Language Acquisition

Steven Pinker
Massachusetts Institute of Technology
Chapter to appear in L. R. Gleitman, M. Lierman, an! ". #. $sherson %&!s.',
(n Invitation to Cognitive Science, )n! &!. *olume +, Language. Camri!ge, M(, MIT Press.
#$#-I#(L *&RSI$#, PL&(S& "$ #$T& ./$T&.
Preparation of the chapter 0as supporte! y #I1 grant 1" +232+ an! #S- grant 4#S 5+675899,
an! y the Mc"onnell6Pe0 Center for Cognitive #euroscience at MIT.
1 Introduction
Language acquisition is one of the central topics in cognitive science. Every theory of cognition has
tried to explain it; probably no other topic has aroused such controversy. Possessing a language is the
quintessentially human trait: all normal humans speak, no nonhuman animal does. Language is the
main vehicle by hich e kno about other people!s thoughts, and the to must be intimately related.
Every time e speak e are revealing something about language, so the facts of language structure
are easy to come by; these data hint at a system of extraordinary complexity. "onetheless, learning a
first language is something every child does successfully, in a matter of a fe years and ithout the
need for formal lessons. #ith language so close to the core of hat it means to be human, it is not
surprising that children!s acquisition of language has received so much attention. $nyone ith strong
vies about the human mind ould like to sho that children!s first fe steps are steps in the right
direction.
Language acquisition is not only inherently interesting; studying it is one ay to look for concrete
ansers to questions that permeate cognitive science:
%odularity. &o children learn language using a 'mental organ,' some of hose principles of
organi(ation are not shared ith other cognitive systems such as perception, motor control, and
reasoning )*homsky, +,-., +,,+; /odor, +,0123 4r is language acquisition 5ust another problem to be
solved by general intelligence, in this case, the problem of ho to communicate ith other humans
over the auditory channel )Putnam, +,-+; 6ates, +,0,23
7uman 8niqueness. $ related question is hether language is unique to humans. $t first glance the
anser seems obvious. 4ther animals communication ith a fixed repertoire of symbols, or ith
analogue variation like the mercury in a thermometer. 6ut none appears to have the combinatorial rule
system of human language, in hich symbols are permuted into an unlimited set of combinations,
each ith a determinate meaning. 4n the other hand, many other claims about human uniqueness,
such as that humans ere the only animals to use tools or to fabricate them, have turned out to be
false. 9ome researchers have thought that apes have the capacity for language but never profited from
a humanlike cultural milieu in hich language as taught, and they have thus tried to teach apes
language:like systems. #hether they have succeeded, and hether human children are really 'taught'
language themselves, are questions e ill soon come to.
Language and ;hought. <s language simply grafted on top of cognition as a ay of sticking
communicable labels onto thoughts )/odor, +,-.; Piaget, +,=>23 4r does learning a language
someho mean learning to think in that language3 $ famous hypothesis, outlined by 6en5amin #horf
)+,.>2, asserts that the categories and relations that e use to understand the orld come from our
+
particular language, so that speakers of different languages conceptuali(e the orld in different ays.
Language acquisition, then, ould be learning to think, not 5ust learning to talk.
;his is an intriguing hypothesis, but virtually all modern cognitive scientists believe it is false )see
Pinker, +,,?a2. 6abies can think before they can talk )*hapter @2. *ognitive psychology has shon
that people think not 5ust in ords but in images )see *hapter @2 and abstract logical propositions )see
the chapter by Larson2. $nd linguistics has shon that human languages are too ambiguous and
schematic to use as a medium of internal computation: hen people think about 'spring,' surely they
are not confused as to hether they are thinking about a season or something that goes 'boing' :: and
if one ord can correspond to to thoughts, thoughts can!t be ords.
6ut language acquisition has a unique contribution to make to this issue. $s e shall see, it is virtually
impossible to sho ho children could learn a language unless you assume they have a considerable
amount of nonlinguistic cognitive machinery in place before they start.
Learning and <nnateness. $ll humans talk but no house pets or house plants do, no matter ho
pampered, so heredity must be involved in language. 6ut a child groing up in Aapan speaks Aapanese
hereas the same child brought up in *alifornia ould speak English, so the environment is also
crucial. ;hus there is no question about hether heredity or environment is involved in language, or
even hether one or the other is 'more important.' <nstead, language acquisition might be our best
hope of finding out ho heredity and environment interact. #e kno that adult language is intricately
complex, and e kno that children become adults. ;herefore something in the child!s mind must be
capable of attaining that complexity. $ny theory that posits too little innate structure, so that its
hypothetical child ends up speaking something less than a real language, must be false. ;he same is
true for any theory that posits too much innate structure, so that the hypothetical child can acquire
English but not, say, 6antu or Bietnamese.
$nd not only do e kno about the output of language acquisition, e kno a fair amount about the
input to it, namely, parent!s speech to their children. 9o even if language acquisition, like all cognitive
processes, is essentially a 'black box,' e kno enough about its input and output to be able to make
precise guesses about its contents.
;he scientific study of language acquisition began around the same time as the birth of cognitive
science, in the late +,.C!s. #e can see no hy that is not a coincidence. ;he historical catalyst as
"oam *homsky!s revie of 9kinner!s Berbal 6ehavior )*homsky, +,.,2. $t that time, $nglo:
$merican natural science, social science, and philosophy had come to a virtual consensus about the
ansers to the questions listed above. ;he mind consisted of sensorimotor abilities plus a fe simple
las of learning governing gradual changes in an organism!s behavioral repertoire. ;herefore language
must be learned, it cannot be a module, and thinking must be a form of verbal behavior, since verbal
behavior is the prime manifestation of 'thought' that can be observed externally. *homsky argued that
language acquisition falsified these beliefs in a single stroke: children learn languages that are
governed by highly subtle and abstract principles, and they do so ithout explicit instruction or any
other environmental clues to the nature of such principles. 7ence language acquisition depends on an
innate, species:specific module that is distinct from general intelligence. %uch of the debate in
language acquisition has attempted to test this once:revolutionary, and still controversial, collection of
ideas. ;he implications extend to the rest of human cognition.
