article in MorH$ a *ur language of the Burkina (aso 25eine et al. '33:;$ in

collouial 5ebrew 21emitic; and in the Fravidian languages ,amil and Iannada 25eine'33;. <xamples of demonstratives becoming definite articles are also common) /atin

ille, illa that became (rench definite articles le, la and 1panish el, la& in Dai 2a Mande

language of /iberia and 1ierra /eone; the demonstrative

this becomes a suffixeddefinite article 25eine et al. '33:;.

>arallel to <nglish will $ a verb meaning want becomes a future marker inBulgarian$ 6umanian and 1erbo+8roatian$ as well as in the Bantu languages of AfricaK

Mabiha$ myene and 1wahili 2Bybee and >agliuca '34$ 5eine et al. '33:;. >arallel to

<nglish "an from to know$ Baluchi 2%ndo+%ranian;$ Fanish 2*ermanic;$ Motu 2>apuaAustronesian;$ Mwera 2Bantu; and Nung 2,ibeto+Burman; use a verb meaning know for

the expression of ability 2Bybee$ >erkins and >agliuca '337;. ,ok >isin$ a creole

language of New *uinea$ uses !n 2from <nglish "an; for ability and also savi from the>ortuguese save he knows for ability. /atin L potere or possum to be able gives (rench

pouvoir and 1panish poder $ both meaning can as auxiliaries and power as nouns. ,hesewords parallel <nglish may ( and past tense might ; , which earlier meant have the physical

power to do something. Derbs or phrases indicating movement towards a goal2comparable to <nglish be going to; freuently become future markers around the world$

found in languages such as (rench and 1panish but also in languages spoken in Africa$

the Americas$ Asia and the >acific 2Bybee and >agliuca '34$ Bybee et al. '337;.

f course$ not all grammaticali&ation paths can be illustrated with <nglish

examples. ,here are also common developments that do not happen to occur in <nglish.(or instance$ a completive or perfect marker 2meaning have -ustC done; develops from a

verb meaning finish in Bantu languages$ as well as in languages as diverse as 8ocama

and ,ucano 2both Andean+<uatorial;$ Ioho 2Mon+Ihmer;$ Buli 2Malayo+>olynesian;$,em and <ngenni 2both Niger+8ongo;$ /ao 2Iam+,ai;$ 5aka and /ahu 2,ibeto+Burman;$8antonese and ,ok >isin 2Bybee et al. '337$ 5eine and 6eh '347;. %n addition$ the same

development from the verb finish has been recorded for American 1ign /anguage$

showing that grammaticali&ation takes place in signed languages the same way as it doesin spoken languages 2Jan&en '33@;.

(or several of these developments % have cited the creole language$ ,ok >isin$ avaiety of Melanesian >idgin <nglish$ which is now the official language of >apua New

*uinea. >idgin languages are originally trade or plantation languages which develop in

situations where speakers of several different languages must interact$ though they share

no common language. At first$ pidgins have no grammatical constructions or categories$ but as they are used in wider contexts and by more people more often$ they begin to

develop grammar. nce such languages come to be used by children as their first

language$ and thus are designated as creole languages$ the development of grammarflowers even more. ,he fact that the grammars of pidgin and creole languages are very

similar in form$ even among pidgins that developed in geographically distant places by

speakers of diverse languages$ has been taken by Bickerton '34' to be strong evidencefor innate language universals. 5owever$ studies of the way in which grammar develops

7



in such languages reveals that the process is the same as the grammaticali&ation process

in more established languages 26omaine '33@$ 1ankoff '33E;.

3. Paths of change and synchronic patterns

,he picture that emerges from the examination of these and the numerous otherdocumented cases of grammaticali&ation is that there are several highly constrained and

specifiable grammati"aliation paths which lead to the development of new grammatical

constructions. 1uch paths are universal in the sense that development along them occursindependently in unrelated languages. (or instance$ the two most common paths for the

development of future tense morphemes in the languages of the world are the following)

27; ,he Movement >ath

movement towards a goal intention future

2@; ,he Dolition >ath

volition or desire intention future

,he first path is exemplified by the development of be going to and the second by will .

New developments along such paths may begin at any time in a languages

history. %n any language we look at$ we find old constructions that are near the end of

such a path$ as well as new constructions that are -ust beginning their evolution and

constructions midway along. *rammar is constantly being created and lost along suchspecifiable and universal tra-ectories.

Fevelopment along the Movement >ath begins when a verb or phrase meaningmovement towards a goal comes to be used with a verb) as in They are going to Windsor

to see the King. At first$ the meaning is primarily spatial$ but a strong inference of

intention is also present. 2Why are they going to Windsor To see the King .; ,he intentionmeaning can become primary$ and from that$ one can infer future actions) *e's going to

(gonna) buy a house can state an intention or make a prediction about future actions.

1uch developments are slow and gradual$ and a grammatici&ing construction onsuch a path will span a portion of it at any given time. ,hus$ <nglish be going to in

1hakespeares time could express both the change of location sense and the intention

sense. %n Modern <nglish$ the intention sense is still present$ but the future sense is also possible$ with no intention or movement implied 2That tree is going to lose its leaves). As

a result of the gradualness of change$ and the fact that in any particular language a future

morpheme might be anywhere on one of these paths$ there is considerable cross+linguisticvariation in the meaning and range of use of a future morpheme at any particular

synchronic period.

