Studying the human language faculty

If you meet someone at a cocktail party and tell them you are a carpenter, or a
veterinarian, or an astronomer, they are likely to be quite satisfied with that, and the
subsequent evolution of the conversation will depend, at least in part, on the depth of
their interest in woodworking, animals, or the universe. But if you tell them you are a
linguist, this is unlikely to satisfy whatever curiosity they may have about you: “Oh,
so how many languages can you speak?” is the most common reply at this point. But
in fact, many-probably even most-linguists actually speak few if any languages in
addition to their native tongue, in any practical sense. A “linguist,” at least in
academic disciplinary terms, is not a person who speaks many languages, but rather
someone concerned with the scientific study of language more generally.
That still doesn’t settle matters, though. As we will discuss below, different
generations of scholars have had rather different notions of what was important
enough about language to warrant study. Languages have histories, and relation ships
with one another that at least superficially parallel genetic connections, and one can
study those things. Most often, languages are spoken, and it is possible to study the
anatomical, acoustic, and perceptual aspects of speech. Different spoken forms can
mean different things, and we might study the kinds of things we can “mean” and the
ways differences in the forms of words are related to differences in their meanings.
Literature, rhetoric, and the texture of ordinary verbal interchange show us that we
can do many different things with words, and we might take an understanding of
these various potential uses as the goal of a scientific study of language.
All of these approaches to language, however, assume that language is an
essentially ubiquitous activity of human beings. As an infant, every human being
normally acquires a knowledge of (at least one) language, knowledge that is acquired
while exposed to a limited amount of speech in the child’s surroundings and that
allows him or her to participate in the verbal life of the community within a relatively
short time. Surely the most basic questions to ask if one wants to understand the
phenomenon of language concern the nature and form of that knowledge, the way it
arises and the way it relates to other aspects of human cognition.
 1 Studying the human language faculty

The study of language and cognition during the past several decades has given
increasing credibility to the view that human knowledge of natural language results
from - and is made possible by - a biologically determined capacity spe cific both to
this domain and to our species. An exploration of the functional properties of this
capacity is the basic program of the present book. These develop along a regular
maturational path, such that it seems appropriate to speak of our knowledge of our
language as “growing” rather than as “being learned.” As with the visual system,
much of the detailed structure that we find seems to be “wired in,” though interaction
with relevant experience is necessary to set the system in operation and to determine
some of its specific properties. We can refer to experience that plays such a role as
triggering the organization of the system, exploiting a term taken from ethology.
The path of development which we observe suggests that the growth of language
results from a specific innate capacity rather than emerging on a purely inductive
basis from observation of the language around us. The mature system incorporates
properties that could not have been learned from observation or any plausibly
available teaching. The deep similarity among the world’s languages also provides
support for the notion that they are the product of a common human faculty.
These fundamental, apparently native properties are shared by the gestural
languages which develop spontaneously in Deaf communities, quite indepen dently of
one another or of the language of the surrounding hearing community. We must
conclude that they are neither the result of simple shared history nor necessary
consequences of the articulatory/acoustic/auditory modality of language in its most
familiar form, spoken language. The development of structurally deficient pidgins
into the essentially normal linguistic systems found in creoles, as a result of
transmission through the natural language learning process in new generations of
children, provides additional evidence for the richness of that process.
The domain-specificity of the language faculty is supported by the many
dissociations that can be observed between control of language structure and other
cognitive functions. Focal brain lesions can result in quite specific lan guage
impairments in the presence of otherwise normal cognitive abilities; and vice versa.
Natural as well as acquired disorders of language also support the proposal that the
human language faculty is a product of our genetically determined biological nature:
there is evidence that certain language deficits show a clear distribution within
families, patterns that epidemiological and other studies show to be just what would
be predicted of relatively simple heritable traits (Gopnik and Crago 1991, Tomblin
1997).
Finally, the species-specificity of the human language faculty is supported by the
very fact that (absent severe pathology) every human child exposed in even
 1.1 Linguistics and the mind/brain
limited ways to the triggering experience of linguistic data develops a full, rich
capacity which is usually more or less homogeneous with that of the surrounding
community. Meanwhile, efforts to teach human languages to individuals of other
species, even those closest to us, have uniformly failed. While a certain capacity for
arbitrary symbolic reference can be elicited in some higher apes (and perhaps even in
other animals, such as parrots), syntactic systems even remotely comparable to those
of human languages seem to be quite outside the capacity of non-humans, despite
intensive and highly directed training.
These considerations make it plausible that human language arises in biologi cally
based ways that are quite comparable to those directing other aspects of the structure
of the organism. The language organ, though, is not to be interpreted as having an
anatomical localization comparable to that of, say, the kidney. Our understanding of
the localization of cognitive function in brain tissue is much too fragmentary and
rudimentary. Certain cortical and subcortical areas can be shown to subserve
functions essential to language, in the sense that lesions in these regions disrupt
language functioning (sometimes in remarkably specific ways), but an inference from
this evidence to a claim that “language is located in Broca’s (and/or Wernicke’s)
area” is quite unwarranted. The linguistic capacity which develops naturally in every
normal human being appears to be best understood in functional and not anatomical
terms, at least for the time being. We will return to these issues of the physical basis
of linguistic knowledge in chapter 10; until then, let us take it as given that some such
physical basis must exist, and concentrate on the nature of linguistic capacities.