2 The Biology of Language Acquisition
7uman language is made possible by special adaptations of the human mind and body that occurred in
the course of human evolution, and hich are put to use by children in acquiring their mother tongue.
=
2.1 Evolution of Language
%ost obviously, the shape of the human vocal tract seems to have been modified in evolution for the
demands of speech. 4ur larynxes are lo in our throats, and our vocal tracts have a sharp right angle
bend that creates to independently:modifiable resonant cavities )the mouth and the pharynx or
throat2 that defines a large to:dimensional range of voel sounds )see the chapter by Liberman2. 6ut
it comes at a sacrifice of efficiency for breathing, salloing, and cheing )Lieberman, +,0?2. 6efore
the invention of the 7eimlich maneuver, choking on food as a common cause of accidental death in
humans, causing >,CCC deaths a year in the 8nited 9tates. ;he evolutionary selective advantages for
language must have been very large to outeigh such a disadvantage.
<t is tempting to think that if language evolved by gradual &arinian natural selection, e must be
able to find some precursor of it in our closest relatives, the chimpan(ees. <n several famous and
controversial demonstrations, chimpan(ees have been taught some hand:signs based on $merican
9ign Language, to manipulate colored sitches or tokens, and to understand some spoken commands
)Dardner E Dardner, +,>,; Premack E Premack, +,01; 9avage:Fumbaugh, +,,+2. #hether one ants
to call their abilities 'language' is not really a scientific question, but a matter of definition: ho far
e are illing to stretch the meaning of the ord 'language'.
;he scientific question is hether the chimps! abilities are homologous to human language :: that is,
hether the to systems sho the same basic organi(ation oing to descent from a single system in
their common ancestor. /or example, biologists don!t debate hether the ing:like structures of
gliding rodents may be called 'genuine ings' or something else )a boring question of definitions2.
<t!s clear that these structures are not homologous to the ings of bats, because they have a
fundamentally different anatomical plan, reflecting a different evolutionary history. 6ats! ings are
modifications of the hands of the common mammalian ancestor; flying squirrels! ings are
modifications of its rib cage. ;he to structures are merely analogous: similar in function.
;hough artificial chimp signaling systems have some analogies to human language )e.g., use in
communication, combinations of more basic signals2, it seems unlikely that they are homologous.
*himpan(ees require massive regimented teaching sequences contrived by humans to acquire quite
rudimentary abilities, mostly limited to a small number of signs, strung together in repetitive, quasi:
random sequences, used ith the intent of requesting food or tickling );errace, Petitto, 9anders, E
6ever, +,-,; 9eidenberg E Petitto, +,-,, +,0-; 9eidenberg, +,0>; #allman, +,,=; Pinker, +,,?a2.
;his contrasts sharply ith human children, ho pick up thousands of ords spontaneously, combine
them in structured sequences here every ord has a determinate role, respect the ord order of the
adult language, and use sentences for a variety of purposes such as commenting on interesting ob5ects.
;his lack of homology does not, by the ay, cast doubt on a gradualistic &arinian account of
language evolution. 7umans did not evolve directly from chimpan(ees. 6oth derived from common
ancestor, probably around >:- million years ago. ;his leaves about 1CC,CCC generations in hich
language could have evolved gradually in the lineage leading to humans, after it split off from the
lineage leading to chimpan(ees. Presumably language evolved in the human lineage for to reasons:
our ancestors developed technology and knoledge of the local environment in their lifetimes, and
ere involved in extensive reciprocal cooperation. ;his alloed them to benefit by sharing hard:on
knoledge ith their kin and exchanging it ith their neighbors )Pinker E 6loom, +,,C2.
2.2 Dissociations between Language and eneral Intelligence
7umans evolved brain circuitry, mostly in the left hemisphere surrounding the sylvian fissure, that
appears to be designed for language, though ho exactly their internal iring gives rise to rules of
language is unknon )see the *hapter by Gurif2. ;he brain mechanisms underlying language are not
1
5ust those alloing us to be smart in general. 9trokes often leave adults ith catastrophic losses in
language )see the *hapter by Gurif, and Pinker, +,,?a2, though not necessarily impaired in other
aspects of intelligence, such as those measured on the nonverbal parts of <H tests. 9imilarly, there is an
inherited set of syndromes called 9pecific Language <mpairment )Dopnik and *rago, +,,1; ;allal,
Foss, E *urtiss, +,0,2 hich is marked by delayed onset of language, difficulties in articulation in
childhood, and lasting difficulties in understanding, producing, and 5udging grammatical sentences.
6y definition, 9pecifically Language <mpaired people sho such deficits despite the absence of
cognitive problems like retardation, sensory problems like hearing loss, or social problems like
autism.
%ore interestingly, there are syndromes shoing the opposite dissociation, here intact language
coexists ith severe retardation. ;hese cases sho that language development does not depend on
fully functioning general intelligence. 4ne example comes from children ith 9pina 6ifida, a
malformation of the vertebrae that leaves the spinal cord unprotected, often resulting in
hydrocephalus, an increase in pressure in the cerebrospinal fluid filling the ventricles )large cavities2
of the brain, distending the brain from ithin. 7ydrocephalic children occasionally end up
significantly retarded but can carry on long, articulate, and fully grammatical conversations, in hich
they earnestly recount vivid events that are, in fact, products of their imaginations )*romer, +,,=;
*urtiss, +,0,; Pinker, +,,?a2. $nother example is #illiams 9yndrome, an inherited condition
involving physical abnormalities, significant retardation )the average <H is about .C2, incompetence at
simple everyday tasks )tying shoelaces, finding one!s ay, adding to numbers, and retrieving items
from a cupboard2, social armth and gregariousness, and fluent, articulate language abilities )6ellugi,
et al., +,,C2.