@



8onsidering -ust synchronic states$ then$ it is extremely difficult to formulate

universals of tense$ or even to give a universal meaning to future that would be valid

across all languages. %ndeed in the '3@Es and '3Es it was common for linguists toexclaim that any attempt to find universals of grammatical meaning would be futile and

ethnocentric 28homsky '3@$ ?einreich '3:;. Now there are attempts to formulate the

innate universals of tense and aspect 21mith '33';$ but it is very difficult to find a smallset of features that accommodate all the cross+linguistic variation in the area of tense and

aspect.

8omparing grammatical categories across languages from only a synchronic

perspective is something like comparing an acorn to an oak tree) they appear to have

distinct and unrelated properties. nly when we observe these entities across the

temporal dimension do we see the relationship between them. 1imilarly with grammaticalcategories and constructions) new relationships are observable when we take into account

where particular grammatical constructions and categories come from and where they are

going.

,he examination of the grammaticali&ation process across many grammatical

domains and many different languages makes it clear that the true language universals areuniversals of change. At one level$ these universals can be stated as paths of change$ such

as those in 27; and 2@;. ,o understand grammar more fully$ however$ we must look behind

these paths of change to the actual mechanisms that cause change and then seek tounderstand these mechanisms in terms of more basic cognitive and interactive processes.

%f we are successful$ we will begin to understand how human language acuires grammar.

4. Conceptual sources for grammatical material

,he examples discussed in the preceding sections showed lexical items entering into

the grammaticali&ation process. ne of the ma-or cross+linguistic similarities noted in the previous section is that the same or very similar lexical meanings tend to grammaticali&e

in unrelated languages. f all the tens of thousands of words in a language$ only a small

set are candidates for participation in the grammaticali&ation process. Are there anygenerali&ations that could be made concerning the members of this set"

6esearchers in this area have made some interesting observations about the lexical

items that are candidates for grammaticali&ation. 5eine et al. '33' have observed that theterms in this set are largely culturally independent$ that is$ universal to human experience.

(urthermore$ they represent concrete and basic aspects of human relations with the

environment$ with a strong emphasis on the spatial environment$ including parts of thehuman body. ,hus we find terms for movement in space$ such as !come# and !go# in

future constructions$ postures such as !sit$# !stand# and !lie# in progressive constructions.

,he relationship in space between one ob-ect and another is freuently expressed in termsof a human body part#s relation to the rest of the body. ,hus the noun for !head# evolves

into a preposition meaning !on top of$# !top# or !on.# !Back# is used for !in back of#

2<nglish provides an example of this derivation;$ !face# for !in front of$# !buttock# or

!anus# for !under$# and !belly# or !stomach# for !in# 25eine et al. '33')'0+':';. %n a



survey of such relational terms in '0@ African languages$ 5eine et al. found that more

than three uarters of the terms whose etymology was known were derived from human

body parts. 1vorou '33:$ using a sample representative of all the language families of theworld$ also finds human body parts to be the most freuent sources of relational terms. 0

/ess concrete$ but nonetheless basic and culturally independent$ notions such as volition$

obligation$ and having knowledge or power also enter into the grammaticali&ation process.

,he relation between locational terms and abstract grammatical concepts has beenrecogni&ed for several decades. Anderson '3' proposes a theory of grammatical cases

2nominative$ accusative$ dative$ etc.; based on spatial relations. ,hus a relational term

meaning !towards# further develops to mean !to# whence it can become a dative marker 2 I

gave the boo! to John) or can even further develop into an accusative 2as in 1panish) +i a Juan !% saw John#;. r$ with a verb$ !to# can signal purpose and eventually generali&e to

an infinitive marker 25aspelmath '343;. ,hus even the most abstract of grammatical

notions can be traced back to a very concrete$ often physical or locational concept

involving the movement and orientation of the human body in space.

Another important observation about the lexical items found in grammaticali&ingconstructions is that they are themselves already highly generali&ed in meaning. ,hus

among motion verbs$ go and come are the most general in meaning$ incorporating only

movement and directionality and not manner 2that is$ the more specific$ saunter$ waddleor run do not grammaticali&e$ though in some cases the most basic form of human

locomotion walk does grammaticali&e;. Among stative verbs$ it is be and have that

grammaticali&e$ and for active verbs$ the most generali&ed$ do 2Bybee et al. '337;.

,he claim here is not that the abstract concepts are forever linked to the more

concrete$ only that they have their diachronic source in the very concrete physical

experience. *rammatical constructions and the concepts they represent becomeemancipated from the concrete 2see section .'; and come to express purely abstract

notions$ such as tense$ case relations$ definiteness$ etc. %t is important to note$ however$

that the sources for grammar are concepts and words drawn from the most concrete and basic aspects of human experience.

5. Grammaticalization as automatization

1ome recent studies of grammatici&ation have emphasi&ed the point that

grammatici&ation is the process of automati&ation of freuently+occurring seuences of

linguistic elements 2Boyland '33$ 5aiman '337$ Bybee$ to appear;. Boyland '33 points out that the changes in form that occur in the grammatici&ation process closely

resemble changes that occur as non+linguistic skills are practiced and become

automati&ed. ?ith repetition$ seuences of units that were previously independent cometo be processed as a single unit or chunk. ,his repackaging has two conseuences) the

identity of the component units is gradually lost$ and the whole chunk begins to reduce in

form. ,hese basic principles of automati&ation apply to all kinds of motor activities)

0 ,he other freuent sources for relational terms are the body parts of livestock and landmarks.



playing a musical instrument$ playing a sport$ stirring pancake batter. ,hey also apply to

grammatici&ation. A phrase such as 2 I'm) going to ( +- ) which has been freuently used

over the last couple of centuries$ has been repackaged as a single processing unit. ,heidentity of the component parts is lost 2children are often surprised to see that gonna is

actually spelled going to;$ and the form is substantially reduced. ,he same applies to all

cases of grammatici&ation.:

%t follows then that grammatical constructions of all types are automati&ed motor

routines and subroutines that can be strung together or embedded in one another to produce fluent speech. ,his conclusion$ arrived at from the study of linguistic data$ is

similar to the proposal of Iimura '33$ '33:$ who argues on the basis of

neuropsychological data for a strong association between grammar and motor skill 2see

also /ieberman '347;. 5owever$ grammar is not -ust motor activity$ but motor activityappropriate to$ and meaningful in$ specific contexts. ,hus it is important to pursue the

uestion of how motor activities and meaning associate to make grammar.

