We can trace a quite obvious continuity
between the primate and human versions of the perceiving, acting, and emoting
minds discussed in previous chapters. Mechanisms revealed in animal experiments
are frequently found in our own brains, and vice versa. However, there is nothing
in the language-like behavior of chimps or other animals that remotely approaches
the language abilities of 3-year-old humans. Thus we have only ourselves to
observe in trying to understanding the development of our language competence
and the underlying cortical mechanisms. Recall that we have touched on issues
surrounding language several times. In Chapter 5 we looked at some speculations
on how language might have arisen during hominid evolution from a precursor
mimetic intelligence, and we also noted that the appearance of language made
possible a mythic intelligence that accelerated the pace of evolution by transmitting
culture and ideas between generations. And in Chapter 7 we briefly noted some
stages in language development in human children.
Clearly, we have some genes and
developmental routines lacking in chimps that ultimately facilitate the formation
of a set of "language organs" universal in all modern humans. These
organs effect a universal grammar or tree-like set of rules that can generate
many possible languages. All these languages use the technique of metaphor
to make varied descriptions of our world. What is not clear is how direct the
linkage is between our genes and the operations of grammar and syntax. Is the
wiring that underlies the distinction among subject, object, and verb categories
instructed by the same sort of genetically programmed molecular markers that
direct cells in the lateral geniculate body to send axons to the visual cortex?
Or does each developing human brain itself invent the wiring of subject, object,
and verb operations as a best solution to problems posed by the environment
of other humans? We address this issue below, after offering evidence that
humans do have an innate language capacity. The sections that follow then deal
with brain mechanisms of language that are revealed by lesion and imaging studies
and with the use of metaphors in constructing language.
The Language Instinct
One line of evidence for a universal
and innate human language capacity comes from the observation that every known
human culture has a language with complex grammar. None of these grammars is
simple, and the complexity of a culture's language seems quite independent
of the complexity of its social organization or technology. Different groups
of humans, isolating themselves from each other, grow different languages. The
inhabitants of the highlands of Papua New Guinea have been isolated from other
humans for 40,000 years and speak about 700--800 different native languages
within an area similar in size to Sweden. No single one of those languages
is spoken by more than 3 percent of the population. One can move 20 miles across
rain forest and cross the territories of three different xenophobic tribes
who, fearing each other, have very little contact. Their languages have fundamentally
different structures, rules of syntax, and grammar. The differences among these
languages are much greater than the differences among modern European languages.
A similar situation exists in West Africa, where more than 700 distinct languages
are encountered. The languages are most numerous in areas of greatest rainfall,
where each group can produce its own food and doesn't need to communicate with
outsiders to trade.
New languages can evolve within
a generation or two from the rapid mixing of existing tongues. The new languages
that arise because different groups need to communicate are called pidgins
and are very crude. But whenever a group that contributes to a pidgin begins
to adopt the pidgin itself as its native language, a creole develops, which
has all the complex features of a normal language. The creole is invented by
the children of the people who speak the pidgin. The fact that they aren't
exposed to a complete language and yet make up one is a strong argument for
a genetic blueprint for language-making ability. This blueprint gets carried
out, however, only in the appropriate milieu of other humans. Language is absent
in feral human children raised by wild animals, and in some severely abused
children deprived of normal human contact.
DESIGN NOTE: IMPORTANT POINT
When two adult populations who
speak different languages mix together and must communicate, a crude form of
a new language takes shape. Within a generation, children of these adult groups
are able to invent a new, hybrid language, sometimes with unique features not
shared by the parent languages.
Creoles, whether invented in South
America, Africa, or the Pacific islands, are remarkably convergent. They place
subject, verb, and object in that order; use singular and plural pronouns for
the first, second, and third person; and employ relative clauses, anterior
tense, and conditional mood. One of the best-documented examples of the evolution
of a creole is a study of language in Hawaii after its annexation by the United
States in 1898. Sugar planters
had imported workers from China, the Philippines, Japan, Korea, Portugal, and
Puerto Rico. The children of the original immigrants, who had maintained their
normal language and a learned pidgin, had the problem of trying to speak with
other children using this pidgin, which was an impoverished and inconsistent
version of a human language. They spontaneously expanded the pidgin into a
consistent and complex creole within a generation, by 1920. Researchers talking
with adults of various ages obtained snapshots of the stages in the pidgin-to-creole
transition, depending on the adults' birth year. Many features of Hawaiian
creole grammar, whose vocabulary is largely English, differ from both English
and the workers' native languages.
