Part 1:Questions and method

Chapter 1

Questions about language and

variation, and where we got them

QUESTIONS ABOUT LANGUAGE

1his book is about the study of language and the search for explanations of why
languages are spoken the way they are. This is a search thar is shared by all types of
linguistic rheory, but in rhe perspective taken in this book (which 1 will variously
refer to as variationist or Labovian linguistics), the goal is not the search for a 'uni-
versal grammar' (although there will be discussion of universals). Rather, the goal of
variationist linguisrics is to understand why language varieries become dif.ferent from
one another (or, in some cases, become more similar). 1he Chomskyan quesrion
'What underlies all human languages?' is thus not replaced, but other questions, ar
least as irnporranr, are added:

• If all human languages share a common universal grammar, why aren'r rhey
all rhe same?
• Why don'r all speakers of a language speak the same way al! rhe time?
• Whar forces produce rhis variation across languages? Are the forces universal?
• How do languages ger from one 'state' to another?
• What does the 'srate' of a language look like if it is changing? Is it differem
frorn a 'stable' one?
• Does variation have a function? Given that in some case ir impedes under-
standing, does variation serve a meaning funcrion?
1he goal of this book is ro enable irs readers ro explain whar our currem besr answers
to these questions are, and ro be prepared ro begin adding ro our knowledge of rhese
quesrions. 1he book is mosdy focused on theory, but since the merhods of vari-
ationist linguistics are closely tied ro the theory, a discussion of methods is essemial.
However, for students and instructors who wish ro have a full merhodological train-
ing, 1 would suggesr supplememing rhis text wirh a merhods book (such as Milray
and Gordon 2003) and a statistics texr (such as Tagliamome 2006, solely focused on
the Varbrul sratisrical method, or Baayen 2008 and ]ohnson 2008, which are good
general staristical inrroducrions for linguisrs).
In rhis firsr chaprer, we will be concerned wirh understanding rhe full scope of
the quesrions asked above. We will rake a hisrorical perspective, because rhe theories,
4 Linguistic Variation and Change

quesrions, and ideas about language discussed in the variationist field are not the
resulr of an ahisrorical process, but, even though by mosr accounts ir is a 'new' field,
are owed to the questions formulared decades and centuries ago. From rhe hisrorical
roots of the variationist programme come quesrions and assurnprions that reverber-
are (even if they are not subscribed rol to this day in variationist studies.

WHERE WE GOT THE QUESTIONS: FROM COMPARATIVE

PHILOLOGY TO VARIATIONIST THEORIES

The roots of rhe variationist programme reach far back, ro the beginning of modern
Western linguistics. This beginning may be uaced to observarions made by Sir
William J ones, who is often credited as the firsr European to remark, in 1786
(Seuren 1998: 79), that the sirnilarities among Sanskrit, Greek, and Latin suggest
that rhey somehow derived from a common so urce (alrhough rnost linguistic his-
rorians believe that these ideas were Boaring around before Jones made his famous
speech; see Weinreich et al. 1968). This insight led to the development of the
comparative method in historicallinguistics (see Seuren 1998), and the nineteenth
century showed a rapid development in the understanding of relarionships among
languages. In addition to the 'discovery' of rhe Indo-European language family, one
of the intellectual currents in European society about this time was a move rowards
rationalism and the discovery of 'natural' laws based on direct observations. This
current led ro a rapid leap in an understanding of the narural world; linguists of
rhe era were therefore inspired to approach language in a similar way. Comparative
hisrorical linguisrics, rhen, arose our of rhe same uends in science that produced
physics, chernístry, and evolurionary biology.
The most irnporranr innovation in linguisrics of this rime was the development
of a method for the comparison of languages. One of the first lessons learned in
hisroricallinguistics is that ir is easy ro find false cognares: words that look similar
in rwo languages but are nor relared, due ro a systematic change that has taken
place in rhe phonology of one or both of the languages. The comparative method
provided a way in which comparisons could be made berween languages in a more
systernaric and less error-prone manner. This research agenda, based on comparison
across languages, also suggesred the need for sysrems of rranscribing sound rhat
were not language-specific, so that such comparisons could be made; this need
gave rise to phoneric alphabets culminaring in the International Phonetic Alphabet
(IPA), which was derived from the work of phonetician Henry Sweet (Sweet 1911).
Another intellectual current arising especially in rhe eighteenth century was the
development of prescriptive grammars. Grammar until this time had mainly been
seen as something that was a property of Latin and Greek, and one learned grammar
by learning to uanslare ancient rexts in rhese languages. As shown by Mugglesrone
(2003), it was in rhe eighreenrh century that European scholars began to write
grammars and dictionaries for 'vulgar' languages such as German, French, and even
English.
All of rhese developmems changed the way people, and especially the developing
 Questions about language and variation 5