5aiman '337$ '334 notes that the chunking and reduction features of thegrammatici&ation process bear a resemblance to non+linguistic rituali&ed behavior$ citing

rituals in both human and non+human species which show chunking and reduction inform. %n addition$ 5aiman cites two other characteristics of rituali&ed behavior that apply

to grammatici&ation. (irst$ repeated practices lead to habituation$ the process by which

an organism ceases to respond at the same level to a repeated stimulus. A repeated wordor phrase tends to lose much of its semantic force 2consider the loss of the power of the /$word when it is used very freuently;. ,hus habituation helps to bring about the

generali&ation or bleaching of semantic content that occurs in grammatici&ation. 1econd$

repeated practices can also change their function$ through the process of eman"ipation, by which the original instrumental function of the practice takes on a symbolic function

inferred from the context in which it occurs. ,hese two processes and other related

processes are crucial to the understanding of how grammatical meaning develops.

6. mancipation and ha!ituation in the creation of grammatical meaning

,he phrase grammatical meaning refers to the type of meaning conveyed by

grammatical morphemes and grammatical constructions. ,his type of meaning is often

contrasted with lexical meaning which is the meaning of nouns$ verbs and ad-ectives.

,he study of grammatici&ation makes it clear that there is no discrete cut+off point between the two types of meaning$ but rather a continuum from one to the other.

5owever$ we can still note the properties of the polar types. /exical meaning is specific

and concrete$ with nouns referring to physical entities and their parts$ and abstract notionsof cultural import. ,he lexical meaning of verbs describes perceived events and relations

among entities$ events that often have concrete physical results. ,he specificity of lexical

meaning is shown by the large number of contrasts that can be made$ i.e. in the number of names for species of trees 2oa!, elm, /ir, pine, willow$ etc.; or the number of designations

for ways to move through space 2wal!, swim, "limb, run, hop, trot, etc.;. ,he more

: Bybee$ >agliuca and >erkins '33' and Bybee$ >erkins and >agliuca '337 demonstrate for a large cross+

linguistic sample that a significant relationship between degree of grammatici&ation in semantic terms and

formal reduction obtains.

4



specific the meaning of a lexical item$ the more stable it remains across differing

contexts.

*rammatical meaning$ on the other hand$ is typically abstract$ referring to large$

abstract domains such as time or modality$ or referring to abstract grammatical relations

such as sub-ect of the verb$ or abstract relations among clauses$ such as although. %t isalso highly general$ being applicable to a large number of contexts. (or instance$ every

<nglish sentence has a grammatical designation of tense$ showing that tense is general

enough to apply to any verb in the language. %t is this type of meaning$ so typical ofhuman language$ that is responsible for the great versatility of language$ making it

applicable to any human communicative context. %t is also this type of meaning that is the

most difficult to describe or explain.

Another important difference between lexical and grammatical meaning concerns

the extent to which language users have conscious access to the meaning of units.

1peakers can often report directly and accurately on the meanings of nouns$ verbs and

ad-ectives$ much as they can report pieces of propositional or declarative knowledge2such as !?ashington$ F. 8. is the capital of the U1#;. 5owever$ grammatical meaning is

much less accessible$ and if speakers can report on uses of grammatical constructions$they often seem aware only of the most concrete of these uses. %n this way$ grammatical

knowledge resembles procedural knowledge or skilled ability 2Boyland '33;$ providing

further evidence that grammatical constructions are automated procedures.

,he approach that studies the way grammatical meaning evolves out of lexical

meaning has a great deal to contribute to the general understanding of grammar and its

meaning. 1ome of the mechanisms for semantic change in grammatici&ation have beenidentified and will be discussed briefly here.

.'. <mancipation

<mancipation in ritualistic language is extremely common. >olite expressions of

concern about a persons well+being in mutual greetings$ such as how are you$ reduce tosimple greetings that reuire no substantive response$ such as hi. %n some varieties of

Black <nglish hi is still answered with /ine, reflecting its source in a uestion$ but in most

dialects it is answered with hi. A string of words that originally had literal meaning or

instrumental function$ has lost its instrumental function and become a symbol for thesituation itself due to repetition in a particular contextKin this case the greeting situation.

,he change from a lexical to a grammatical function in grammatici&ation involvesa process that is uite parallel and could also be considered emancipation. As % mentioned

above$ in 1hakespeares <nglish be going to had its literal meaning of movement in space

towards some goal. 5owever$ given an apparent interest by human beings in goals and purposes$ even in 1hakespeares <nglish$ the information value of be going to was less

about movement in space and more about purpose. 8onsider example 2;)

2; 0u!e 1ir Dalentine$ whither away so fast"

3



+al. >lease it your grace$ there is a messenger

,hat stays in to bear my letters to my friends$

And % am going to deliver them.