Not only is having a language universal;
so also is the fundamental design. Noam Chomsky is credited with uncovering
universal properties of natural human languages, and some accord him a position
in the history of ideas comparable to that of Darwin or Descartes. We
have an inherited neural capacity for generating a system of rules into which
words are plugged---a universal grammar underlying all languages. Every sentence,
according to Chomsky, has a deep structure, a pattern in the mind, that is
mapped onto surface structure, the actual spoken utterance, according to rules
or "transformations." The details of the transformations vary from
language to language, but all share certain abstract properties that constitute
the universal grammar.
All languages use nouns and verbs,
subjects and objects, cases and agreement, and vocabularies in the thousands
or tens of thousands of words. Almost all languages have the same basic word
order of subject and object. The 4000--6000 languages known today are sufficiently
alike that an extraterrestrial observer might consider them all one language.
A sentence in any spoken language can be diagrammed as an inverted modular
tree containing noun phrases, verb phrases, and prepositional phrases that
can be fitted inside each other. The technical jargon (N-bar, V-bar, X-bar)
used to describe sentence structure is fully as intricate as the vocabulary
used to describe visual pathways in the brain.
The Case for Language as an Evolutionary
Both Chomsky and the evolutionary
biologist Stephen Jay Gould deny that language is an evolutionary adaptation
but suggest, rather, that it is an accidental consequence of having a complex
cortex, or perhaps a by-product of some other faculty on which selection acted,
as in Calvin's throwing model (see Chapter 9). The psychologist Steven Pinker
and many other linguists argue, to the contrary, for the more plausible view
that we have a language instinct that has appeared as a consequence of natural
selection, just like our liver and our ears. In
addition to the universality of language and its design, the stereotypic development
of language in children of all cultures supports this view. In the uterus,
human embryos react to the melody, stress, and timing of the mother's native
speech, and immediately after birth they suck much harder when hearing their
native language than when hearing a foreign one. Children begin to babble during
their first year, and words appear after about 12 months. Inborn linguistic
mechanisms (such as simple word combinations like "more candy") take
off at about 18 months, are fully operating at about 3 years, and decline by
puberty. Over a relatively narrow window of 6 months to around the age of 2,
the entire grammar of a language appears.
The language instinct does not
require conventional spoken language for its expression. The fact that congenitally
deaf babies can babble in sign language and the fact that those who can neither
hear nor speak communicate in a distinctly human way make a convincing case
for the ability of humans to use abstract symbols independently of conventional
language as a means of communicating with each other. Manual babbling occurs
in deaf children who are exposed to signed languages from birth. The similarities
between manual and vocal babbling suggest that babbling is a product of an
amodal, brain-based language capacity. The
speech modality is not critical. Rather, babbling is tied to the abstract linguistic
structure of language and to an expressive capacity capable of processing different
types of signals (signed or spoken). One interesting report is of a deaf boy,
the son of deaf parents who signed incorrectly (using American Sign Language
learned after they were adults). The boy nevertheless appears to have learned
correct grammar on his own, a remarkable result consistent with an innate ability
to construct grammar. Even more
dramatic is the invention of a complex sign language by a group of more than
500 deaf children in a Nicaraguan school. This
is similar to the invention of creoles by children of parents who speak a more
rudimentary, pidgin language.
DESIGN NOTE: IMPORTANT POINT
The appearance of language in infants
follows a stereotyped time course that is similar across cultures. The ability
to manipulate symbols, which underlies language competence, is amodal: It can
be expressed by vocal or manual signs.
The Learning of Language
Saying that language is an innate
ability of humans, invented and expressed as human children grow up together,
is not the same thing as saying that there is genetic control over formation
of the detailed cortical modules, domains, or representations that make up
linguistic competence. The alternative model is that our developing brains
contain robust learning devices that permit each of us invent, independently,
very similar arrays of "language organs." This distinction brings
us back to the nature-nurture debate discussed in Chapter 7, a debate that
comes into sharpest focus as we try to understand the relationship between
genes and language.