2.! "aturation of the Language #yste$
$s the chapter by "eport and Dleitman suggests, the maturation of language circuits during a child!s
early years may be a driving force underlying the course of language acquisition )Pinker, +,,?,
*hapter ,; 6ates, ;hal, E Aanosky, +,,=; Locke, +,,=; 7uttenlocher, +,,C2. 6efore birth, virtually
all the neurons )nerve cells2 are formed, and they migrate into their proper locations in the brain. 6ut
head si(e, brain eight, and thickness of the cerebral cortex )gray matter2, here the synapses
)5unctions2 subserving mental computation take place, continue to increase rapidly in the year after
birth. Long:distance connections )hite matter2 are not complete until nine months, and they continue
to gro their speed:inducing myelin insulation throughout childhood. 9ynapses continue to develop,
peaking in number beteen nine months and to years )depending on the brain region2, at hich
point the child has .CI more synapses than the adult. %etabolic activity in the brain reaches adult
levels by nine to ten months, and soon exceeds it, peaking around the age of four. <n addition, huge
numbers of neurons die in utero, and the dying continues during the first to years before leveling off
at age seven. 9ynapses ither from the age of to through the rest of childhood and into adolescence,
hen the brain!s metabolic rate falls back to adult levels. Perhaps linguistic milestones like babbling,
first ords, and grammar require minimum levels of brain si(e, long:distance connections, or extra
synapses, particularly in the language centers of the brain.
9imilarly, one can con5ecture that these changes are responsible for the decline in the ability to learn a
language over the lifespan. ;he language learning circuitry of the brain is more plastic in childhood;
children learn or recover language hen the left hemisphere of the brain is damaged or even surgically
removed )though not quite at normal levels2, but comparable damage in an adult usually leads to
permanent aphasia )*urtiss, +,0,; Lenneberg, +,>-2. %ost adults never master a foreign language,
especially the phonology, giving rise to hat e call a 'foreign accent.' ;heir development often
fossili(es into permanent error patterns that no teaching or correction can undo. ;here are great
individual differences, hich depend on effort, attitudes, amount of exposure, quality of teaching, and
plain talent.
?
%any explanations have been advanced for children!s superiority: they can exploit the special ays
that their mothers talk them, they make errors unself:consciously, they are more motivated to
communicate, they like to conform, they are not xenophobic or set in their ays, and they have no
first language to interfere. 6ut some of these accounts are unlikely, based on hat e learn about ho
language acquisition orks later in this chapter. /or example, children can learn a language ithout
the special indulgent speech from their mothers; they make fe errors; and they get no feedback for
the errors they do make. $nd it can!t be an across:the:board decline in learning. ;here is no evidence,
for example, that learning ords )as opposed to phonology or grammar2 declines in adulthood.
;he chapter by "eport and Dleitman shos ho sheer age seems to play an important role.
9uccessful acquisition of language typically happens by ? )as e shall see in the next section2, is
guaranteed for children up to the age of six, is steadily compromised from then until shortly after
puberty, and is rare thereafter. %aturational changes in the brain, such as the decline in metabolic rate
and number of neurons during the early school age years, and the bottoming out of the number of
synapses and metabolic rate around puberty, are plausible causes. ;hus, there may be a neurologically:
determined 'critical period' for successful language acquisition, analogous to the critical periods
documented in visual development in mammals and in the acquisition of songs by some birds.
! The %ourse of Language Acquisition
$lthough scholars have kept diaries of their children!s speech for over a century )*harles &arin as
one of the first2, it as only after portable tape:recorders became available in the late +,.C!s that
children!s spontaneous speech began to be analy(ed systematically ithin developmental psychology.
;hese naturalistic studies of children!s spontaneous speech have become even more accessible no
that they can be put into computer files and can be disseminated and analy(ed automatically
)%ac#hinney E 9no, +,0., +,,C; %ac#hinney, +,,+2. ;hey are complemented by experimental
methods. <n production tasks, children utter sentences to describe pictures or scenes, in response to
questions, or to imitate target sentences. <n comprehension tasks, they listen to sentences and then
point to pictures or act out events ith toys. <n 5udgement tasks, they indicate hether or hich
sentences provided by an experimenter sound 'silly' to them.
$s the chapter by #erker shos, language acquisition begins very early in the human lifespan, and
begins, logically enough, ith the acquisition of a language!s sound patterns. ;he main linguistic
accomplishments during the first year of life are control of the speech musculature and sensitivity to
the phonetic distinctions used in the parents! language. <nterestingly, babies achieve these feats before
they produce or understand ords, so their learning cannot depend on correlating sound ith meaning.
;hat is, they cannot be listening for the difference in sound beteen a ord they think means bit and a
ord they think means beet, because they have learned neither ord. ;hey must be sorting the sounds
directly, someho tuning their speech analysis module to deliver the phonemes used in their language
)Juhl, et al., +,,=2. ;he module can then serve as the front end of the system that learns ords and
grammar.
9hortly before their first birthday, babies begin to understand ords, and around that birthday, they
start to produce them )see *lark, +,,1; <ngram, +,0,2. #ords are usually produced in isolation; this
one:ord stage can last from to months to a year. *hildren!s first ords are similar all over the
planet. $bout half the ords are for ob5ects: food )5uice, cookie, body parts )eye, nose2, clothing
)diaper, sock2, vehicles )car, boat2, toys )doll, block2, household items )bottle, light, animals )dog,
kitty2, and people )dada, baby2. ;here are ords for actions, motions, and routines, like )up, off, open,
peekaboo, eat, and go, and modifiers, like hot, allgone, more, dirty, and cold. /inally, there are
routines used in social interaction, like yes, no, ant, bye:bye, and hi :: a fe of hich, like look at
that and hat is that, are ords in the sense of memori(ed chunks, though they are not single ords
.
for the adult. *hildren differ in ho much they name ob5ects or engage in social interaction using
memori(ed routines, though all children do both.
$round +0 months, language changes in to ays. Bocabulary groth increases; the child begins to
learn ords at a rate of one every to aking hours, and ill keep learning that rate or faster through
adolescence )*lark, +,,1; Pinker, +,,?2. $nd primitive syntax begins, ith to:ord strings like the
folloing:
All dry. All messy. All wet.
I sit. I shut. No bed.
No pee. See baby. See pretty.
More cereal. More hot. Hi Calico.
Other pocket. Boot off. Siren by.
Mail come. Airplane allone. Bybebye car.