scientífic communiry (or 'natural philosophers' as they called thernselves), rhoughr

about language. For example, language needed to become an object that could
be studied, and ir needed to acquire regularities that one could write laws abour.
Biology, physics, and mathernatics were dominant fields in the burgeoning scientiíic
world, so ir is nor surprising that their study gave rise ro rwo ways of looking at lan-
guage. An early way of thinking about language was through a biological meraphor,
in which languages were seen as organisms. Another view held that language was the
'personal properry' of an individual. 1his argument was debated vigorously, with the
opposing view being that language is a communiry objecr. 1his argument continues
to this day, with generative linguistics in some ways taking the former view, and
variationists the latter.
In the late nineteenth century a group of young, mosdy German men had the
audaciry to argue against their teachers, and radically rethink rhe relationship oflan-
guage and the individual. 1hese junggrammatikers or Neogrammarians, as they have
become known, postulared that ir is nor individual words that change in a language,
and.at the same time rejected the biological meraphor. Rather, rhey argued, linguists
should be searching for regular and predictable laws of sound change, just as physi-
cists search for laws of nature. That is, languages change rhrough individual sounds
(or categories of sound) rather than word by word. Alrhough ir may seem obvious ro
modern linguisrs that change should affecr classes of sounds, rhe eogrammarians
were working without the benefit of the theoretical notion of the abstraer phoneme,
and in facr rheir work may have helped to spread the terrn more widely. 1he
Neogrammarians were (and still are) central to variationist work because rhey began
to look for 'laws' of sound change. 1hey began to describe changes in ways rhar
look similar to phonetic and phonological changes as they are described in modern
historicallinguistics.
1he first revolution in linguistic thinking in the rwentieth century (Seuren 1998:
157) was started by Saussure's Cours de la linguistique generale (1983 [1916]) and
especially by the notion of the phoneme. Phonemes are defined by their srructural
relationship to other phonemes in the linguisric sysrern, and nor by their phoneric
characrer. In this view, the significance of Ipl in 'pin' is not irs phoneric characrer
combined with the other phonetic characters of the word, but its difference from
orher phonemes such as Ibl or Id/, and this can easily be seen by the fact that replac-
ing Ipl wirh these other rwo consonams produces words with complerely different
meanings. A strucrural relarionship is one of difference that produces meaning,
crucially only within a particular linguistic system (language or variery). Although
the way we now discuss the phoneme is not identical to that described in the Cours,
the notion of a category of sounds has its beginning there.
While Saussure worked in Geneva, in America the field of anthropology was
beginning ro develop. Franz Boas (Boas 1911) saw the project of linguisrics as
part of anthropology and as revealing the folly of racism, challenging me view thar
American Indians were somehow inferior to Anglo-Americans. One of his tools
in this endeavour was to describe American Indian languages in a rigorous way.
1his concern with method can be traced in the present-day variationist focus on
6 Linguistic Variation and Change

merhod as a central pan of me research prograrnme. In addition, Boas srressed the