Genetic Determinants of Language
If it could be shown that a mutant
gene influenced a detailed aspect of grammar, this might provide evidence that
the gene in question was necessary for some fairly detailed brain wiring. It
would not necessarily be a gene for grammar, but it presumably would be required
for the relevant neuronal circuits to form. Showing that a gene mutation has
an effect that is specific to language, sparing other abilities, would constitute
powerful evidence for a segregated and genetically specified language faculty.
It appeared for a while that just
such a mutation had been found, expressed in 16 of 30 individuals in a British
family over three generations. Though
they were able to memorize words and their meanings and to perform some grammar
operations, these individuals frequently failed to perform past tense operations
(such as generate -> generated) and pluralization
operations (cat -> cats). A number of studies have
now reported other families in which a form of developmental dysphasia (reflected
by inability to distinguish singular from plural, and tenses of regular, but
not irregular, verbs) is distributed in a way consistent with inheritance as
a single autosomal-dominant mutation. However,
subsequent studies of the British family members, as well as of other cases,
have revealed that the deficits are not restricted to specific aspects of language.
Rather, they cause low performance on a wide variety of language tests and
are correlated with lower verbal and nonverbal IQ and poor control of oral-facial
Another candidate for genetic specification
of grammar has been suggested by studies of abnormal development: the syndrome
known as specific language impairment, or SLI. The syndrome, which has a clear
genetic basis, delays language development and causes abnormal grammatical
morphology, but nonlinguistic intelligence can be normal. It appears now that
affected individuals are deficient in the processing of rapid temporal sequences
of auditory, visual, and motor systems. These
deficits are general, even though their most obvious effect is on language.
Thus far, no inherited language deficits have been shown to be independent
of more general brain developmental or processing mechanisms.
Rather than searching for genetic
links to specific language deficits, another approach would be to seek out
genetic defects that spare language as a "module of mind" but compromise
other aspects of cognition. This would suggest the presence of genes that permit
the development of a separate language competence even when other intelligences
are compromised. One such case is found in Williams's syndrome, a
genetic defect that occurs in about one of every 20,000 births. It involves
deletion of a region of chromosome 7 that
contains a gene for the protein elastin and a gene for an enzyme-modifying
protein (a protein kinase) whose mutation has been shown to correlate with
impaired visuospatial constructive cognition. Subjects have an elfin appearance,
a narrowed aortic valve, and an IQ of around 50. They are unable to perform
simple chores for themselves, have impoverished spatial abilities, and maintain
the animism (belief that all moving objects are alive) that is left behind
in normal development. Their language, however, is rich and fluent, and they
are very sociable.
Those with Williams's syndrome
are similar in general cognitive abilities to those with Down's syndrome, which
is caused by another chromosomal abnormality (trisomy of chromosome 21). Autopsies
and brain imaging show that the cerebral cortex has shrunk on both sides but
that the frontal lobes, medial temporal lobes, and neocerebellum are closer
to normal size. The neocerebellum is a thin layer of cells atop the older cerebellum
brain region. It evolved more recently, along with the prefrontal cortex. Humans
are the only animals with large versions of these two brain regions and the
only animals with language. Moreover, brain imaging studies show that the neocerebellum
is activated only when semantic reasoning is required. Perhaps this recently
evolved circuit underlying language and sociability is spared in Williams's
syndrome, and the gene deletion that marks the syndrome impairs an older set
of instructions for brain development, while leaving the newer language circuit
intact. It is interesting that in many autistic children who are antisocial
and poor at language, but good at spatial tasks, these areas are small compared
to their size in nondisabled children and those with syndrome.
DESIGN NOTE: IMPORTANT POINT
Observations on some language idiot
savants provide further support for a language faculty that can remain intact
in a brain that has developed abnormally. There is the case of the 29-year-old
subject with a nonverbal IQ of 65 who could not draw simple figures, add 2
and 2, or tie his shoes. He could speak 16 languages. His language abilities
were independent of his cognitive ability; he never mulled over the meaning
of passages and was not able to think about what he translated.
In spite of striking cases such
as language idiot savants and children with Williams's syndrome, there is still
no evidence that the language faculty is a wholly independent function. The
current view is that it is more likely to consist of refinements of systems
that coordinate nonlinguistic elements with each other. This suggestion is
reinforced by the brain localization work described below.