Our car. !apa away. "ry pants.
4ur car. Papa aay. &ry pants. *hildren!s to:ord combinations are highly similar across cultures.
Everyhere, children announce hen ob5ects appear, disappear, and move about, point out their
properties and oners, comment on people doing things and seeing things, re5ect and request ob5ects
and activities, and ask about ho, hat, and here. ;hese sequences already reflect the language
being acquired: in ,.I of them, the ords are properly ordered )6raine, +,->; 6ron, +,-1; Pinker,
+,0?; <ngram, +,0,2.
Even before they put ords together, babies can comprehend a sentence using its syntax. /or example,
in one experiment, babies ho spoke only in single ords ere seated in front of to television
screens, each of hich featured a pair of adults dressed up as *ookie %onster and 6ig 6ird from
9esame 9treet. 4ne screen shoed *ookie %onster tickling 6ig 6ird; the other shoed 6ig 6ird
tickling *ookie %onster. $ voice:over said, '47 L44JKKK 6<D 6<F& <9 ;<*JL<"D *44J<E
%4"9;EFKK /<"& 6<D 6<F& ;<*JL<"D *44J<E %4"9;EFKK' )4r vice:versa.2 ;he children
must have understood the meaning of the ordering of sub5ect, verb, and ob5ect, because they looked
more at the screen that depicted the sentence in the voice:over )7irsh:Pasek E Dolinkoff, +,,+2.
*hildren!s output seems to meet up ith a bottleneck at the output end )6ron, +,-1; 6loom, +,-C;
Pinker, +,0?2. ;heir to: and three:ord utterances look like samples dran from longer potential
sentences expressing a complete and more complicated idea. Foger 6ron, one of the founders of the
modern study of language development, noted that although the three children he studied intensively
never produced a sentence as complicated as %other gave Aohn lunch in the kitchen, they did produce
strings containing all of its components, and in the correct order: )6ron, +,-1, p. =C.2:
Aent Action #ecipient Ob$ect %ocation
&Mother a'e (ohn lunch in the kitchen.)
Mommy fi*.
Mommy pumpkin.
Baby table.
+i'e doie.
!ut liht.
!ut floor.
I ride horsie.
,ractor o floor.
+i'e doie paper.
!ut truck window.
Adam put it bo*.
6eteen the late to!s and mid:three!s, children!s language blooms into fluent grammatical
conversation so rapidly that it overhelms the researchers ho study it, and no one has orked out
the exact sequence. 9entence length increases steadily, and because grammar is a combinatorial
system, the number of syntactic types increases exponentially, doubling every month, reaching the
>
thousands before the third birthday )<ngram, +,0,, p. =1.; 6ron, +,-1; Limber, +,-1; Pinker, +,0?2.
/or example, here are snapshots of the development of one of 6ron!s longitudinal sub5ects, $dam, in
the year folloing his first ord combinations at the age of = years and 1 months )Pinker, +,,?a2:
=;1: Play checkers. 6ig drum. < got horn.
=;?: 9ee marching bear go3 9cre part machine.
=;.: "o put boots on. #here rench go3 #hat that paper clip doing3
=;>: #rite a piece a paper. #hat that egg doing3 "o, < don!t ant to sit seat.
=;-: #here piece a paper go3 &ropped a rubber band. Fintintin don!t fly, %ommy.
=;0: Let me get don ith the boots on. 7o tiger be so healthy and
fly like kite3 Aoshua thro like a penguin.
=;,: #here %ommy keep her pocket book3 9ho you something funny.
=;+C: Look at that train 8rsula brought. Lou don!t have paper. &o you ant little bit, *romer3
=;++: &o ant some pie on your face3 #hy you mixing baby chocolate3 < said hy not you coming
in3 #e going turn light on so you can!t : see.
1;C: < going come in fourteen minutes. < going ear that to edding. ;hose are not strong mens. Lou
dress me up like a baby elephant.
1;+: < like to play ith something else. Lou kno ho to put it back together. < gon! make it like a
rocket to blast off ith. Lou ant : to give me some carrots and some beans3 Press the button and
catch : it, sir. #hy you put the pacifier in his mouth3
1;=: 9o it can!t be cleaned3 < broke my racing car. &o you kno the light ents off3 #hen it!s got a
flat tire it!s need a go to the station. <!m going to mail this so the letter can!t come off. < : ant to have
some espresso. *an < put my head in the mailbox so : the mailman can kno here < are and put me
in the mailbox3 *an < : keep the scredriver 5ust like a carpenter keep the scredriver3
"ormal children can differ by a year or more in their rate of language development, though the stages
they pass through are generally the same regardless of ho stretched out or compressed. $dam!s
language development, for example, as relatively leisurely; many children speak in complex
sentences before they turn to.
&uring the grammar explosion, children!s sentences are getting not only longer but more complex,
ith fuller trees, because the children can embed one constituent inside another. #hereas before they
might have said Dive doggie paper )a three:branch Berb Phrase2 and 6ig doggie )a to:branch "oun
Phrase2, they no say Dive big doggie paper, ith the to:branch "P embedded inside the three:
branch BP. ;he earlier sentences resembled telegrams, missing unstressed function ords like of, the,
on, and does, as ell as inflections like :ed, :ing, and :s. 6y the 1!s, children are using these function
ords more often than they are omitting them, many in more than ,CI of the sentences that require
them. $ full range of sentence types floer :: questions ith ords like ho, hat and here, relative
clauses, comparatives, negations, complements, con5unctions, and passives. ;hese constructions
appear to display the most, perhaps even all, of the grammatical machinery needed to account for
adult grammar.
-
;hough many of the young 1:year:old!s sentences are ungrammatical for one reason or another, it is
because there are many things that can go rong in any single sentence. #hen researchers focus on a
single grammatical rule and count ho often a child obeys it and ho often he or she versus flouts it,
the results are very impressive: for 5ust about every rule that has been looked at, three:year olds obey
it a ma5ority of the time )9tromsold, +,,C; Pinker, +,0?, +,0,; *rain, +,,=; %arcus, et al., +,,=2. $s
e have seen, children rarely scramble ord orders and, by the age of three, come to supply most
inflections and function ords in sentences that require them. ;hough our ears perk up hen e hear
errors like mens, ents, *an you broke those3, #hat he can ride in3, ;hat!s a furniture, 6utton me the
rest, and Doing to see kitten, the errors occur in anyhere from C.+I to 0I of the opportunities for
making them; more than ,CI of the time, the child is on target. ;he next chapter follos one of those
errors in detail.