'unconscious' narure oflanguage, and aligned this characrer wirh 'culture', Indeed,
ir is central to Boas's argument rhar language should be aligned with culture, since
borh have caregories thar develop organically rarher rhan rhrough conscious rhoughr
and discussion. Boas's student Edward Sapir elaborared rhis notion. He developed
Boas's ideas and expanded the anthropologicallinguistics rradirion of descriprion
in his inrluenrial 1921 book, Language. This current became an irnportanr one that
in rhe larer rwentierh century influenced sociolinguisrics considerably (see Shuy
1990), borh because of its focus on descriprive merhods and because in this tradition
language was always seen as a cultural rarher rhan individual properry.
The rension berween language as an absrracr objecr and as the properry of an
individual, however, is srill presento In a central early programmaric anide on
variarionisr linguisrics which we will discuss extensively, Weinreich et al. (1968;
henceforrh WLH) spend a considerable amount of rime engaging wirh rhe rheories
of change ourlined by Hermann Paul (PauI1970, originally published in 1880, wirh
rhe fifrh edirion published in 1920). Paul was an exrremely influentiallinguisr of rhe
lare nineteenth and early twentieth centuries, whose view oflinguisric change resred
on a concepr of language as somerhing that belonged to an individual: he argued
that rhe 'idiolecr' is the proper objecr of linguisric srudy. For him, individuals are
fairly consistent in their language knowledge (or language 'image', a rranslation of
Paul's Vorstellungen); however, pronunciation randomly varies around a rarger. A
dialect (or language) for Paul is a collecrion - an average - of idiolects. Language
changes when idiolects change in parallel, and dialects splir from one another when
groups of idiolects diverge. These changes are explained by conract with orher dia-
lects and accommodation by speakers ro rhar new dialecr. WLH spend considerable
space refuring rhis view in favour of a view of a dialect as a communiry objecr. They
argue that Paul's view cannot account for the realiry oflanguage change, and indeed
can accounr only for phoneric change and nor for change in the case of lexical
replacement. There are other unanswerable questions, as wel!: why do speakers move
rheir idiolecrs in the same direction (in parallel)? If everyone is varying randomly,
why do they end up moving in rhe same direction? In addition, if change tends to
come from idiolects in contact, why are rhose idiolecrs different in the first place?
WLH delve deeply into many orher approaches to change, and we will not go
into al! of thern here. But rheir cririque of other rheories constitutes the real innova-
tion in the srudy of language considered in this book. The best-known passage from
the anide srares their central daim most eloquendy; it is worth quoting in full and
spending some pages on rhis passage [O understand rhe innovations in the srudy of
language change established by WLH (pp. 100-1):

The generative model for the descriprion of language as a homogeneous object

... is irself needlessly unrealistic and represents a backward srep from srructural
rheories capable of accommodating the facts of orderly heterogeneiry. Ir seems [O
us quite pointless [O construct a rheory of change which accepts as its input need-
lessly idealized and counterfacrual descriptions of language states. Long before
 Questions about language and variation 7

predicrive rheories of language change can be arrempted, it will be necessary ro

learn ro see language - whether from a diachronic or a synchronic vantage - as an
objecr possessing orderly hererogeneiry.
1he facrs of hererogeneiry have nor so far jibed well wirh rhe srrucrural
approach ro language ... For rhe more linguisrs became impressed wirh rhe
exisrence of srrucrure of language and rhe more rhey bolsrered rhis observaríon
wirh deducrive arguments abour rhe funcrional advantages of srrucrure, rhe more
mysrerious became rhe rransirion of a language from srare ro srare. Afrer al!, if a
language has ro be srrucrured in order ro funcrion efficienrly, how do people con-
rinue ro ralk while rhe language changes, rhar is, while ir passes rhough periods of
lessened sysremariciry? Alrernarively, if overriding pressures do force a language
ro change, and if communicarion is less efficient in rhe inrerirn (as would deduc-
rively follow from rhe rheory), why have such inefficiencies nor been observed in
pracrice?
1his, ir seems ro us, is rhe fundamental quesrion with which a rheory of
language change musr cope. 1he solurion, we will argue, lies in rhe direcrion of
breaking down rhe identihcation of srrucruredness wirh homogeneiry. 1he key
ro a rarional conceprion of language change - indeed of language irself - is rhe
possibiliry of describing orderly differenriation in a language serving a comrnu-
niry. We will argue rhar narivelike command of hererogeneous srrucrures is nor
a marrer of mulridialecralism or 'rnere' performance bur is pan of unilinguallin-
guisric comperence. One of rhe corollaries of our approach is rhar in a language
serving a complex (i.e., real) communiry, ir is absence of srrucrured hererogeneiry
rhar would be dysfuncrional. [ernphasis addedJ

1he focus of rhis passage is on rhe hererogeneous narure of language, as opposed

ro rhe homogeneous srares rhar mosr orher rheories of language up ro rhis point
assume. Underlying rhis focus is an undersranding of language as exisring in a
communiry of individuals, rarher rhan complerely and perfecrly in one of those
individuals. 1his view is embedded in the key srarement fram rhe second paragraph
above: '1he key ro a rarional conceprion of language change - indeed of language
irself - is rhe possibiliry of describing orderly differentiarion in a fanguage serving a
community.' 1he main difference berween WLH's view and that of previous (and
many current) rheories is rhar a language or language variery (dialect, etc.) is nor rhe
averaging of individuals, or even somerhing locarable in a single individual's mind,
bur rarher an external, communiry 'objecr'. In rhis view, individual variarion is nor
ro be averaged, bur rarher seen as rhe parrern or rexrure of rhe variery: variation is
pan of the very descriprion of a language.
WLH argue rhar 'the idenrificacion of srrucruredness with homogeneiry' is one
of rhe main conceptual problems prevenring linguistic rheories fram developing
an adequare account oflinguisric change (and as we will see, ir also fails to account
for many so-called synchronic linguisric faces). Really, there is no observarional or
scienrific reason ro assume rhar language is homogeneous (whether in a communiry
or in sorneone's mind). 1his view is rarher an ideology abour language rhar has
8 Linguistic Variation and Change