Language Development and Brain
We have already mentioned several
times, in Chapters 3 and 5, the two areas in the left frontal and temporal
lobes (Broca's and Wernicke's areas) that are associated, respectively, with
the production and comprehension of language. These are illustrated again in
Figure 11-1. If these locations had the same constancy from individual to individual
as do the primary sensory and motor cortices, this would provide an argument
for their strong genetic specification. Quite to the contrary, however, children
who have left focal lesions or who have undergone removal of the left hemisphere
can develop language functions "within the normal range" using the
right hemisphere, although deficits in syntactic and phonological tasks are
frequently found. Deficits that the lesions cause in the development of spatial
cognition do not show the same plasticity. Part of the explanation for this
may lie in the fact that spatial cognition is an ancient system, having evolved
throughout the vertebrate line, whereas the language system is much more recent.
Although this system is usually located in the left hemisphere, the underlying
genetic equipment permits other placements.
Some of the cortical areas implicated
in language functions. The external view of the left hemisphere shows Broca's
and Wernicke's areas, which are involved in the generation and comprehension
of speech, respectively. The right hemisphere has been displaced in the drawing
to show its inside surface with the anterior cingulate cortex, which is involved
in arousal and attentional control in speech, and the supplementary motor cortex,
which plays a role in speech initiation. Also shown is the location of the
corpus callosum, which connects corresponding areas of the two hemispheres.
Anatomical studies on other primate brains have shown rich reciprocal interconnections
between these areas (arrows), but the corresponding anatomy in humans is not
so well known.
Different languages can be put
in different places. A stroke in one part of the brain can knock out a native
language and leave intact languages learned later, or vice versa. Imaging of
normal bilingual brains shows that two languages, such as French and English,
can activate slightly different areas of cortex. The
first language learned by a child is normally tightly organized in terms of
nerve circuits in the left hemisphere, but later languages are more loosely
organized. This is why it often takes longer to find words in them.
The normal placement of language
in the left hemisphere usually involves both spoken and written language, but
there is the fascinating case of a woman whose right hemisphere controlled
writing. This was revealed because she underwent an operation for epilepsy
that cut the bundle of fibers (corpus callosum) connecting the left and right
hemispheres. Words that were presented to her right visual field (and thus
to the left hemisphere) could be spoken out loud but could not be written.
Words presented to her left visual field (right hemisphere) could not be spoken
but could be written down with her left hand. Perhaps
reading and writing, because they were invented much more recently than the
evolved capacity for spoken language, can more easily be wired up in the brain
wherever there are "spare areas."
A closer look at language development
in children reveals that the developing language system is not the same as
an immature version of the adult system. Massive reorganization occurs with
experience. Brain imaging and electrical recordings demonstrate that word recognition
tasks that at 13 months recruit both right and left cortexes condense by 20
months to activate mainly parietal and temporal regions of the left cortex.
Left hemisphere injury that causes delays and deficits in the perception and
production of fine-grained perceptual detail correlates with difficulties in
expressive language, perceptions, and storage of the many sounds and meanings
that are crucial for articulate speech. Right hemisphere injury that causes
difficulties with integrating details into a coherent whole has a greater impact
on word and sentence comprehension.
One theory is that children do
not begin with innate representation for language but rather have innate left-right
differences in the speed and style of processing, including a predisposition
for extraction of perceptual detail on the left and for integration across
inputs on the right. Under normal circumstances, the left temporal area recruits
left frontal cortex into an integrated system for the comprehension and generation
of speech. Early focal brain injury can delay or redirect this process.
Language Development as Invention
One argument for innate specification
of language mechanisms in the brain has been that the linguistic environment
faced by a human baby is such a disordered cacophony that rules for extracting
what is relevant must be preformed, not discovered. Against this argument,
it has recently been found that 8-month-old human infants who are exposed for
only 2 minutes to unbroken strings of nonsense syllables are able to detect
the difference between three-syllable sequences that appeared as a unit and
sequences that also appeared in their learning set but did so in random order.
This means that infants are using simple statistical procedures to discover
word boundaries in connected speech, at just the age when systematic evidence
of word recognition starts to appear. The
ability to recognize words after only a few exposures and to retain this knowledge
for a long time might be taken as evidence for a dedicated language-learning
mechanism, but it turns out that this capacity is also seen in other domains
than language learning, such as matching novel objects and sounds.