2'@3@$ 1hakespeare$ Two 1entlemen o/ +erona %%%.i.@';

25opper and ,raugott '33:;

Note that even though the Fuke asks about movement 2O?here are you going so fast"O;$

what he really wants to know is Dalentines intention or purpose. Note also that althoughDalentine answers in terms of movement$ he also includes the appropriate information

about his intention.

,he freuent association of be going to with contexts in which the intentions ofthe sub-ect are also being revealed leads to its gradual emancipation from the earlier

meaning of movement in space. ,he new function of expressing a goal or intention also

gradually becomes the main function of the construction.

.0. >ragmatic inference

%n the grammatici&ation literature$ the mechanism of change in this example has

been called pragmatic inference 2,raugott '343$ 5opper and ,raugott '33:;. %t is widely

accepted that an important feature of the communication process is the ability to makeinferences) the speaker must be able to -udge which details the hearer can supply and

formulate his=her utterances accordingly$ and the hearer must fill in details not supplied

by the speaker. ,hus$ the hearer is constantly asking why is s=he asking me or telling me

this" %n the example above$ Dalentine knew that the Fukes uestion was not -ust aboutmovement in space but also about intention and answered appropriately. ?hen the same

pattern of inferences occurs freuently with a particular grammatical construction$ those

inferences can become part of the meaning of the construction. %f be going to isfreuently used to talk about intentions$ it begins to have intention as part of its meaning.

,he literature on grammatici&ation is full of such instances 2,raugott '343$ Bybee et al.

'337;.

,raugott 2'343$ '33@; has identified several important patterns of inferencing that

create semantic change in grammatici&ation and lead to the expression of more abstract

grammatical meaning. 1he argues that more concrete propositional 2lexical; meaning$describing external situations$ such as spatial movement or location$ is regularly inferred

to have meanings that describe internal 2evaluative$ perceptual or cognitive; situations$

such as intention. A second trend which she posits is that both external and internalmeanings can be reinterpreted as having meanings based in the textual situation$ such as

meanings that place the described situation in time before or after the moment of speech$

i.e. tense. A third trend infers the speaker#s sub-ective beliefs or attitudes from the othertwo types of meaning. ,he claim is that the abstract meanings of grammatical

constructions arise from common patterns of inference. ,he types of meanings that arise

in this way suggest that hearers are commonly working to infer as much as possible about

the relations of narrated descriptions to the current speech situation and to the speakers

'E



sub-ective evaluation of it. ,he content of grammar$ then$ is directly related to$ and arises

from$ the very act of communication. %t is not autonomous from the meanings it conveys

or the purposes it serves.

Note further that common paths of change$ such as those shown in 27; and 2@;$

would not be attested across languages unless users of these languages made very similarinferences under similar conditions. ,hat is$ the repetition across languages of the change

in meaning from !movement towards a goal# to !intention# is evidence that speakers in

different cultures tend to infer intentionsG similarly$ changes from temporal seuence 2as<nglish sin"e, originally meaning !after the time that#; to causation indicate that language

users are prone to infer causation.

.:. *enerali&ation or bleaching as habituation

Another important mechanism of change in grammatici&ation is related to

habituation. ,he earliest discussion of grammaticali&ation recogni&ed that grammatical

morphemes lose components of their original lexical meaning and are much more generaland abstract. (or instance$ will loses the volitional aspect of its meaning and be going to

loses the spatial movement components. ,his process has been called bleaching orgenerali&ation of meaning. ,he latter term is especially appropriate because the loss of

specificities of meaning makes a morpheme applicable in a more general range of

contexts. (or example$ if will does not signal volition$ it can be used with a wider rangeof sub-ects$ including inanimate ob-ects.

6epetition itself diminishes the force of a word$ phrase or construction. <xamples

are legion. Iterate doesnt seem to mean repeat uite strongly enough$ so we tend to addre$& with repetition the strength of that fades and we have to say reiterate again. 2ou guys

generali&es to include females and the word guy now can be used in collouial speech

even for inanimate ob-ects. %n grammatici&ation$ the generali&ation or bleaching of themeaning of a construction is caused by freuency$ but it also contributes to additional

freuency$ since a generali&ed construction can be used in more contexts$ leading to an

almost inexorable movement along a grammatici&ation path.

8onstructions that have been bleached of their more specific lexical meaning are

more likely to pick up inferential meaning from the context$ that is$ grammatical meaning

2Bybee et al. '337;. ,he (rench negative phrase ne ( +- ) pas originally contained both anegative element ne and the noun pas meaning step and was used only with motion

verbs$ with the sense not 2go; a step. Now pas has lost all of its independent meaning in

the construction and has absorbed the negative meaning from the construction. As ne isgradually being deleted$ pas becomes the sole negative marker.

.7. 8ategori&ation

An important feature of generali&ation is the expansion of contexts in which a

construction can occur. (or instance$ the (rench construction ne ( +- ) pas was originally

restricted to use with motion verbs. ,he verb slot in this construction gradually expanded

''



to include all verbs. ,he be going to construction in <nglish originally reuired human

sub-ects and active$ agentive verbs$ but now its use has expanded to all types of sub-ects

and verbs.

8onstructions always contain a variable slot 2otherwise we consider them set

phrases;$ and the variable slot is restricted to items of a certain category. ,hese categoriesare usually defined semantically$ with terms such as human$ volitional$ change of state

and so on. 1ome constructions reuire uite specific categories. (or instance$ the

construction typified by the phrase to wend one's way allows a verb of motion$ or a verbconstruable as describing movement or the creation of a path 2 swiggled his way, made

our way, "ut their way;. ,he position after the verb reuires a pronoun that is

coreferential with the sub-ect.