1.1 Linguistics and the mind/brain

The major puzzles in linguistics concern the fact that our biological nature makes it
possible to acquire a richly structured kind of knowledge, much of which is only
indirectly evidenced (if at all) in the data available to the learner. This is the
“poverty-of-stimulus” problem, which will occupy us in chapters below. It follows
that any sensible student of language would want to take as his or her basic agenda
the investigation of the details of this knowledge, and the foundation for its growth in
the individual. This understanding of what linguistics is all about, however, with the
focus on language organs, is actually a relatively recent one.
In the Introduction, we noted the problem of where linguistics departments fall in
the geography of modern academia. As the central concerns of the field have moved
away from historical issues and the study of languages for their own sake, and toward
a more general understanding of the phenomenon of language, the reaction has often
been to move linguistics out of the humanities and into the social sciences. This has
not been particularly comfortable, however,
 1 Studying the human language faculty
since the most important questions about the nature oflanguage do not really re spond
to the assumptions and methods of the disciplines concerned with group phenomena
that arise fundamentally as a part of social reality (anthropology, economics,
sociology, political science, etc.). The problem of relating group behavior to the
properties of the individual, and its appropriate resolution, was already prefigured
some time ago in an even broader form by Edward Sapir, a notable visionary in his
approach to language and to culture more generally.
In his classic article in the first number of the journal Psychiatry (Sapir 1938) and
in many other writings, Sapir urged that a true understanding of the nature and effects
of culture must necessarily be founded on an understanding of the individuals who
participate in culture and society: “In the psychological sense, culture is not the thing
that is given us. The culture of a group as a whole is not a true reality. What is given -
what we do start with - is the individual and his behavior.” 1 And the central term in
this understanding is the nature of the mind and personality of the individual, not an
external characterization of his actions and responses or some system that somehow
exists outside of any particular person.
Trained by Franz Boas as a cultural anthropologist, Sapir devoted most of his
professional life to the study of language and the development of the nascent
discipline of linguistics (see Anderson 1985, chap. 9 and Darnell 1990 for sketches of
his personal and professional life). For Boas, as for Sapir, language was a key to all
other understanding of cultural realities, since it is only through an appreciation of the
particularities of an individual’s language that we can hope to gain access to his
thoughts and conception of the world, both natural and social. Sapir, indeed, is widely
associated with the notion of “linguistic relativity,” according to which the structure
of an individual’s language not only reflects but even contributes to determining the
ways in which he construes his world. 2 Language thus occupies a central place among
the phenomena that can lead us to an understanding of culture; and it must follow that
the way to study language is in terms of the knowledge developed in individual
speakers, not in terms of such externalities as collections of recorded linguistic acts.
In the history of linguistics, Sapir is remembered especially as one who emphasized
the need to study what speakers know and believe (perhaps unconsciously) about their
language, not simply what they do when they speak.
In addition to his primary focus on linguistics, Sapir also wrote widely of more
general issues in the nature of society, culture, and personality. His Psychiatry piece
was far from isolated in the sympathy it showed with the project of psy chiatry and
psychoanalysis. This interest in psychiatric issues and approaches

1
Sapir 1994, p. 139. The quotation is from a recent reconstruction of Sapir’s lectures on these topics.

2
This is the so-called “Sapir-Whorf Hypothesis.”
 1.2 Linguistics as history
was certainly not isolated from his work as a linguist and anthropologist. On the
contrary, as reflected in the title of his article (“Why cultural anthropology needs the
psychiatrist”), he felt that the mode of understanding essayed by the psychiatrist was
really the only path to a true appreciation of cultural phe nomena, given the claim
above that the individual is the basic reality in this study.
Why the psychiatrist, in particular? One must remember that the 1930s, when Sapir
was active in general studies of personality and culture, was the time when ideas from
the positivist and behaviorist traditions predominated in scientific investigations of
psychological questions, and of course these were precisely antagonistic to a
sympathetic investigation of the nature and contents of the mind and personality.
Psychiatry, in contrast, was centrally occupied with exactly this, and so the
fundamental place of the individual in language and culture entailed a need for the
kind of light that could only be shed on core issues by psychiatric investigation and
understanding.
While no one denied Sapir’s stunning brilliance as a linguist, both as a theoretician
and as an analyst, many of his colleagues at the time considered this “mentalist”
aspect of his thought to be an eccentricity - even an aberration - something to be
excused rather than imitated. After all, linguistics was on its way to attaining genuine
status as a science precisely through adopting the be haviorism of the day, focusing on
purely mechanical methods for collecting and arranging linguistic data so as to arrive
at a purely external analysis of linguistic behavior, eschewing all metaphysical talk
about “minds” and such-like unobservables (cf. Bloomfield 1933). Over time,
however, the field of linguistics has arrived at essentially the same conclusion Sapir
did, by its own path and making only a little use of the insight he had to offer.