There appears to be an increasing
consensus that genetic determination, or innateness, applies to the basic principles
of brain organization, such as maturational schedules, input (sensory) and
output (motor) pathways, processing, and representation. These are the basis
of learning devices that generate higher cognitive functions, including language.
Phonetic or grammatical structures may have evolved to exploit particular natural
biases of the brain's perceptual and motor systems. This could make it appear
that our brains were especially adapted to acquire natural language, but in
fact it would be natural language that has had to change to fit the mechanisms
of our brains.
Children are not just passively
soaking up bits of language; they are actively constructing and inventing language
as the brain develops. The modern cortex, including the areas on the left side
that normally develop language functions, is multipotential in its early stages
of development. Functional specificity derives from the specificity of sensory
input and motor output, and linguistic knowledge and processing result from
the rich connectivity that develops in between. The
plausibility of these constructive models is enhanced by recent work on artificial
neural networks simulated in computers, showing that they can detect patterns
in input, extract phonetic and phonological structures from raw speech, and
extract grammatical regularities. Models
that grow by small and gradual changes show sudden vocabulary spurts of the
sort that are observed in young children. Lesions in artificial networks can
cause dissociations between, or differently influence, abilities to form regular
and irregular verbs. Such dissociations are indeed observed in patients with
Brain Mechanisms of Language
Normally, the language faculty
that each human child develops localizes the bulk of its operations in the
left cortex. Sign language, like spoken language, also normally emanates from
the left cerebral hemisphere, even though sign language relies on visuospatial
signals that are usually processed in the right hemisphere. The
lateralization of linguistic gesture (spoken or signed language) and nonlinguistic
gesture (pantomime) has been compared in deaf and hearing individuals and in
deaf signers with left hemisphere aphasia. The observations demonstrate a linguistic
specialization of the left hemisphere that is distinct from motor or symbolic
communication. A deaf signer with left hemisphere aphasia could spontaneously
substitute pantomime for signs. This differential disruption of the linguistic
gesture of signing and the symbolic gesture of pantomime emphasizes their functional
separability and reinforces the idea that the mimetic intelligence discussed
in Chapter 5 can be distinguished as an entity separable from language.
DESIGN NOTE: SELF-EXPERIMENT
You can demonstrate one consequence
of the lateralization of language functions to one hemisphere with the following
simple exercise. Ask a friend to start speaking, and then repeat what is said
while he or she is saying it---that is, "shadow" your friend's speech.
Now, at the same time, tap the third finger of your right hand in a regular
rhythm. Then try the same thing tapping your left finger instead. Is tapping
with the right finger a bit harder? For most people it is, because the right
finger competes with language for the resources of the left hemisphere. The
same thing happens for most deaf people when then shadow one-handed signs in
American Sign Language while tapping their fingers.
Multiple Language Areas in the
The brain location of specific
language functions is being studied in increasingly fine detail by the analysis
of discrete brain lesions that compromise some functions, and by the imaging
of normal living human brains while they are performing those same functions.
The results often are complementary. The number of "language areas" identified
is multiplying on an almost daily basis as the temporal and spatial resolution
of imaging techniques improve, just as is the case with other areas of perception
and cognition. You might look
back at the imaging experiment shown in Figure 3-8, which illustrates that
speaking words activates motor areas of the left frontal lobe whose damage
(as in Broca's aphasia) can interfere with speaking but not understanding.
Recognition of spoken words engages the temporal-parietal areas whose damage
(as in Wernicke's aphasia) can compromise understanding but not speaking. Visual
presentation of words recruits the visual areas mentioned earlier.
Frontal and temporal areas become
active when meaning must be attached to words, as in generating verbs to go
with nouns or in grouping together related words or concepts, and the effect
of practice can be monitored. An instruction to associate a verb with each
noun projected on a screen initially activates frontal lobes. But after 15
minutes of practice, the activation has contracted mainly to those areas used
in simply reading a word out loud. Not surprisingly, brain activation increases
with the complexity of a task being faced; complex sentences cause more activation
of Broca's and Wernicke's areas than simple sentences.
Different regions of the left temporal
lobe, outside the classical language areas, are involved in retrieving words
from our mental dictionary. Studies of patients with brain lesions and imaging
studies on unimpaired patients suggest that knowledge of unique persons is
associated with the anterior portion of the temporal lobe, knowledge of animals
with activity in the middle part of the inferior temporal cortex, and knowledge
of tools with the posterior part of the inferior temporal cortex (see Figure
11-2). Thus there appears to
be a partial segregation of the systems required to bring these different classes
of words or concepts into their spoken form.