,he various positions in a construction$ then$ reuire categori&ation. ,hese

categories are based on the experience one has with the construction$ -ust as the

categories we create for classifying cultural and natural ob-ects are based on our

experience with the world 26osch '34$ /akoff '34;. /inguistic categories$ both those based on form and those based on meaning$ have a prototype structure. ,hey cannot be

defined adeuately as a series of necessary and sufficient conditions$ but rather must becharacteri&ed in terms of more central and more peripheral members 2see ,aylor$ '334;.

,he possibility of adding new peripheral members to a category allows productivity and

change. New items can be used in a construction if they are perceived as similar in someway to existing members. Accumulated change in membership will lead to change in

what is considered central and what is considered peripheral.

,he productive use of constructions$ or automated subroutines$ is what allowsspeakers to generate new utterances$ as speech consists of these routines concatenated

and embedded in one another. ,hrough practice ones fluency in manipulating and

stringing together constructions increases. /inguistic categori&ation determines theappropriateness of particular combinations of constructions and takes the same form as

categori&ation for non+linguistic percepts. Note that all the component processes that lead

to the development of new grammatical constructions come out of language use incontext and they involve cognitive skills and strategies that are also used in non+linguistic

tasks.

". #he grammaticalization of su!ordinate clauses

?ork on grammaticali&ation has not been restricted to the development of

grammatical morphemes$ but has also involved the larger structures within which thesemorphemes function. %n the late '3E#s *ivon noted that over time loosely -oined

structures or clauses tended to become more tightly fused resulting at times in new

subordinate clauses. An example studied by /ord '3 2see also 5eine et al. '33'$5opper and ,raugott '33:; shows the development of a complementi&er and complement

clauses in some related ?est African languages.

'0



%n <we$ with the main verb b3 !to say#$ two clauses can be loosely -oined as in

2;)

2; me$b3 me$w $e

%+say %+do+it !% said$ P% did itQ=% said that % did it#

?ith other verbs of saying$ however$ b3 occurs without the pronominal prefix and

functions more like a complementi&er)

24; me$gbl b3 me$w $e

%+say say %+do+it !% said that % did it#

(rom uses such as 24;$ b3 has been extended$ and it can be used as a complementi&er with

many different main verbs$ including verbs of cognition 2ny4 !know#;$ perception 2!p

!see#; and verbs of desire$ as in 23;)

23; me$d5 b3 m4ple awua dewoi%+want say %+1UBJUN8,+buy dress some

!% want to buy some dresses#

Not only has b3 lost its original meaning of !say# and taken on an abstract grammatical

function$ it has also produced a tighter syntactic structure$ as it introduces a subordinateclause.

%n <nglish it is no accident that the complementi&er that has the same phonological form as the demonstrative that $ as textual evidence points to the derivation

of the complementi&er from the demonstrative pronoun. 5opper and ,raugott '33:)'4@+

43 present examples from ld <nglish where the complementi&er that still shows somecharacteristics of being a pronoun and the structures it occurs in are more loosely

connected 2more hypotactic; than today. 8onsider the use in 2'E; where a fronted

demonstrative pronoun corresponds to the complementi&er later in the sentence)

2'E; Tht ge/remede 0iulius hiora "onsul, tht tht angin wearth tidli"e

F<M arranged Fiulius their consul$ 8M> F<M beginning was in+time

thurhtogen.

achieved

,he loose or hypotactic structure of this sentence is reflected in my translation in 2'';.,oday$ a subordinate structure$ such as 2'0;$ would be used.

2''; !,hat$ their consul Fiulius arranged$ that the beginning was on time.#

2'0; !,heir consul Fiulius arranged 2it; that it was started on time.#

':



Another type of <nglish complement structure that has developed within the

documented period is the to infinitive. ,he resemblance between the preposition to and

the infinitive marker is again no accident. 5aspelmath '343 demonstrates that it isextremely freuent cross+linguistically for a preposition meaning !to$ towards# to develop

into an infinitive marker. %n ld and Middle <nglish to could be used with an infinitive

2marked with a suffix; to indicate purpose$ and optionally after verbs such as !begin#$!cease# or !teach# but was otherwise not used with verbs 21weet '440;. %n Middle <nglish

to was used in purpose clauses but it had also expanded to clauses with main verbs of

desiring or ordering 2MossH '3@0;. Note that the infinitive suffix 6en still appeared on theverb)

2':; thanne wolde he ma!en hem to dryn!en

!then he wanted to make them drink#

,he use of to before an infinitive continued to expand to a variety of contexts as the

infinitive suffix was lost. ,he result is the loss of much of the prepositional meaning of to

in this context and the development of a new form of the verb for use in a variety ofsubordinate clauses. 21ee 5aspelmath '343 for more details about this development in

*erman and other languages.;

8ertain constructions in <nglish use the infinitive without to$ in particular the

modal auxiliaries 2will, shall, may, "an, would, "ould, etc.;. ,he difference betweenconstructions such as will go without to before the infinitive and want to go with the to

infinitive can be partly explained in terms of the period in which the constructions

developed and began grammaticali&ing. Already in ld <nglish$ the modern modalauxiliaries were used with infinitives$ which at that time were marked with a suffix.

Furing the Middle <nglish period such constructions increased in freuency and became

entrenched. ,he new infinitive marker to was also gradually expanding during the same period$ but because the MFA/ AUR%/%A6S T D<6BC construction was already formed and

entrenched without to$ the new marker has not affected these constructions. %n contrast$

the construction of want T %N(%N%,%D<C developed much later and so uses the infinitivemarker that was current when it developed. ther conservative features of the modal

auxiliary construction are discussed in Bybee 2to appear;.