*hildren do not seem to favor any particular kind of language )indeed, it ould be pu((ling ho any
kind of language could survive if children did not easily learn itK2. ;hey siftly acquire free ord
order, 94B and B94 orders, rich systems of case and agreement, strings of agglutinated suffixes,
ergative case marking, and hatever else their language thros at them, ith no lag relative to their
English:speaking counterparts. Even grammatical gender, hich many adults learning a second
language find mystifying, presents no problem: children acquiring language like /rench, Derman, and
7ebre acquire gender marking quickly, make fe errors, and never use the association ith
maleness and femaleness as a false criterion )Levy, +,012. <t is safe to say that except for
constructions that are rare, predominantly used in ritten language, or mentally taxing even to an
adult )like ;he horse that the elephant tickled kissed the pig2, all parts of all languages are acquired
before the child turns four )9lobin, +,0.M+,,=2.
& E'(laining Language Acquisition
7o do e explain children!s course of language acquisition :: most importantly, their inevitable and
early mastery3 9everal kinds of mechanisms are at ork. $s e sa in section )2, the brain changes
after birth, and these maturational changes may govern the onset, rate, and adult decline of language
acquisition capacity. Deneral changes in the child!s information processing abilities )attention,
memory, short:term buffers for acoustic input and articulatory output2 could leave their mark as ell.
<n the next chapter, < sho ho a memory retrieval limitation :: children are less reliable at recalling
that broke is the past tense of break :: can account for a conspicuous and universal error pattern,
overregulari(ations like breaked )see also %arcus, et al., +,,=2.
%any other small effects have been documented here changes in information processing abilities
affect language development. /or example, children selectively pick up information at the ends of
ords )9lobin, +,-12, and at the beginnings and ends of sentences )"eport, et al, +,--2, presumably
because these are the parts of strings that are best retained in short term memory. 9imilarly, the
progressively idening bottleneck for early ord combinations presumably reflects a general increase
in motor planning capacity. *onceptual development )see *hapter @2, too, might affect language
development: if a child has not yet mastered a difficult semantic distinction, such as the complex
temporal relations involved in Aohn ill have gone, he or she may be unable to master the syntax of
the construction dedicated to expressing it.
;he complexity of a grammatical form has a demonstrable role in development: simpler rules and
forms appear in speech before more complex ones, all other things being equal. /or example, the
plural marker :s in English )e.g. cats2, hich requires knoing only hether the number of referents is
singular or plural, is used consistently before the present tense marker :s )he alks2, hich requires
knoing hether the sub5ect is singular or plural and hether it is a first, second, or third person and
hether the event is in the present tense )6ron, +,-12. 9imilarly, complex forms are sometimes first
used in simpler approximations. Fussian contains one case marker for masculine nominative )i.e., a
0
suffix on a masculine noun indicating that it is the sub5ect of the sentence2, one for feminine
nominative, one for masculine accusative )used to indicate that a noun is a direct ob5ect2, and one for
feminine accusative. *hildren often use each marker ith the correct case, never using a nominative
marker for accusative nouns or vice:versa, but don!t properly use the masculine and feminine variants
ith masculine and feminine nouns )9lobin, +,0.2.
6ut these global trends do not explain the main event: ho children succeed. Language acquisition is
so complex that one needs a precise frameork for understanding hat it involves :: indeed, hat
learning in general involves.
&.1 Learnability Theory
#hat is language acquisition, in principle3 $ branch of theoretical computer science called
Learnability ;heory attempts to anser this question )Dold, +,>-; 4sherson, 9tob, E #einstein, +,0.;
Pinker, +,-,2. Learnability theory has defined learning as a scenario involving four parts )the theory
embraces all forms of learning, but < ill use language as the example2:
+. $ class of languages. 4ne of them is the 'target' language, to be : attained by the learner, but
the learner does not, of course, kno : hich it is. <n the case of children, the class of
languages ould : consist of the existing and possible human languages; the target : language
is the one spoken in their community.
=. $n environment. ;his is the information in the orld that the learner has to go on in trying to
acquire the language. <n the case of children, it might include the sentences parents utter, the
context in hich they utter them, feedback to the child )verbal or nonverbal2 in response to the
child!s on speech, and so on. Parental utterances can be a random sample of the language, or
they might have some special properties: they might be ordered in certain ays, sentences
might be repeated or only uttered once, and so on.
1. $ learning strategy. ;he learner, using information in the environment, tries out 'hypotheses'
about the target language. ;he learning strategy is the algorithm that creates the hypotheses
and determines hether they are consistent ith the input information from the environment.
/or children, it is the 'grammar:forming' mechanism in their brains; their 'language
acquisition device.'
?. $ success criterion. <f e ant to say that 'learning' occurs, presumably it is because the
learners! hypotheses are not random, : but that by some time the hypotheses are related in some
systematic : ay to the target language. Learners may arrive at a hypothesis : identical to the
target language after some fixed period of time; : they may arrive at an approximation to it;
they may aiver among a : set of hypotheses one of hich is correct.
;heorems in learnability theory sho ho assumptions about any of the three components imposes
logical constraints on the fourth. <t is not hard to sho hy learning a language, on logical grounds
alone, is so hard. Like all 'induction problems' )uncertain generali(ations from instances2, there are
an infinite number of hypotheses consistent ith any finite sample of environmental information.
Learnability theory shos hich induction problems are solvable and hich are not.