1.2 Linguistics as history

While the study of language up through about the 1870s was as much a branch of
philosophy as an empirical science, scholars began to adopt genuinely scientific
methods around that time. Linguistics grew, in part, out of the study of philology:
how do we establish the correct form of an ancient text, given that we have perhaps
only several corrupted manuscripts to go on? A method developed for comparing
manuscripts to establish the most likely ancestral form from which they all derived. 3
When nineteenth-century scholars came to the conclusion that whole languages,
not just individual texts, could be regarded as related in this way, the comparative method
came to be applied to the problem of how to

3
The connection between philological methods in dealing with ancient manuscripts and the Stamm- baum or
“tree” model of inter-linguistic relationships is discussed by Hoenigswald 1963, who suggests that the
method of reconstructing the geneology of manuscripts developed by F. Ritschl may have influenced the
conceptions of his student A. Schleicher.
 1 Studying the human language faculty
reconstruct an ancestral language (e.g., Proto-Indo-European) from which a number
of attested modern languages could be derived. In the case of texts, it is clear how an
original form comes to be corrupted: scribes make mistakes. But what is the analog in
the case of whole languages? Nineteenth-century linguists came up with the
conception of “sound change”: systematically over time within part of a speech
community, certain sounds change into others. Thus, voiced stops like b, d, g may be
uniformly replaced by the corresponding voiceless stops p, t, k, for example (see (1.2)
below). When we reconstruct an earlier form of language, what we are trying to do is
to undo the sequence of sound changes by which the modern language came about.
Where this became scientific was when people suggested that there was a gen eral
lawfulness to these changes, and proposed the principle of regularity of sound change. That
is, unlike the random pattern of scribal errors, sound changes operate in a uniform
way, and do not affect individual words idiosyn- cratically. Furthermore, scholars
hoped that it would be possible to derive the range of possible sound changes from a
study of the physics and physiology of speech, assuming that sound changes reflected
regularities that could ultimately be deduced from natural laws of these domains, and
not purely arbitrary, random events.
The point of these observations is that at this stage, the object of scientific inquiry
about language was pretty much defined by the scope of the available genuinely
explanatory principles, and this meant that linguistics was “about” the histories of
languages (and of individual words within those languages). The basic question of the
field could be put as: how do languages come to be as they are through ( diachronic)
change over time? As a result, historical linguistics was the only kind of linguistics
there was at the time. Towards the end of the nineteenth century, Hermann Paul
(1880, p. 20) was able to pontificate that “Es ist eingewendet, dass es noch eine ander
wissenschaftliche Betrachtung der Sprache gabe, als die geschichtliche. Ich muss das
in Abrede stellen” (It has been objected that there is another scientific view of
language possible besides the historical. I must contradict this).
Linguistics was thus resolutely about change, and for nineteenth-century lin-
guistics, what changed was a “language”: essentially, an inventory of words. The
systematicity of linguistic change arose from the nature of its basic mecha nism:
change of sounds, which thereby affected all words in which the relevant sounds (or
sound combinations) occurred. There was more to language than just words, of
course, but everything else was attributable either to a universal “logic” or to
individually variable “habits,” and these matters did not greatly interest the linguists
of the time. The job of the linguist, then, was to write the history of words, their
pronunciations, and their meanings. This understanding of what linguistics is about
still characterizes some parts of the field, though classical historical linguistics is a
relatively small subdiscipline today.
 1.2 Linguistics as history
Words are transmitted from one generation to the next, and they may change their
form in the course of that transmission. Latin pater “father” became padre, pere,
patre, pai, etc. in the modern Romance languages. One could characterize such
changes by writing “sound laws” such as the principle that a dental stop came to be
pronounced with vocal fold vibration (that is, [t] came to be pro nounced as [d])
between a vowel and a vocalic r at some point in the transition from Latin to Italian
and Spanish. The important observation which made it possible to envision a science
of this sort of thing was the fact that while such changes affected particular words,
they were in principle formulable in terms of phonetic environments alone, without
direct reference to the words themselves. Changes in the set of words, then, were the
consequence of the working of sound laws, and linguistics could potentially develop a
science of these that could hope to explain the development of languages over time.
In this view, languages are the basic objects of reality, entities existing in their own
right “out there,” waiting to be acquired by speakers. Linguists sought to determine
and quantify the degree of historical relatedness among sets of lan guages, and this
relatedness was expressed through tree diagrams or cladograms such as that in (1.1),
introduced by August Schleicher (1861-62).