$. Grammatical change in grammaticalization%

,he recent surge in research on grammaticali&ation has made it clear that the same

set of processes and mechanisms are responsible for all aspects of grammar. Allgrammatical morphemes have developed out of lexical morphemes$ principally nouns

and verbs$ and all grammatical structures have developed out of more loosely organi&ed

constituents. %n this section$ the processes of decategoriali&ation and reduction ofconstituent structure will be illustrated.

4.'. Fecategoriali&ation

Fecategoriali&ation is the term applied to the set of processes by which a noun or

verb loses its morphosyntactic properties in the process of becoming a grammatical

'7



element 25opper '33';. %n some cases$ the lexical item from which a grammatical

morpheme arose will remain in the language 2 go retains many lexical uses$ despite the

grammaticali&ation of be going to;$ and in other cases$ the lexical item disappears andonly the grammatical element remains 2"an is grammaticali&ed$ and the main verb from

which it developed$ "unnan !to know$# has disappeared;. %n both cases the

grammaticali&ing element ceases to behave like a regular noun or verb.

Derbs lose canonical verbal properties when they become auxiliaries. 8onsider the

auxiliary "an which derives from the ld <nglish main verb "unnan !to know#. %n ld<nglish$ "unnan could be used with a noun phrase ob-ect$ but today "an occurs only with

a verb complement) L I "an that and L I "an her are ungrammatical. ,he <nglish modal

auxiliaries have lost all their inflected or derived forms and are invariable. ,here is no

infinitive 7to "an$ no progressive or gerund form L"anning, and the past form of "an$which is "ould $ is developing non+past uses 2 I "ould do it tomorrow; and will perhaps

lose its function as the past of "an$ -ust as should no longer expresses the past of shall.

,he auxiliaries rarely modify one another. ?hile the use of shall "an was possible in

Middle <nglish$ such constructions have disappeared from Modern <nglish. %n otherwords$ "an has no main verb uses.

An example of a noun which has lost much of its categoriality is the con-uncion while

which was previously a noun meaning a length of time. ,oday it is very limited in its use as a

noun. ?hen it is clause+initial and functioning as a con-unction$ it has no noun properties. ,hus itdoes not take articles nor can it be modified as in 2'7;.

2'7; L% was there the same while you were.

%n other contexts$ its use as a noun is restricted to set phrases such as all the while, a long while.

%t cannot be freely used as a nounG thus 2'@; through 2'; are unacceptable.

2'@; L%ve been there many whiles.

2'; L% waited a boring while.

2'; L,he while was very long.

4.0. /oss of constituent structure

,he elements in constructions that are grammaticali&ing become more tightlyfused together and the internal constituent structure of the construction tends to reduce.

,hus two clauses become one$ two verb phrases become one$ and so on. A few illustrative

examples follow.

5eine et al. '33' report that in ,eso 2a Nilo+1aharan language of western Ienya

and eastern Uganda; the negative construction 2'4; derived from a construction with amain clause and subordinate clause$ as in 2'3;.

2'4; mam petero e$!oto e!i o!

not >eter :1*+want dog

'@



!>eter does not want a dog#

2'3; e$mam petero e$!tot e!i o!

:1* is not >eter 2who; :1*+want dog

!%t is not >eter who wants a dog#

,he sentence in 2'3; consists of the main verb 6mam which originally meant !not to be#

with >eter as its ob-ect$ and a relative clause modifying >eter. %n the current construction

as in 2'4;$ the verb is grammaticali&ed to a negative particle and negative sentenceconsists of one clause rather than two.

,he <nglish main plus complement clause construction exemplified in 20E; isundergoing grammaticali&ation which results in the loss of the main clause status for the

phrase I thin! 2,hompson and Mulac '33';.

20E; % think that we#re definitely moving towards being more technological.

8ommon uses of I thin! (that) and I guess (that) show that the actual assertion in

utterances like 20E; is contained in the complement clause and the introductory !mainclause# -ust gives an epistemic evaluation of how much confidence the speaker has in the

truth of the assertion. %t is with these two erstwhile main verbs that the omission of that

occurs most freuently$ as in 20';. Also these phrases can be used parenthetically$ inwhich case the former complement clause is the main clause in every respect$ as in 200;.

2<xamples from ,hompson and Mulac '33'.

20'; % think exercise is really beneficial$ to anybody.

200; %t#s -ust your point of view$ you know$ what you like to do in your spare time$ %

think.

,hus a complex clause consisting of a main verb and a complement clause has become a

single main clause modified by an epistemic expression.

%n other cases in which verbs grammaticali&e the result is the reduction of two

verb phrases to one. 8onsider the case of a verb becoming an auxiliary. As illustrated in

20:;$ as the original main verb becomes an auxiliary$ the embedded verb takes on mainverb status. %n any case$ the result is a single D> where formerly there were two.