$ key factor is the role of negative evidence, or information about hich strings of ords are not
sentences in the language to be acquired. 7uman children might get such information by being
corrected every time they speak ungrammatically. <f they aren!t :: and as e shall see, they probably
aren!t :: the acquisition problem is all the harder. *onsider /igure +, here languages are depicted as
circles corresponding to sets of ord strings, and all the logical possibilities for ho the child!s
language could differ from the adult language are depicted. ;here are four possibilities. )a2 ;he child!s
hypothesis language )72 is dis5oint from the language to be acquired )the 'target language,' ;2. ;hat
ould correspond to the state of child learning English ho cannot say a single ell:formed English
sentence. /or example, the child might be able only to say things like e breaked it, and e goed,
,
never e broke it or e ent. )b2 ;he child!s hypothesis and the target language intersect. 7ere the
child ould be able to utter some English sentences, like he ent. 7oever, he or she also uses strings
of ords that are not English, such as e breaked it; and some sentences of English, such as e broke
it, ould still be outside their abilities. )c2 ;he child!s hypothesis language is a subset of the target
language. ;hat ould mean that the child ould have mastered some of English, but not all of it, but
that everything the child had mastered ould be part of English. ;he child might not be able to say e
broke it, but he or she ould be able to say some grammatical sentences, such as e ent; no errors
such as she breaked it or e goed ould occur. ;he final logical possibility is )d2, here ;he child!s
hypothesis language is a superset of the target language. ;hat ould occur, for example, if the child
could say e broke it, e ent, e breaked it and e goed.
<n cases )a:c2, the child can reali(e that the hypothesis is incorrect by hearing sentences from parental
'positive evidence,' )indicated by the 'N' symbol2 that are in the target language but not the
hypothesi(ed one: sentences such as e broke it. ;his is impossible in case )d2; negative evidence
)such as corrections of the child!s ungrammatical sentences by his or her parents2 ould be needed. <n
other ords, ithout negative evidence, if a child guesses too large a language, the orld can never
tell him he!s rong.
;his has several consequences. /or one thing, the most general learning algorithm one might conceive
of :: one that is capable of hypothesi(ing any grammar, or any computer program capable of
generating a language :: is in trouble. #ithout negative evidence )and even in many cases ith it2,
there is no general:purpose, all:poerful learning machine; a machine must in some sense 'kno'
something about the constraints in the domain in hich it is learning.
%ore concretely, if children don!t receive negative evidence )see 9ection 2 e have a lot of explaining
to do, because overly large hypotheses are very easy for the child to make. /or example, children
actually do go through stages in hich they use to or more past tense forms for a given verb, such as
broke and breaked :: this case is discussed in detail in my other chapter in this volume. ;hey derive
transitive verbs from intransitives too freely: here an adult might say both ;he ice melted and <
melted the ice, children also can say ;he girl giggled and &on!t giggle meK )6oerman, +,0=b;
Pinker, +,0,2. <n each case they are in situation )d2 in /igure +, and unless their parents slip them
some signal in every case that lets them kno they are not speaking properly, it is pu((ling that they
eventually stop. ;hat is, e ould need to explain ho they gro into adults ho are more restrictive
in their speech :: or another ay of putting is that it!s pu((ling that the English language doesn!t allo
don!t giggle me and she eated given that children are tempted to gro up talking that ay. <f the orld
isn!t telling children to stop, something in their brains is, and e have to find out ho or hat is
causing the change.
Let!s no examine language acquisition in the human species by breaking it don into the four
elements that give a precise definition to learning: the target of learning, the input, the degree of
success, and the learning strategy.
) *hat is Learned
;o understand ho @ is learned, you first have to understand hat @ is. Linguistic theory is thus an
essential part of the study of language acquisition )see the *hapter by Lasnik2. Linguistic research
tries do three things. /irst, it must characteri(e the facts of English, and all the other languages hose
acquisition e are interested in explaining. 9econd, since children are not predisposed to learn English
or any other language, linguistics has to examine the structure of other languages. <n particular,
linguists characteri(e hich aspects of grammar are universal, prevalent, rare, and nonexistent across
languages. *ontrary to early suspicions, languages do not vary arbitrarily and ithout limit; there is
by no a large catalogue of language universals, properties shared exactly, or in a small number of
+C
variations, by all languages )see *omrie, +,0+; Dreenberg, +,-0; 9hopen, +,0.2. ;his obviously bears
on hat children!s language acquisition mechanisms find easy or hard to learn.
$nd one must go beyond a mere list of universals. %any universal properties of language are not
specific to language but are simply reflections of universals of human experience. $ll languages have
ords for 'ater' and 'foot' because all people need to refer to ater and feet; no language has a
ord a million syllables long because no person ould have time to say it. 6ut others might be
specific to the innate design of language itself. /or example, if a language has both derivational
suffixes )hich create ne ords from old ones, like :ism2 and inflectional suffixes )hich modify a
ord to fit its role in the sentence, like plural :s2, then the derivational suffixes are alays closer to the
ord stem than the inflectional ones. /or example, in English one can say &arinisms )derivational
:ism closer to the stem than inflectional :s2 but not &arinsism. <t is hard to think of a reason ho this
la ould fit in to any universal la of thought or memory: hy ould the concept of to ideologies
based on one &arin should be thinkable, but the concept of one ideology based on to &arins )say,
*harles and Erasmus2 not be thinkable )unless one reasons in a circle and declares that the mind must
find :ism to be more cognitively basic than the plural, because that!s the order e see in language2.
8niversals like this, that are specifically linguistic, should be captured in a theory of 8niversal
Drammar )8D2 )*homsky, +,>., +,0+, +,,+2. 8D specifies the alloable mental representations and
operations that all languages are confined to use. ;he theory of universal grammar is closely tied to
the theory of the mental mechanisms children use in acquiring language; their hypotheses about
language must be couched in structures sanctioned by 8D.
;o see ho linguistic research can!t be ignored in understanding language acquisition, consider the
sentences belo. <n each of the examples, a learner ho heard the )a2 and )b2 sentences could quite
sensibly extract a general rule that, hen applied to the )c2 sentence, yield version )d2. Let the result is
an odd sentence that no one ould say:
+. )a2 Aohn sa %ary ith her best friend!s husband.
)b2 #ho did Aohn see %ary ith3
)c2 Aohn sa %ary and her best friend!s husband.