(1.1) Proto-Indo-European

Aryan-Greek-Italo-Celtic Germano-Slavic

Greek-Italo-Celtic

Aryan Italo-Celtic Balto-Slavic

Indic Iranian Greek Albanian Italic Celtic Slavic Baltic Germanic

(After Schleicher1861-62; note that many of the details in this model would be contested by
scholars today.)

As has often been pointed out, the Schleicher-style cladograms, familiar from all
textbooks on historical linguistics, capture only what biologists call “homologies,”
similarities which result from common historical origins, and not “analogies,”
similarities which arise from common developments. This was recognized very early,
and nineteenth-century linguists supplemented the “tree- theory” picture with “wave-
theory” models representing common innovations shared by geographically
contiguous languages.
 1 Studying the human language faculty
These cladograms assume that languages are like species, as articulated by
Schleicher 1863, and that they derive from one another in some coherent historical or
evolutionary development. But we should realize some of the idealizations that are
involved. Trees of this kind idealize away from the fact that languages do not split
sharply at some specific point and suddenly emerge in their full individuality. The
splitting process is more gradual and is initiated by minor divergences. We might say
that the first change which affected, say, Latin and not any of the other languages, is
the bifurcation point, the point at which Latin suddenly splits away. That is not
enough, though: saying that French and Italian are both (separately) descended from
Latin glosses over the fact that they are descended from different forms of Latin, and
that “Latin” is a cover term for many different forms of speech. In fact, the analogy
between languages and species is not ultimately a very useful one, and the
conventional tree models of historical linguists would require vast elaboration to be
equivalent to modern biological cladograms based on the molecular genetics of
organisms.
Schleicher-style family tree models also do not provide a very insightful way of
talking about relatedness, since they deal only with historically derived similarities.
These models were developed to capture relatedness among lexicons, which were
taken to be pretty much unstructured inventories of words (see chapter 7 below). If
one goes beyond the mere word-stock, however, and envisions relatedness more
broadly than just in terms of similarities resulting from a common history, very
different relations would emerge.
If one thinks of grammars, individual systems of knowledge, language organs,
developed on the basis of exposure to some relevant linguistic experience and
emerging according to the requirements of the linguistic genotype ( universal grammar) as
children set predefined parameters in appropriate ways, one can then compare the
grammars of English speakers with those of German speakers, and ask whether those
grammars are more or less similar to each other than either are to the grammars of
Italian speakers. After all, the child acquiring a language has no access to the
historical sources of the linguistic forms in the input, and accordingly no bias toward
having linguistic ontogeny recapitulate phylogeny, in Ernst Hackel’s famous phrase.
German grammars are quite different from English grammars: they have the object
preceding rather than following the verb (Ich glaube dass Hans Maria liebt “I
believe that Hans loves Maria”); they require the finite verb to appear in second
position in main clauses (Im Dorfgibt es nicht zu viele Hunde “In the town there are
not many dogs”); they have very different word order possibilities within the verb
phrase. In fact it is quite possible - even likely - that English grammars might be more
similar to grammars with which there is less historical connection. From this
perspective, looking at the parameters in the current linguistic literature, English
grammars may be more similar to
 Nineteenth-century linguists focused on the surface forms of words, which are the products of human behavior, rather than on internal processes that un derlie and
shape that behavior. They thus dealt with ^-language rather than /-language in the terminology of Chomsky 1986. Not all aspects of language have satisfying
accounts in these terms, though, and it is often necessary to invoke underlying processes and abstract systems that are not manifest parts of the observable facts of ^-
language. This is true for such famous instances of regular sound change as Grimm’s Law, which affected many types of conso nants in a related way in a sort of
cycle (cf. (1.2a)); or the Great Vowel Shift in English (cf. (1.2b)), which changed all of the long vowels in another sort of cycle, raising all vowels one step and
making diphthongs of the highest vowels. Thus, [swe:t] “sweet” (pronounced with a vowel similar to that of modern skate) became modern English [swi:t]; [ti:m]
“time” (pronounced rather like modern team) became [taim]; [hu:s] “house” (with a vowel similar to that of modern loose) became [haus], etc.). Grimm’s Law and
the Great Vowel Shift affect many sounds at the same time, and represent changes in systems, not simply in inventories.