20:;

1 &' 1

N> D> N> D>

D D> AUR D N>

'



>6* D D N>

1ally is going to see Jack 1ally !s gonna see Jack

Another interesting case of the reduction of two D>s to one occurs in languagesthat allow serial verb constructions. ,he following example from Soruba illustrates this

nicely 2*ivon '3@$ 5eine and 6eh '347$ 1tahlke '3E;. %n 207; there are two verbs

which each have direct ob-ects and approximately eual status)

207; mo /i 8d3 g3 igi

% took machete cut tree

,his can either be interpreted as !% took the machete and cut the tree# or$ since /i is

grammaticali&ing as an instrumental preposition$ it is more likely to be intrepreted as !%

cut the tree with the machete.# ,he fact that the serial verb construction has become a

single D> with the grammaticali&ation of /i is underscored by examples such as 20@;)

20@; mo /i g3 igi

% took=with cleverness cut tree

!% cut the tree cleverly#

Almost every case of grammaticali&ation involves such a change in constituent

structure. ?hen viewed in terms of a structural analysis of the successive synchronic

states$ it is tempting to say that a reanalysis has taken place. (or example$ in the twocases -ust examined$ what was a verb is reanaly&ed as an auxiliary in one case and a

preposition in the other. But it is important to note that even these reanalyses take place

gradually$ which means that when grammaticali&ation is occurring$ it may not be possibleto uniuely assign elements to particular grammatical categories or structures. 5eine'33: argues that the reason there is so much controversy surrounding the category of

auxiliary verb$ in that some linguists argue that they are verbs and others argue that they

are a separate category$ is that auxiliaries derive gradually from verbs and have notalways lost all their verbal properties even though they have become grammaticali&ed.

(. )mplications of grammaticalization

Now that the researchers mentioned in this chapter 2and others working on this

topic; have studied the grammaticali&ation process and its outcome from a cross+linguistic perspective as well as in language+specific detail$ we have a much clearer

picture of the nature of grammar. ?e have seen that grammatical constructions arise

through freuent repetition and their meanings change through processes ofgenerali&ation and pragmatic inferencing. *rammatical constructions are automated$

conventionali&ed units$ which % claim$ are also processing units. ,he meanings and

functions of constructions are not fixed and categorical$ but allow variation which leads

'



to gradual change over time. An essential factor in the development of grammatical

constructions is language use.

1ince all grammatical categories and constructions are derivable from experience

with language$ there is no reason to suppose that they are innate. %n fact$ the notion of

innate grammatical rules is incompatible with the gradual$ usage+driven nature ofgrammatical change. %nnate rules and categories would be unchangeable over time and

over generations$ or if change occurred$ an abrupt shift from one discrete category to

another would be reuired.7

1trong universals of grammaticali&ation give rise to similarities among languages.

Underlying these universals of change are cognitive and communicative universals. ,he

cognitive processes in grammaticali&ation discussed in this paper are)

2'; the ability to automate neuromotor seuences through repetition

20; the ability to categori&e recurrent linguistic elements

2:; the tendency to infer more than is actually said27; the tendency to habituate to repeated stimuli

ther mechanisms operative in the process of the creation of grammar may be identified

by further research$ perhaps by experimental means$ or by the further examination of

grammatical change in progress.

*eferences

Anderson$ John M. '3'. The grammar o/ "ase9 towards a lo"alist theory. /ondon)8ambridge University >ress.

Bickerton$ Ferek. '34'. -oots o/ language. Ann Arbor) Iaroma.

Boyland$ Joyce ,ang. '33. :orphosynta"ti" #hange in ;rogress9 < ;sy"holinguisti"

<pproa"h. Fissertation) University of 8alifornia at Berkeley.

Browman$ 8atherine >.$ and /ouis M. *oldstein. '330. Articulatory phonology) an

overview. ;honeti"a 73) '@@+'4E.

Bybee$ Joan /. ,o appear. Mechanisms of change in grammatici&ation) the role of

repetition.

Bybee$ J. /. and >agliuca$ ?. 2'34;. ,he evolution of future meaning. %n A. *. 6amat$

7 1ee /ightfoot#s '33 account of the development of the <nglish modal auxialiaries$ and >lank#s '347

response.

'4



. 8arruba and *. Bernini 2eds;$ >apers from the D%%th %nternational 8onference

on 5istorical /inguistics 2pp. 'E3+00;. Amsterdam) John Ben-amins.

Bybee$ J.$ >agliuca$ ?.$ and >erkins 6. 2'33';. Back to the future. %n ,raugott$ <.8. and

5eine$ B. 2eds;$ Approaches to *rammaticali&ation 2pp. '+@4;. Amsterdam) John

Ben-amins.

Bybee$ Joan$ 6evere >erkins$ and ?illiam >agliuca. '337. The volution o/ 1rammar9

Tense, <spe"t and :odality in the =anguages o/ the World. 8hicago) University of 8hicago >ress.

8homsky$ Noam. '3@. >ynta"ti" >tru"tures. ,he 5ague) Mouton.

*ivon$ ,almy. '3@. 1erial verbs and syntactic change) Niger+8ongo. %n 8harles N. /i$

2ed.;$ Word order and word order "hange$ 7+''0. Austin$ ,R) University of

,exas >ress.

5aiman$ John. '337. P6ituali&ation and the Fevelopment of /anguage. %n ?m. >agliuca

2ed.;$ ;erspe"tives on 1rammati"aliation. Amsterdam) John Ben-amins$ :+04.

5aiman$ John. '334. Tal! is "heap9 sar"asm, alienation, and the evolution o/ language.

New Sork) xford University >ress.

5aspelmath$ Martin. '343. (rom purposive to infinitive) A universal path of

grammatici&ation. ?olia =inguisti"a *istori"a 'E.04+:'E.

5eine$ Bernd. '33:. <u@iliaries. New Sork) xford University >ress.

5eine$ Bernd. '33. #ognitive /oundations o/ grammar . New Sork) xford University>ress.

5eine$ Bernd$ Ulrike 8laudi and (rederike 59nnemeyer. '33'. (rom cognition togrammar) evidence from African languages. %n <li&abeth ,raugott and Bernd

5eine 2eds.; <pproa"hes to 1rammati"aliation, +ol A$ '73+'4. Amsterdam)

John Ben-amins.