Regions of the temporal lobe that
become active when information about persons, animals, or tools is being processed
and verbally expressed.
Each of these systems, at its deepest
level (the mental dictionary, or concept level), associates an object (such
as a dog) with its semantic features (furry, domestic, pet). The next level
assigns proper syntax (grammatical features such as gender), and a final level
matches the syntactic elements to the sounds required to make an utterance.
It is at this final stage that the "tip-of-the-tongue" phenomenon
can occur, when you can describe an object (furry, has four legs) but can't
quite say the word.
DESIGN NOTE: IMPORTANT POINT
Language relies on our current
perceptions and actions as well as on our memories of objects and actions,
and so it is not surprising that imaging experiments show that large areas
of cortex that process sensory information and control motor output are involved
not only in memory and learning but also in language. For
example, speaking color words can selectively activate a region in the ventral
temporal lobe just anterior to the area involved in the perception of color,
and speaking action words activates an area just anterior to visual areas involved
in motion perception, as well as parts of motor areas of the frontal cortex.
Language as Accessory to Other
Fundamental Brain Mechanisms
It seems reasonable to imagine
that subject-verb-object ensembles (as in "I throw rock") were categories
and actions that were chunked, localized, or given domains in the prelanguage
brain. Language would then have been a useful adaptation that could "piggyback" on
these fundamental operations by applying labels to what was already going on.
Recall the evidence already cited that language is some sort of accessory module.
Uniquely human capacities are found in the complete absence of language; in
abnormal cases such as brain lesions and congenital lack of hearing and speech;
and also in mime, tool making, and social intelligence (including the ability
to comprehend complex events and remember roles, customs, and appropriate behavior).
The conventional language system might be viewed as one kind of input and output
domain, parallel to a mainstream of thought and awareness, possibly derivative
of the basic perceiving, acting, and emotional intelligence we have discussed
in the last several chapters.
The implication is that knowledge
of objects that is accessible to language is stored in a distributed way, with
attributes of objects stored close to the regions of cortex that perceive those
attributes. At the level of
word meaning, language systems have to be coordinated with nonlinguistic brains
areas that represent object and perceptual meanings. Brain lesion studies show
that extensive systems in the left and right hemispheres process nonlanguage
interactions between the body and its environment, categorizing representation
(such as shape, color, sequence, and emotional state) and also more complicated,
symbolic representations of objects, events, and relationships. There seems
to be substantial overlap in the processing of the meanings of words and visual
images. A word, with all the
different kinds of information related to it, might be represented by a very
far-flung net of neurons that code the thing it stands for, its sound, its
syntax, its visual image, and its logic. Areas of the brain devoted to different
aspects of language, grammatical rules, or storage of word meanings might be
spread out in patterns, like the stripes or blobs of the visual cortex, but
with finer resolution than can be revealed by brain lesions or current brain
Studies on the classical aphasias
we mentioned before (sensory, motor, and conduction aphasia) have been expanded
to show just how discrete lesions can be, behaviorally and anatomically. The
use of different aspects of language, such as proper nouns, common nouns, and
irregular or regular verbs, can be specifically disrupted, which suggests that
they are processed in different areas of the brain. These
areas vary from person to person. The reading and/or writing of language can
be disrupted by brain damage (alexia and agraphia) without damage to the hearing
or speaking of language. The damage is usually in the parietal-temporal-occipital
association cortex concerned with integration of visual, auditory, and tactile
DESIGN NOTE: IMPORTANT POINT
An interesting point comes from
studies of reading and writing disturbances among the Japanese. There are two
distinct systems of writing Japanese. Kata kana is phonetic, with 71 graphemes
in the system. Kanji is ideographic, based on the Chinese character system,
with over 40,000 ideograms. Lesion of the angular gyrus of the parietal-temporal-occipital
association cortex can disrupt reading of kana but leave comprehension of kanji
The different functions and brain
areas listed here do not send their signals to a common destination for integration,
as though language appeared on a movie screen in the brain. There is no evidence
for a central place where it all comes together to form the "I" that
we subjectively experience when we listen and speak. Rather, language and other
aspects of our cognition, such as vision, are governed by some yet-to-be-defined
mechanisms that bind different brain areas together in time. How are our perception
and generation of language unified? Some interesting clues on how we normally
generate sentences come from normal slips of the tongue. In one rare form of
aphasia, jargon aphasia, patients seem entirely normal and have normal intelligence,
yet they utter sentences that are complete gibberish, like a "word salad." They
do not perceive any problem as they emit complete but unrelated phrases and
sentence fragments. It seems as though the underlying process that normally
orders the fragments into longer, more meaningful sequences has gone awry.