1.2 Linguistics as history

Italian than to German, and French grammars may be more similar to German than to
Spanish.4 There is no reason to believe that structural similarity should be even an
approximate function of historical relatedness - assuming that there is in fact a non-
circular notion of historical relatedness to be discovered.
Nonetheless, nineteenth-century linguists focused on languages, seen as inventories
of words composed from basic inventories of sounds, which could change over time.
Languages so conceived appeared to change in systematic ways which could be
formulated as regular correspondences, each the product of some combination of
sound changes that could be expressed as sound laws independent of the identities of
the individual words affected. By the end of the nineteenth century, linguists knew
that this was not the entire story: there were other regularities of language change
which could not be stated in purely phonetic terms, suggesting that at least in these
cases it was not the language (construed as an inventory) or its sounds that were
changing, but rather some kind of more abstract system. This was dealt with in a
terminological move: there were regularities of “sound change,” but there could be
other sorts of change that worked differently. These were called “analogical change,”
and were assumed to be governed by quite a different, more mysterious, kind of
regularity.

4
Linguists idealize and speak of French grammars, but a “French grammar” has much the same status as a
“French liver,” a convenient fiction. Individuals have livers and individuals have grammars. Grammars may
be similar to one another, and we may seek similarities between the grammars of “English speakers,”
whoever they may be exactly. In doing so, we must be prepared to find differences between the grammars of
a man in Houston and a woman in Leeds.
 1 Studying the human language faculty

(1.2) a. Grimm’s Law

voiced -

aspi voiceless
rate
d
fric

b. The Great English Vowel Shift

Because the historical linguists of this period were working with the external forms
of language, rather than with internal, underlying processes and abstract systems, it
makes sense that their attention was confined to phonological and morphological
aspects of language, and they paid little attention to change in syntactic systems. It
makes no sense to think of (sets of) sentences, products of behavior, as being
transmitted from one generation to another, because language acquisition is clearly
not a matter of learning sets of sentences. This limitation was not seen as a matter of
concern: they simply worked on what they felt they had useful tools to elucidate. The
debates of the time were primarily about the nature and causes of sound change.
A part of what made the study of sound change appealingly scientific was the fact
that its systematicities could be related to those of another domain: the physics and
physiology of speech. When we describe Proto-Indo-European /p,t,k/ as becoming
/f,9,%/ in the Germanic languages (e.g., prehistoric patr becoming modern English
father, tres becoming three, etc.), we can unify these by saying that one kind of
formational occlusion (complete blockage of the oral cavity followed by a rapid
release, accompanied by a brief burst of noisy airflow) is replaced by another
(incomplete blockage of airflow, allowing for noisy, turbulent flow during the period
of occlusion). With this unitary formulation in hand, we can hope to find reasons why
one sort of occlusion should be replaced by another, a rather more hopeful task than
that of finding reasons why one arbitrary set of symbols should be replaced by
another equally arbitrary one.
It is natural that the development of rigorous methods in historical linguistics more
or less coincided with the beginnings of serious investigation of the
 1.2 Linguistics as history
phonetics of speech using instrumental techniques that were precise and quantifiable
(if quite primitive by today’s standards). It was widely assumed that the terms of an
explanatory theory of speech production and perception would serve as the
underpinnings of an equally explanatory theory of sound change, and it is no accident
that among the important figures in the late-nineteenth- century development of
comparative Indo-European studies was Eduard Sievers, author of the basic text of
the period on the science of phonetics, Sievers 1881.
But in spite of the promising start provided by increasingly precise accounts of
speech, sound laws were clearly not reducible to measurable aspects of the physical
world. Grimm’s, Grassmann’s, and Verner’s Laws (three changes whose discovery
and formulation played particularly important roles in the development of the
doctrine of the regularity of sound change) were not general laws like Boyle’s Law:
they were rather particular contingent events that had taken place at a given point in
history, but were not universally valid. As such, they required a deeper explanation.
Changes were taken to be directional, as in biology. Rask (1818) held that languages
became simpler. Schleicher (1848) identified a progression from isolating to
agglutinating to inflectional types, although this was said to hold for preliterate
societies, whereas Rask’s drive to simplicity was relevant for literate societies.
Darwin (1874, pp. 88-92) thought that languages changed in the direction of having
shorter, “easier” forms.
There was widespread agreement that language change followed fixed de-
velopmental laws, and that there was a direction to change, but there was active
disagreement about what that direction was. By the end of the nineteenth century
there was an enormous body of work on sound correspondences among historically
related languages, and vast compendia of changes that had taken place, especially in
the Indo-European languages. There were, however, few ideas about why those
changes had taken place.
First, the notion that languages became simpler, more natural, easier to pro nounce,
etc. was quite circular: “simpler,” etc. was what languages change to, and there was
no independent definition in a framework dealing entirely with historical change.
Second, the idea that languages change toward greater simplicity (or whatever) gives
no account of why a given change takes place when it does - as opposed to the laws
of gravity, which apply to all objects at all times. To that extent, invoking notions of
directionality was no more lawlike than the laws of Grimm, Grassmann, and Verner,
which directionality was intended to explain.
There were attempts to breakout of this circularity, by invoking psychological
causes, but the psychology itself was implausible and unexplanatory. Grimm (1848)
attributed his consonant changes to the courage and pride associated with the German
race’s advance over all parts of Europe. Jespersen (1909) ascribed the shift in the
meanings of verbs like like, repent etc. (whereby Middle English
 1 Studying the human language faculty