5eine$ Bernd$ ,om *9ldemann$ 8hrista Iilian+5at&$ Fonald A. /essau$ 5ein& 6oberg$

Mathias 1chladt and ,homas 1tol&. '33:. #on"eptual shi/t9 a le@i"on o/

grammati"aliation pro"esses in </ri"an languages. Afrikanistische Arbetispapier :7=:@. University of 8ologne.

5eine$ Bernd$ and Mechtild 6eh. '347. 1rammati"aliation and reanalysis in </ri"anlanguages. 5amburg) 5elmut Buske Derlag.

5opper$ >aul J. '33'. n some principles of grammatici&ation. %n <li&abeth ,raugott and

Bernd 5eine$ eds.$ <pproa"hes to 1rammati"aliation, +ol A$ '+:@. Amsterdam)

'3



John Ben-amins.

5opper$ >aul and <li&abeth ,raugott. '33:. *rammaticali&ation. 8ambridge) 8ambridgeUniversity >ress.

Jan&en$ ,erry. '33@. ,he polygrammaticali&ation of (%N%15 in A1/. M1$University of Manitoba.

Iimura$ Foreen. '33. Neuromotor mechanisms in the evolution of humancommunication. %n 1teklis$ 5. F. and 6aleigh$ M. J. 2eds.; Beurobiology o/ so"ial

"ommuni"ation in primates. 1an Fiego) Academic >ress.

Iimura$ Foreen. '33:. Beuromotor me"hanisms in human "ommuni"ation. xford)xford University >ress.

/akoff$ *eorge. '34. Women, /ire and dangerous things. 8hicago) University of

8hicago >ress.

/ehmann$ 8hristiane. '340. ,houghts on *rammaticali&ation. A programmatic 1ketch.Dol. % 2Arbeiten des Ilner Universalien+>ro-ekts 74;. Iln) UniversitVt &u Iln.

%nstitut f9r 1prachwissenschaft.

/ieberman$ >hillip. '347. iology and the evolution o/ language. 8ambridge$ Mass)

5arvard University >ress.

/ightfoot$ Favid. '33. ;rin"iples o/ dia"hroni" synta@. 8ambridge) 8ambridgeUniversity >ress.

/ord$ 8arol. '3. <vidence for syntactic reanalysis) from verb to complementi&er inIwa. %n 1anford B. 1teever$ 8arol A. ?alker$ and 1alikoko 1. Mufwene$ eds.$

;apers /rom the ;arasession on 0ia"hroni" >ynta@. 8hicago) 8hicago /inguistic

1ociety.

Meillet$ A. '3'0. '3@4C. /#Hvolution des formes grammaticales. %n /inguistiue

5istoriue et /inguistiue *HnHrale$ pp. ':E+74. >aris) 8hampion.

MossH$ (ernand. '3@0. < handboo! o/ :iddle nglish. Baltimore) ,he Johns 5opkins

>ress.

Mowrey$ 6ichard$ and ?illiam >agliuca. '33@. ,he reductive character of articulatory

evolution. -ivista di =inguisti"a 2';) :+'07.

>lank$ (rans. '347. ,he modals story retold. >tudies in =anguage 4.:E@+7.

6omaine$ 1usan. '33@. ,he grammaticali&ation of irrealis in ,ok >isin. %n Joan

0E



Bybee and 1u&anne (leischman 2eds.; :odality in grammar and dis"ourse$ :43+

70. Amsterdam) John Ben-amins.

6osch$ <. '34. >rinciples of categori&ation. %n <. 6osch and B. B. /loyd 2eds.;

#ognition and "ategoriation. 5illsdale$ NJ) <rlbaum.

1ankoff$ *illian. '33E. ,he grammaticali&ation of tense and aspect in ,ok

>isin and 1ranan. =anguage variation and "hange 0.03@+:'0.

1mith$ 8arlota 1. '33'. The parameter o/ aspe"t . Fordrecht and Boston) Iluwer

Academic >ublishers.

1tahlke$ 5. '3E. 1erial verbs. ><= '2';.E+33.

1vorou$ 1oteria. '33:. The grammar o/ spa"e. Amsterdam) John Ben-amins.

1weet$ '440.

,aylor$ John 6. '334. 1yntactic constructions as prototype categories. %n Michael,omasello 2ed.;$ The new psy"hology o/ language$ '+0E0. Mahwah$ NJ)

/awrence <rlbaum.

,hompson$ 1andra and Mulac. '33'. %n <li&abeth ,raugott and Bernd 5eine 2eds.;

<pproa"hes to 1rammati"aliation, +ol A$ '73+'4. Amsterdam) John Ben-amins.

,raugott$ <. 8. 2'343;. n the rise of epistemic meanings in <nglish) an example ofsub-ectification in semantic change. /anguage @) :'+@@.

,raugott$ <. 8. 2'33@;. 1ub-ectification in grammaticalisation. %n Feiter 1tein and 1usan?right 2eds.; >ubCe"tivity and subCe"tivisation9 =inguisti" perspe"tives.

8ambridge University >ress$ pp. :'+@7.

,raugott$ <. 8. to appear.

?einreich$ Uriel. '3:. n the 1emantic 1tructure of /anguage. %n Universals of

/anguage$ edited by Joseph *reenberg$''7+''. 8ambridge$ Mass.) M%, >ress.

(onte) http)==www.unm.edu=W-bybee=8ognitiveX0EprocessesX0EinX0Egrammaticali&ation.doc

Cognitive Processes in Grammaticalization - Bybee

Documents