persisred rhroughour rhe hisrory ofWesrern linguistics, and it is relared to the ways
rhar language is assumed ro work more widely. In facr, as Anderson (2006) shows,
rhe idea of a homogeneous, national language was one of rhe notions thar helped
creare rhe modern nation-state and rhe very idea of 'a natiori' and 'a language' (see
also Agha 2007).
One of rhe mosr obvious examples of this ideology of homogeneiry is rhe focus
of srrucruralist (and to some extent generarive) phonology on exceptionless (what
1 will also call caregorical), predicrable rules or consrrainrs. Thar is, in describing
a language, rhe focus is on predicring caregorically rhe phoneric characrer of pho-
nemes. If rhey cannor be caregorically predicred, then any phonetic alternarion is
rhrown inro rhe 'free variatiori' dustbin, and ofren arrribured to issues of 'perform-
ance'. In this view rhen, if somerhing is not caregorically predicrable - Le., if rhe rule
does nor apply always - rhen ir is 'free' or wíthout arder.
So WLH inrroduced rhe rerm 'orderly hererogeneiry' as rhe key for unlocking rhe
problem. Orderly hererogeneity is essenrially rhe parrern or rexrure discussed above.
This terrn addresses rhe faer that earlier rheories had viewed any rype of hereroge-
neity as disorderly. Of course, we see hererogeneous order all rhe time in almosr
all human praerices. For example, when do you ger our of bed every morning? At
dawn? In faer, ir would silly ro suggesr rhar everyone gers up exaetly at dawn, and if
we were to eonsrruer a rheory of eircadian rhythrn, we wouldri't argue that everyone
has this as rhe neeessary ideal. But this doesri't mean that there isn't a predierable,
generalisable srruerure to rhe way a cornmuniry, a eulture, or an individual wakes
up eaeh day: In faer, researehers have found rhar rhere is an approximately four-
hour window in whieh mosr people wake, and rhe waking rimes of those outside
those parrerns ean be aecounred for in other ways (Horne and Ostberg 1976). The
poinr of rhis non-linguisric exarnple is ro show thar ir is nor hard ro think of human
knowledge and behaviour as being hererogeneous, or variable, bur srill possessing
predierable srruerure.

ORDERLY HETEROGENEITY AND CONSTRAINTS ON ITS

FORM

So whar rhen is 'orderly hererogeneiry' in language? Ir refers ro rhe fact rhar speakers
of a language have many ehoiees in how ro assemble any urreranee (rhar's rhe het-
erogeneity pan), bur rhar rhese ehoices form predictable parrerns in rerms of the lin-
guistie sysrem and social facrors (that's the orderliness part}. This is rhe basie insight
from which all the research discussed in rhis book flows. Ler's look ar this difference
with an example. One of rhe earliesr and besr-known variation srudies is Labov's
(1966) invesrigarion into linguistic change in New York City, whieh we will discuss
in deraillarer on. As an example here we can focus on his study of posrvocalie Ir!
(the pronunciarion of Irl afrer vowels in words such asfloor or flurth). In New York
rhe Ir! is nor always pronounced, unlike in mosr other varieries of North American
English. Labov srudied wherher rhere was a parrern in who pronounced the Ir! and
when. Pan ofhis merhod consisred of going to three differenr department srores and
 Questions about language and variation 9