sentences like pears like me became modern I like pears) to the greater interest taken
in persons than in things at one stage in the history of English. For other examples,
see Lightfoot 1999. Hardly anyone was satisfied that the result of such speculations
provided a genuine science of (the history of) language.

1.3 Linguistics as the study of ^-language

In the early years of the twentieth century, then, the data of linguistics consisted of an
inventory of sound changes occurring for no good reason and tending in no particular
direction. The historical approach had not brought a scientific, Newtonian-style
analysis of language, of the kind that had been hoped for, and there was no
predictability to the changes. One of the greatest figures in historical linguistics at the
time was Ferdinand de Saussure, whose revolutionary work on the vowel system of
Indo-European (de Saussure 1879) established much of the research agenda in
historical linguistics for the twentieth century; and it is something of an irony that he
was the first to argue forcefully that these historical issues were only a very small part
of the story, and that the science of language would have to provide an account of the
nature of the linguistic systems that we find today in their own terms, not just their
histories.
Saussure (de Saussure 1974) argued that when we look at language, we must
distinguish the system of a language (called “langue”) from concrete acts of speaking
or language use (“parole”). What is systematic about language is to be sought in an
analysis of langue, but for Saussure, the locus of linguistic change is not there, but
rather in usage, or parole. It follows that linguistic change itself cannot be the object
of a fundamental science of language.
Following Saussure’s lead, the historicist paradigm - the notion that there are
principles of history to be discovered, which would account for a language’s
development - was largely abandoned by the 1920s (see Lass 1980 for thorough
discussion). Indeed, there was a virulent anti-historicism in the writing of several
major figures of this period, including not only Saussure but also Franz Boas, Edward
Sapir, and Leonard Bloomfield. They worked to some extent on language change, but
abandoned historicism, and with it the earlier program of seeking to explain how it
was that languages come to be as they are.
Perceived problems included the circularity of the historical principles invoked,
and the kinds of psychological claims that had been made. Sapir (1929, p. 208) wrote
that the psychological interpretation of language change was “desirable and even
necessary,” but that the existing psychological explanations were unhelpful and “do
not immediately tie up with what we actually know about the historical behavior of
language.” Bloomfield (1933, p. 17) complained about the circularity of
psychologizing, saying that there was no independent evidence for the mental
processes other than the linguistic processes they were supposed to explain. The
historicist paradigm was not really
 1.3 Linguistics as the study of ^-language

refuted or shown to be wrong: rather, it was abandoned as yielding diminish ing

returns and not asking the truly fundamental questions about the nature of language.
Work on language change continued to flourish, but subsequent generations did not,
by and large, appeal to historicist explanations.
In chapter 8 below, we will ask what accounts of language history we can give if
we take a different, more contingent approach. We shall shift away from a study of
the results of behavior and toward a study of the states and properties of the
mind/brain that give rise to those products, and we will see that a rather deeper (if
quite different) understanding of change results. But at this point, let us continue to
trace the evolution of thought about what linguistics ought to concern itself with.
Saussure’s concern to re-orient linguistics from the study of parole to that of
langue gave rise to structural linguistics, or the study of linguistic systems: the
sounds and grammatical structures of, e.g., French, as opposed to the history of how
Latin originals became the individual words of Modern French. Saussure himself
argued that the right basis for such an understanding was to construe the linguistic
system as something with a reality that was primarily social: a set of conventions
among speakers of a given language to do things in a certain way. For him, langue
was a system existing outside of any individual, located rather in the speech
community that employs a particular code.
The nature of that code was that of a structured inventory of S I G N S , where each
sign consists of the unity of some expression with the semantic content which it
signifies (thus, “[kst] means ‘Felis catus’ ”). The structural value of any unit in the
system derives from the principle that each sign is distinct from every other sign:
thus, the value of cat is provided by the facts that in terms of expression, [kffit] is
different from [rat], [kot], [ksn], etc.; while in terms of content, a cat is distinct from
a lion, tiger, dog, etc. The system of signs (langue) is what makes communication
possible within a speech community; it constitutes the common knowledge of that
community, and is therefore independent of the particular properties of any individual
member of the community, or of any individual utterance that may be produced on a
particular occasion by a speaker of the language (an act of parole).
While Saussure’s conception of langue as a primarily social entity was not widely
adopted, his replacement of diachronic change by synchronic system as the object of
study in the field was. Structural linguistics evolved in various ways through the first
half of the century, and a consideration of those developments is well beyond the
scope of the present book (see Anderson 1985, among many other sources, for further
discussion).
With the rise of strongly empiricist, verificationist views in the philosophy of
science during this period, the study of language eventually became mired down in
the assumptions of behaviorist (or as Leonard Bloomfield, a major influence in this
development, preferred to put it, “mechanist”) psychology. On
 1 Studying the human language faculty