Speech errors such as Freudian slips (the unintended expression of a repressed
sentiment) and spoonerisms (the transposition of parts of two or more words)
point to how grammatical and lexical rules are working. They suggest that the
act of speaking involves the generation of many possible options that vie with
each other to be in a final utterance. Most of the options never reach awareness
but can be revealed, for example, if we happen to say something like "barn
door" when we mean "darn bore." The
idea is that there is a competition among many sequences, or candidates, held
in short-term unconscious memory, an interpretation not unlike the Darwin Machine
model for choosing movement sequences that we described in the last section
of Chapter 9. Most of these sequences are discarded after the winner---the
actual utterance---has emerged.
Metaphor and the Construction of
We have said little thus far about
the school of generative semanticists---heirs to Chomsky who place major emphasis
on how context directs the shape and form of language. They
claim that much of language is constructed by metaphoric operations (projections
of a pattern from one domain of experience in order to structure another domain
of a different kind) based on fundamental bodily functions. The idea is that
the logic of muscle action in the world (up, down, sideways, in, out, containing,
rotation) might be a foundation upon which logic operations of language are
built. Recurring dynamic patterns of our perceptual interactions and motor
programs give coherence and structure to our experience. A verticality scheme,
for instance, emerges from our tendency to employ an up-down orientation in
picking out meaningful structures of our experience (trees, stairs, flagpoles,
other humans). Our brains during development form themselves in an interaction
with vertical stimuli; some nerve cells are wired to specialize in them. Consider
the simple and pervasive metaphorical understanding that "more is up." This
derives from the verticality schema---we are understanding quantity in terms
of the verticality schema---as in "prices keep going up," and "his
earnings fell." The metaphorical projection from up to more is natural,
because it is related to common everyday experiences. When we add more liquid
to a container, the level goes up. When we add more objects to a pile, the
level goes up. There is, then, a physical basis for our abstract understanding
of quantity. In this and many other cases, embodied human understanding is
indispensable for meaning and rationality.
The in-out scheme gives rise to
container metaphors based on the human body, our own in-out orientation. Examples
include our speaking of moving into or out of a forest and our asking, "Are
you in the race on Sunday?" (where the race is spoken of as a container.)
We use metaphors as a central tool in talking about our own mental states and
the minds of others, usually without being aware of how this biases our inquiries
at the outset. Most of us view our minds as containers and conceive of thoughts
as being like physical objects inside them; we also take thoughts to be natural
language utterances inside our heads. If you say "part of me doesn't believe
he is telling the truth," you are using the convention of talking about
mind parts as persons.
The hypothesis is that many of
these operations are so fundamental to living in a human body that they lead
to a class of perceptions, behaviors, and language conventions that are universal.
They are part of the basis of the "universal mind" that we discussed
in Chapter 5. The metaphorical operations used to build language, however,
can be tied to specific cultural surroundings. For most of us, time not only
has its basic meaning but also is treated as though it were money. ("You're
wasting my time." "This gadget will save you hours." "He's
living on borrowed time." "I've invested a lot of time in her.").
Some cultures don't use this metaphor, just as they don't equate active with
up and passive with down. In many cultures there is a metaphoric mapping of
masculine and feminine onto many other dimensions, such as light-dark, public-private,
science-arts, rational-intuitive, active-passive, and hard-soft. One
counter-reaction to our immersion in metaphor is to argue that there is no "correct" way
to perceive the world---that all the truths of our culture, including scientific
dogma, are relative. Arguments against this position were outlined at the end
of Chapter 7. Metaphors function as mind-tools, or as a scaffolding on which
to arrange our complex physical and social reality, but they should not be
taken as our only means of approaching and understanding it.