)d2 O#ho did Aohn see %ary and3
=. )a2 <rv drove the car into the garage.
)b2 <rv drove the car.
)c2 <rv put the car into the garage.
)d2 O<rv put the car.
1. )a2 < expect the fur to fly.
)b2 < expect the fur ill fly.
)c2 ;he fur is expected to fly.
)d2 O;he fur is expected ill fly.
?. )a2 ;he baby seems to be asleep.
)b2 ;he baby seems asleep.
)c2 ;he baby seems to be sleeping.
)d2 O;he baby seems sleeping.
++
.. )a2 Aohn liked the pictures of 6ill that %ary took.
)b2 Aohn liked %ary!s pictures of 6ill.
)c2 Aohn liked the pictures of himself that %ary took.
)d2 OAohn liked %ary!s pictures of himself.
;he solution to the problem must be that children!s learning mechanisms ultimately don!t allo them
to make hat ould otherise be a tempting generali(ation. /or example, in )+2, constraints that
prevent extraction of a single phrase out of a coordinate structure )phrases 5oined by a ord like and
or or2 ould block ould hat otherise be a natural generali(ation from other examples of
extraction, such as +)a:b2. ;he other examples presents other pu((les that the theory of universal
grammar, as part of a theory of language acquisition, must solve. <t is because of the subtlety of these
examples, and the abstractness of the principles of universal grammar that must be posited to explain
them, that *homsky has claimed that the overall structure of language must be innate, based on his
paper:and:pencil examination of the facts of language alone.
+ In(ut
;o understand ho children learn language, e have to kno hat aspects of language )from their
parents or peers2 they have access to.
+.1 ,ositive Evidence
*hildren clearly need some kind of linguistic input to acquire a language. ;here have been occasional
cases in history here abandoned children have someho survived in forests, such as Bictor, the #ild
6oy of $veyron )sub5ect of a film by /rancois ;ruffaut2. 4ccasionally other modern children have
gron up ild because depraved parents have raised them silently in dark rooms and attics; the
chapter by "eport and Dleitman discuss some of those cases. ;he outcome is alays the same: the
children, hen found, are mute. #hatever innate grammatical abilities there are, they are too
schematic to generate concrete speech, ords, and grammatical constructions on their on.
*hildren do not, hoever, need to hear a full:fledged language; as long as they are in a community
ith other children, and have some source for individual ords, they ill invent one on their on,
often in a single generation. *hildren ho gre up in plantations and slave colonies ere often
exposed to a crude pidgin that served as the lingua franca in these 6abels of laborers. 6ut they gre
up to speak genuinely ne languages, expressive 'creoles' ith their on complex grammars
)6ickerton, +,0?; see also the *hapter by "eport and Dleitman2. ;he sign languages of the deaf
arose in similar ays. <ndeed, they arise spontaneously and quickly herever there is a community of
deaf children )9enghas, +,,?; Jegl, +,,?2.
*hildren most definitely do need to hear an existing language to learn that language, of course.
*hildren ith Aapanese genes do not find Aapanese any easier than English, or vice:versa; they learn
hichever language they are exposed to. ;he term 'positive evidence' refers to the information
available to the child about hich strings of ords are grammatical sentences of the target language.
6y 'grammatical,' incidentally, linguists and psycholinguists mean only those sentences that sound
natural in colloquial speech, not necessarily those that ould be deemed 'proper English' in formal
ritten prose. ;hus split infinitives, dangling participles, slang, and so on, are 'grammatical' in this
sense )and indeed, are as logical, systematic, expressive, and precise as 'correct' ritten English,
often more so; see Pinker, +,,?a2. 9imilarly, elliptical utterances, such as hen the question #here
are you going3 is ansered ith ;o the store2, count as grammatical. Ellipsis is not 5ust random
snipping from sentences, but is governed by rules that are part of the grammar of one!s language or
+=
dialect. /or example, the grammar of casual 6ritish English allos you to anser the question #ill he
go3 by saying 7e might do, hereas the grammar of $merican English doesn!t allo it.
Diven this scientific definition of 'grammatical,' do e find that parents! speech counts as 'positive
evidence'3 ;hat is, hen a parent uses a sentence, can the child assume that it is part of the language
to be learned, or do parents use so many ungrammatical sentences random fragments, slips of the
tongue, hesitations, and false starts that the child ould have to take much of it ith a grain of salt3
/ortunately for the child, the vast ma5ority of the speech they hear during the language:learning years
is fluent, complete, and grammatically ell:formed: ,,.,1I, according to one estimate )"eport,
Dleitman, E Dleitman, +,--2. <ndeed, this is true of conversation among adults in general )Labov,
+,>,2.
;hus language acquisition is ordinarily driven by a grammatical sample of the target language. "ote
that his is true even for forms of English that people unthinkingly call 'ungrammatical,' 'fractured,'
or 'bad English,' such as rural $merican English )e.g., them books; he don!t; e ain!t; they drug him
aay2 and urban black English )e.g., 9he alking; 7e be orking; see the *hapter by Labov2. ;hese
are not corrupted versions of standard English; to a linguist they look 5ust like different dialects, as
rule:governed as the southern:England dialect of English that, for historical reasons, became the
standard several centuries ago. 9cientifically speaking, the grammar of orking:class speech ::
indeed, every human language system that has been studied :: is intricately complex, though different
languages are complex in different ays.
+.2 -egative Evidence
"egative evidence refers to information about hich strings of ords are not grammatical sentences
in the language, such as corrections or other forms of feedback from a parent that tell the child that
one of his or her utterances is ungrammatical. $s mentioned in 9ection 2, it!s very important for us to
kno hether children get and need negative, because in the absence of negative evidence, any child
ho hypothesi(es a rule that generates a superset of the language ill have no ay of knoing that he
or she is rong Dold, +,>-; Pinker, +,-,, +,0,2. <f children don!t get, or don!t use, negative evidence,
they must have some mechanism that either avoids generating too large a language the child ould be
conservative :: or that can recover from such overgeneration.