that view, scientific investigation can only concern itself with external observables.
Properties of mind (if indeed there are such) are intrinsically beyond the reach of
science. Bloomfield in particular (contrary to his subsequent reputa tion in some
circles) did not deny the existence of the mind, but maintained that its properties were
so far beyond the reach of (current) science as to make its investigation pointless.
What was consistently true through this period was that the primary object of
inquiry in the field was not something historical, but something present to
contemporary speakers of a given language. This was always something ex ternal, and
the basic question of the field had become: what are the properties of the sets of
sounds, words, sentences, etc. recorded in attested acts of speaking? Whether thought
of as deriving from social conventions or as the external responses corresponding to
particular stimuli, linguistic objects were consistently seen in this external mode. The
commitment to an ^-language view of what linguistics studies was thus axiomatic and
complete.

1.4 Linguistics as the study of /-language

The study of ^-language entails the study of languages as sets of externally observable
elements - sounds, morphological units, words, phrases, sentences, texts, etc. But there
is no reason to believe that such sets are actually coherent or well-defined objects of
inquiry. That is, there is no precisely definable notion of a “language,” such that one
can show in some non-circular way that a given sentence is “a sentence of English”
(see Chomsky 1986, Lightfoot 1999, and others). A given sentence might be used by a
particular speaker in Arkansas, but not by another speaker in Yorkshire. Does that
make it a sentence of English? Apparently the answer is “yes” for at least some
Arkansas speakers, but “no” for at least some Yorkshire speakers.
There is no general algorithm for characterizing the sentences of English, and there
is no reason to expect to find one. If different speakers of the “same” language speak
differently, there can be no general algorithm. Languages in this extensional sense are
conglomerations of the outputs of various grammars, where each grammar is
represented in the mind/brain of (at least) one individual speaker. It is not, of course,
necessary that people who consider themselves to speak the “same” language should
have different grammars, different language organs, but it is certainly not necessary
that their grammars be the same, either, and empirical evidence suggests considerable
variation within particular speech communities. Some of this variation, as it is
reflected in differences in output of the grammars concerned, may be identified by
speakers as reflecting different “languages,” but this is a sociological notion, not a
linguistic one. For example, speakers across a large part of the former Yugoslavia used
to consider that they spoke “Serbo-Croatian,” but as a result of changes in political
identity, some of
 1.4 Linguistics as the study of /-language
these came to be speakers of “Serbian,” others of “Croatian,” “Bosnian,” etc. - all of
this in the absence of any significant change in the grammars of the speakers involved.
The ^-language set of sentences in use in a speech community, then, is not a
coherent object of study because it corresponds to the outputs of a diverse range of
(possibly very different) grammars. Indeed, even the ^-language notion of the set of
sentences produced by a given speaker cannot serve this purpose. Ignoring the fact that
an individual speaker will typically control a number of (possibly very different)
styles, corresponding to different grammars, the actual set of sentences produced
and/or understood by a speaker is a completely accidental consequence of the range of
things s/he has had occasion to say or hear. Insofar as there are systematic properties
of this collection, they are systematicities either of non-linguistic experience or of the
underlying grammar, but not of the set of sentences itself. “Languages” in the ^-
language sense are not well-defined entities. They do not “exist” independent of
contingent, concrete acts of speaking; they do not “descend” from one another; and
there is no reason to believe that their study will reveal any interesting properties. They
are convenient fictions, like the rising and the setting of the sun. It follows that if we
want to develop a genuine science of language, this must be concerned instead with the
/-language notion of grammars, the properties of a person’s language organ.
In the 1950s, the intellectual coherence and scientific status of American
structuralist theories of (£-)language were not seriously in doubt. When an alternative
to the distribution-based theories of linguistic structure then in vogue 5 developed
around the work of Noam Chomsky, this did not initially challenge the notion that the
goal of linguistics was to provide rigorous and principled descriptions of the set of
possible utterances that constitute a particular language. As Chomsky worked out his
ideas, however, it became clear that the conceptual underpinnings of the field were in
principle incapable of providing a substantive account of language.
The central difficulty for existing theories was the problem of how lan guage could
be acquired. Chomsky (1959) presented a devastating critique of Skinnerian
behaviorism that centered on this issue, an assessment that eventually had far-reaching
effects in psychology, but whose immediate consequence was a shift of seismic
proportions in the expectations scholars had for linguistic theory. He observed that
when we consider language use (“verbal behavior” in Skinner’s terms), we must
recognize that
[t]he child who learns a language has in some sense constructed the grammar for him self on the
basis of his observation of sentences and non-sentences (i.e. corrections by the verbal community).
Study of the actual observed ability of a speaker to distinguish