asking where so me itern was located in the store, knowing full well that the answer
was the fourth Hoor. He pretended not to hear the response, causing the clerk to
repeat 'fourrh Hoor', and thus got four producrions of /r!. Both the pronunciations
of individual speakers and pronunciations across speakers were variable: sometimes
people had a 'constricted' approxírnant [1] pronunciation (or 'r-ful'), and sornetimes
a vocalic pronunciation ('r-Iess'; in the case of fourth it would be [fo: ej). By looking
at the historical record and by comparing older and younger speakers, Labov de ter-
mined that the r-ful pronunciation was increasing in ew York.
We can use this example to see that there are rwo dimensions of orderly hetero-
geneiry. First, speakers behave differently (heterogeneiry), but which speakers and
which utterances are more r-ful will be statistically predictable (orderliness), In the
departrnenr store study, for example, Labov found that salespeople in all stores had
variable pronunciations, but that those who worked in the high-end store used the
r-ful pronunciation more than those in the other rwo stores. 5econd, pronunciation
will be affected by linguistic factors: in this case, the pronunciation of Ir/ will be
statistically predictable on the basis of some other linguistic fact, such as whether the
l i] is followed by a consonant in the syllable (this is orderliness). Labov invesrigared
rwo phonetic environrnents, and while there was variation in both fourth and floor,
the fourth pronunciations were more often vocalic than the floor examples. N Ot all
such effects were phoneric. When Labov had rhe salespeople repeat their response,
so that rhey were more 'empharíc' the second time, the utterances that were more
emphatic were more likely to have r-ful pronunciations. So Labov found heteroge-
neiry, or variation, both across individuals and within individuals, and this variation
was not random, but patterned in statistically predictable ways. Orderly heteroge-
neiry allows us to see a change such as in Ir/ nor as an abrupt change in which one
generation is all r-less and the next generation suddenly r-ful, but as a more gradual
and organised affair, and this remo ves much of its rnysrery.This reconceptualisation
also sers up new questions, such as whether there are universal patterns of variabiliry.
WLH articulate a ser of problems thar a theory of linguistic change must answer,
and one riddle that may be unanswerable:

l. 1he constraints problem, or that of 'the set of possible changes and possible
conditions for changes which can take place in a structure of a given rype'
(p. 101): In what linguistic and social conditions are certain changes likely
or unlikely? For the New York example, whar linguistic structural factors
might have led to the 'srrengthening' of Ir/? Were there vowel changes rhat
supported the shift?
2. 1he transition problem: What imervening stages 'can be observed, or must
be posited, berween any rwo forms of a language defined for a language
communiry at differem times'? 1he New York example tells us that there
needs to be some kind of variabiliry in the transition. We might also observe
phonetic gradiems such that some speakers exhibit very definite [1], so me
sound a little less so, and some cornpletely r-Iess. 1heoretically, we will want
to try to predict what these stages will look like for all changes, or perhaps
10 Linguistic Variation and Change

for differenr kinds of changes (or changes in phonetícs, phonology, rnorphol-

ogy, or syntax). Finally, rhe transition problem encompasses the regularity
question (also known as the Neogrammarian controversy): do changes move
regularlyas changes to phonemes (and, relatedly, by syntactic category), or do
they move at differenr rates from word to word (and for synrax, by individual
constructions) ?
3. 1he embedding problem: 'How are changes embedded in the matrix of
linguistic and exrralinguistic concomitanrs of the forms in question (what
other changes are associated with the given changes in a manner that cannot
be attributed to chance?)?' (p. 101). 1his is the problem on which the vast
majority of variation studies have focused. 1he New York city study provides
several examples of embedding. First, the observation that speakers in the
high-end departrnent store were more r-ful is an embedding of the change
to r-fulfulness in the social fabric of the ciry. In addition, rhe change is
embedded in the linguistic system in that Ir/ is more likely to be r-less when
followed by a consonanr. 1he theoretical goal in this area is thus to find regu-
larities such that we can predicr the ways that changes are likely to expand
through a community and through a language.
4. 1he evaluation problem: 'How can the observed changes be evaluated - in
rerrns of rheir effects upon linguistic structure, upon communicative effi-
ciency, and on the wide range of nonrepresenrational facrors involved in
speaking?' (p. 10 1). 1his problem is almost as much srudied as the ernbed-
ding problem and rhey are almost by necessity srudied rogerher. In New
York, we want to know what a shíft to /r!-fulness does to the structure of the
language (for example, in the way that vowels before /r! are pronounced), and
whether or not rhe shift leads to misundersrandings. More common is a focus
on the social evaluarion: is the newer pronunciation valued in some way or
viewed wirh disdain? Theorerically, we wanr to look for predictable, repeated
parterns of how such changes are usually evaluated in the cornrnuniry, and
wherher there is agreement or disagreemenr on how and even wherher to
evaluate the new and old ways of speaking.
5. 1he actuation riddle: How and why do changes begin? 1his is a riddle
because changes probably begin unobserved, making it difficulr ro identify
their start.