The take-home message of this chapter
is that the genetic endowment of all modern humans equips them to generate
brain modules for processing language if they grow up in the company of other
humans. The complex social organization that language makes possible is a quantum
leap beyond that of our primate precursors. Studies of both human development
and comparative anthropology show that isolated groups of humans can invent
new languages with complex grammars and communicate them either by speaking
or by making manual signs. Although we will never know exactly how and when
the ability to deal with symbols in verbal or manual form first appeared, it
is clear that the many specializations of our vocal apparatus that support
spoken language are adaptations that have enhanced its usefulness. The same
is probably true of the brain mechanisms that permit processing and analysis
of rapid sound sequences. There are no obvious "language genes" whose
appearance might have correlated with the evolution of language competence.
It seems likely that this evolution entailed the refinement and expansion of
many of the cognitive faculties already involved in sensing, acting, memory,
and communication. Observations on brain lesions and imaging data both suggest
that many areas of the cortex are recruited during language operations. The
fact that the usual brain locations of language operations can be changed by
genetic or developmental perturbations, as well as by injury, suggests that
there is great plasticity in the mechanisms that put language components in
place. These components appear to generate our language utterances by yet another "Darwin
Machine" mechanism, in which an array of candidates for expression vie
for predominance. The streams of information we emit then apply a variety of
techniques, such as the use of extended metaphors, to stack meanings upon meanings
to categorize our experience of the external world. It is this process that
permits us to engage the topics of Part IV of this book---to share our thinking
about how our self consciousness and modern minds are integrated into a unitary
Questions for Thought
1. A robust current debate centers
on whether language is an evolutionary adaptation, or whether it is an accidental
byproduct of humans having developed a very complex cerebral cortex. Can you
offer some arguments relevant to one or the other side of this issue?
2. Several lines of evidence suggest
that interacting children spontaneously invent the elements of language. Language
development is not observed in the several known examples of feral children
raised by animals in the wild. Can you develop the position that these observations
are not relevant (or, alternatively, that they are relevant) to the question
of whether stages in language development require the expression of "language
enabling genes" unique to humans?
3. Brain lesions can cause specific
language deficits, such as inabilities to understand, speak, write, or deal
with specific categories of knowledge. Genetic mutations which influence human
language ability do cause such specific deficits. What does this suggest about
how the more specific faculties are constructed?
4. Roughly normal language development
can be observed in children who have had one hemisphere removed at an early
age. Can you offer a suggestion for why, then, during normal language development,
a number of language functions distribute between the two hemispheres?
Suggestions for further general
Pinker, S. 1994. The Language Instinct.
New York: William Morrow. If you are going to read only one popular book on
language, this should probably be that book. It describes the construction
of languages, argues that all humans have a similar evolved capacity to learn
language, and discusses some of the brain mechanisms involved.
Diamond, J. 1992. The Third Chimpanzee.
New York: Harper Collins. Chapter 8 of this book describes anthropological
studies that elucidate how different groups of humans grow different languages.
Crystal, D. 1987. The Cambridge
Encyclopedia of Language. Cambridge, England: Cambridge University Press. A
presentation of the structures and varieties of languages.
Lakoff, G., & Johnson, M. 1980.
Metaphors We Live By. Chicago: University of Chicago Press. This book argues
that many linguistic operations derive from basic physical operations of the
human body and are further developed by the use of culturally specific metaphors.
Reading on more advanced or specialized
Gazzaniga, M.S., Ivry, R.B., & Mangun,
G.R. 1998. Cognitive Neuroscience. New York: Norton. Chapter 8 of this book
is devoted to language and the brain and is an up-to-date summary of what we
have learned from lesion and imaging studies.
Elman, J.L., Bates, E.A., Johnson,
M.H., Karmiloff-Smith, A., Parisi, D., & Plunkett, K. 1996. Rethinking
Innateness: A Connectionist Perspective on Development. Cambridge, MA: M.I.T.
Press. This book takes a position opposing the argument of Pinker and others
that many detailed aspects of language are genetically determined.
Nobre, A.C., & Plunkett, K.
1997. The neural system of language: structure and development. Current Opinions
in Neurobiology 7(2):262--268. This is a summary of neuroimaging and brain
lesion studies that suggest that the neural system for language is widely distributed
and shares organizational principles with other cognitive systems in the brain.
1992, Ch. 8