5
Harris 1951 was perhaps the epitome of this approach.
 1 Studying the human language faculty

sentences from non-sentences, detect ambiguities, etc., apparently forces us to the con clusion that
this grammar is of an extremely complex and abstract character, and that the young child has
succeeded in carrying out what from the formal point of view, at least, seems to be a remarkable
type of theory construction. Furthermore, this task is accom plished in an astonishingly short time,
to a large extent independently of intelligence, and in a comparable way by all children. Any theory
of learning must cope with these facts. (Chomsky 1959, p. 57)

These remarks had a number of profound consequences. For one, they made it
clear that the generalized mechanisms of behaviorist learning theory, based on
attributing as little structure as possible to the organism in particular domains, was
quite incapable of dealing with the acquisition of human language. For another, they
brought the problem of learning to the forefront in the study of language: where
previously linguists had generally been content to characterize the language itself,
Chomsky made it clear that an understanding of language cannot proceed without
asking how a speaker’s ability to use it arises in development.
But for linguists, the most profound effect of these arguments was a shift in our
conception of the object of study in the field. Chomsky stressed that the basic
problem is not one of characterizing what people do: it is rather one of characterizing
what they know. The central reality of language is the fact that we call someone a
speaker of, say, Japanese, because of a certain kind of knowledge that he or she has.
If that is the case, linguists need to find a way to study the structure of this
knowledge, and while the things people say and do can constitute important evidence,
that is not all there is, or even the most important thing. This knowledge is what we
are calling a person’s language organ.
In this focus on the nature of language as a form of knowledge, an aspect of the
structure of the mind, linguists have thus returned to a conception much like Sapir’s
of the centrality of the individual to an understanding of linguistic and cultural
phenomena. The decline of behaviorist assumptions in the last several decades of the
twentieth century necessarily led to a much broader consensus about the need to
understand the mind in its own terms: if Bloomfield was indeed right that the mind
was beyond the reach of current science, the thing to do was to develop the relevant
science, not study something else.
Much of academic psychology still finds itself preoccupied with externalist issues,
and for one reason or another rejects the validity or utility of conceiving its object in
terms of the minds of individuals described at some appropriate level of abstraction.
The result has been the rise of cognitive science as a discipline whose goal is
precisely a science of the mind. Combining ideas from linguistics, computer science,
philosophy, anthropology, and cognitive psychology, this emerging field focuses
squarely on the nature of mental and cognitive life.
 1.4 Linguistics as the study of /-language
Linguists and linguistics have been important in these developments, in view of the
fact that it is in this sphere that the central role of the individual can be studied
particularly concretely and productively. Cognitive science thus plays for the
contemporary linguist the kind of role Sapir foresaw for psychiatry: one that makes it
possible to study phenomena that emerge from the individual by providing methods
for investigating the content of the individual’s mind and the structure of mental life.
If the insight of Sapir’s 1938 paper now appears in somewhat different form, its
validity has only been reinforced by subsequent developments in the study of
language.
The result of the developments surveyed above has been the rise of a con sensus
that linguistics really is (a branch of) cognitive science: the object of study, /-
language, is an aspect of the structure of the mind/brain, the language organ, defined
at some level of abstraction, which must be studied as such and whose study can lead
to further insights about the architecture of cognition. This view is certainly not a
universal one: some linguists, for a variety of reasons, continue to focus on external
objects and events rather than on the form of linguistic knowledge. It is also
important to note that many linguists who would assent in principle to the view we
have presented nonetheless continue to pursue in practice the study of particular
linguistic phenomena in ways that remain unchanged from their origins within an ^-
language conception of the discipline. In what follows, we will attempt to draw out
the differences in approach that follow from this shift of attention, and the relation
between such abstract, general considerations and the everyday practice of
linguistics.
Here we have sketched some aspects of the path by which the field of linguistics
has arrived in its own way at what were essentially Sapir’s conclusions. In
subsequent chapters, we will outline some of the basic properties and sub structures of
language organs, considering as we proceed how we have been able to discover them.