Once we take variability to be a basic design feature oflanguage, we then need a way
of theoretically, or at least heuristically, representing ir. In variationist approaches,
this representation has been accomplished rhrough the notion of the linguistic vari-
able. 1he traditional way of defining a linguistic variable is to see it as more than one
way of saying 'the same thing'. As we will see in the next chapter, the notion of 'the
same thing' is slippery. But the essential idea is that there is one, isolable linguistic
feature that carries meaning, such as the pronunciation of a phoneme like Ir/, and
the community has more than one way of representing it in language. 1he 'variable'
notion is discussed in detail in the next chapter.
 Questions about language and variation 11

Note that a view of language that takes ordered heeerogeneiry as a given does
noe rule out caeegoriciry; in fact rhe field assumes thar there are some pares of the
grammar that may be caeegorical. In describing a grammar with probabíliries, ir is
trivial ro describe a pan of that grammar that is caeegorical: it simply has a prob-
abiliry of one (acring 100 per cent of the rime). For example, at some point in the
future, New Yorkers may use a consonamal pronunciation for all of their Iris.
WLH end their essay with so me general principies for the srudy of language
change, which summarise rhe variarionist perspecrive and suggest issues irnportant
in variationist linguistics:

1. '[Language change) is not random drifr proceeding from inherenr variariori'

(p. 187). This is a reaction to an idea that Saussure suggested, and is still the
view of so me linguists (who argue that more radical change is always the
produce of language or dialect conract). However, there seems to be more
directionaliry in language change rhan would be prediceed if change were
random and unstructured.
2. 'lhe association berween strucrure and homogeneiry is an illusion' (pp.
187-8). As discussed above, we can find structure, or orderliness, in variabil-
iry, if we know how to look and how ro describe ir.
3. 'Nor all variabiliry and hererogeneiry in language strucrure involves change;
but all change involves hererogeneiry and variabiliry' (p. 188). In other words,
variabiliry is a 'natural', even srable state for language, but if language do es
undergo change, we are logically going ro see some variabiliry.
4. "Ihe generalizarion of a change is neirher uniform nor instantaneous' (p.
188). lhis is a corollary ro (3); changes happen fasrer in so me parrs of the
language and for some speakers, and rhere will be many speakers who show
variability in the transition.
5. 'Grammars of change are grammars of speech communities' (p. 188). lhis
seems simple, but it is irnportant and one of rhe most difficule principIes
for students to undersrand, because ir requires ehinking about our objecr of
study very diflerently. We cannot understand language change by assuming a
single, ideal individual who represems rhe only grammar during rhe change.
Rather, rhe grarnmars we represem are rhose of a speech communiry. Labov
described IrI in New York Ciry, nor in a particular New Yorker. Given the
variabiliry across individuals such a task would be problematic. Of course,
speakers do have linguistic knowledge, and the problem of how ro represent
variable knowledge is an irnportanr one.
G. 'Linguisric change is transrnitted within the communiry as a whole; ir is
nor confined to discrere sreps in the family' (p. 188).lhe word 'discrete' is
irnportant here, because rhere are panerns by generaeion. Bur there are no
big jumps from morher to son, for example, and the change is not restricted
ro families but affecrs - and is moved forwards by - the entire communiry.
7. 'Linguiseic and social facrors closely interrelate' (p. 188). Linguiseic change
do es not happen in every speaker in the same way, but rarher cerrain rypes
12 Linguistic Variation and Change

of speakers will adopt the change before others. The linguistic and the social
constraints (or factors) affecting a change are to so me extent independent (at
least theorerically), but they will interact. Thus, a1most everyone in New York
will show the effect of a following consonant (everyone's rate of Ir/-Iessness
before consonants will be greater), but their overall rate of Ir/-fulness will
vary as well.

In addition, I will include a third rype of factor. These are cognirive factors, and
they constrain what speakers cognitively perceive and produce. Thar is, how are such
perceprion and production cognirively organized in the individual? I include this
factor in order to bring research on individual 'abilities' into our discussion. One of
the central problems in variationist linguisrics is to determine exacdy how different
factors relate to one another. For example, do es the following environment for Ir! in
New York work the same for everyone? Why does the following environrnent have
rhar effecr and how mighr ir be 'overridden' (for example, if you poinr the partern
out to someone, can they consisrently use one variant or another)?
A11 three of the major constraints (structural, social, and cognitive) acr on lan-
guage simultaneously, and ir is the tension among thern that ends up driving and
derermining the direction of changes. The problems and principies outlined by
WLH and expiained above focus the questions we ask, and the three main con-
straints will help to structure the answers. W<':, will begin with a more in-depth look
at the concepr of the variable.