Foger 6ron and *amille 7anlon )+,-C2 attempted to test 6. /. 9kinner!s behaviorist claim that
language learning depends on parents! reinforcement of children!s grammatical behaviors. 8sing
transcripts of naturalistic parent:child dialogue, they divided children!s sentences into ones that ere
grammatically ell:formed and ones that contained grammatical errors. ;hey then divided adults!
responses to those sentences into ones that expressed some kind of approval )e.g., 'yes, that!s good'2
and those that expressed some kind of disapproval. ;hey looked for a correlation, but failed to find
one: parents did not differentially express approval or disapproval to their children contingent on
hether the child!s prior utterance as ell:formed or not )approval depends, instead, on hether the
child!s utterance as true2. 6ron and 7anlon also looked at children!s ell:formed and badly:formed
questions, and hether parents seemed to anser them appropriately, as if they understood them, or
ith non sequiturs. ;hey found parents do not understand their children!s ell:formed questions better
than their badly:formed ones.
4ther studies )e.g. 7irsh:Pasek, ;reiman, and 9chneiderman, +,0?; &emetras, Post, and 9no, +,0>;
Penner, +,0-; 6ohannon E 9tanoic(, +,002 have replicated that result, but ith a tist. 9ome have
found small statistical contingencies beteen the grammaticality of some children!s sentence and the
kind of follo:up given by their parents; for example, hether the parent repeats the sentence
verbatim, asks a follo:up question, or changes the topic. 6ut %arcus )+,,12 has found that these
patterns fall far short of negative evidence )reliable information about the grammatical status of any
+1
ord string2. &ifferent parents react in opposite ays to their children!s ungrammatical sentences, and
many forms of ungrammaticality are not reacted to at all :: leaving a given child unable to kno hat
to make of any parental reaction. Even hen a parent does react differentially, a child ould have to
repeat a particular error, verbatim, hundreds of times to eliminate the error, because the parent!s
reaction is only statistical: the feedback signals given to ungrammatical signals are also given nearly
as often to grammatical sentences.
9tromsold )+,,?2 has an even more dramatic demonstration that parental feedback cannot be crucial.
9he studied a child ho, for unknon neurological reasons, as congenitally unable to talk. 7e as a
good listener, though, and hen tested he as able to understand complicated sentences perfectly, and
to 5udge accurately hether a sentence as grammatical or ungrammatical. ;he boy!s abilities sho
that children certainly do not need negative evidence to learn grammatical rules properly, even in the
unlikely event that their parents provided it.
;hese results, though of profound importance, should not be too surprising. Every speaker of English
5udges sentences such as < dribbled the floor ith paint and ;en pounds as eighed by the boy and
#ho do you believe the claim that Aohn sa3 and Aohn asked %ary to look at himself to be
ungrammatical. 6ut it is unlikely that every such speaker has at some point uttered these sentences
and benefited from negative feedback. ;he child must have some mental mechanisms that rule out
vast numbers of 'reasonable' strings of ords ithout any outside intervention.
+.! "otherese
Parents and caretakers in most parts of the orld modify their speech hen talking to young children,
one example of ho people in general use several 'registers' in different social settings. 9peech to
children is sloer, shorter, in some ays )but not all2 simpler, higher:pitched, more exaggerated in
intonation, more fluent and grammatically ell:formed, and more directed in content to the present
situation, compared to speech among adults )9no E /erguson, +,--2. %any parents also expand
their children!s utterances into full sentences, or offer sequences of paraphrases of a given sentence.
4ne should not, though, consider this speech register, sometimes called '%otherese,' to be a set of
'language lessons.' ;hough mother!s speech may seem simple at first glance, in many ays it is not.
/or example, speech to children is full of questions :: sometimes a ma5ority of the sentences. <f you
think questions are simple, 5ust try to rite a set of rules that accounts for the folloing sentences and
non:sentences:
+. 7e can go somehere.
#here can he go3
O#here can he go somehere3
O#here he can go3
O#here did he can go3
=. 7e ent somehere.
#here did he go3
7e ent #7EFE3
O#here ent he3
O#here did he ent3
O#here he ent3
O7e did go #7EFE3
1. 7e ent home.
#hy did he go home3
7o come he ent home3
+?
O#hy he ent home3
O7o come did he go home3
Linguists struggle over these facts )see the *hapters by Lasnik and Larson2, some of the most
pu((ling in the English language. 6ut these are the constructions that infants are bombarded ith and
that they master in their preschool years.
;he chapter by "eport and Dleitman gives another reason for doubting that %otherese is a set of
language lessons. *hildren hose mothers use %otherese more consistently don!t pass through the
milestones of language development any faster )"eport, et al, +,--2. /urthermore, there are some
communities ith radically different ideas about children!s proper place in society. <n some societies,
for example, people tacitly assume that that children aren!t orth speaking to, and don!t have anything
to say that is orth listening to. 9uch children learn to speak by overhearing streams of adult:to:adult
speech )7eath, +,012. <n some communities in "e Duinea, mothers consciously try to teach their
children language, but not in the style familiar to us, of talking to them indulgently. Father, they ait
until a third party is present, and coach the child as to the proper, adultlike sentences they should use
)see 9chieffelin E Eisenberg, +,0+2. "onetheless, those children, like all children, gro up to be
fluent language speakers. <t surely must help children hen their parents speak sloly, clearly, and
succinctly to them, but their success at learning can!t be explained by any special grammar:unveiling
properties of parental babytalk.
+.& ,rosody
Parental speech is not a string of printed ords on a ticker:tape, nor is it in a monotone like science:
fiction robots. "ormal human speech has a pattern of melody, timing, and stress called prosody. $nd
motherese directed to young infants has a characteristic, exaggerated prosody of its on: a rise and
fall contour for approving, a set of sharp staccato bursts for prohibiting, a rise pattern for directing
attention, and smooth, lo legato murmurs for comforting. /ernald )+,,=2 has shon that these
patterns are very idespread across language communities, and may be universal. ;he melodies seem
to attract the child!s attention, mark the sounds as speech as opposed to stomach grolings or other
noises, and might distinguish statements, questions, and imperatives, delineate ma5or sentence
boundaries, and highlight ne ords. #hen given a choice, babies prefer to listen to speech ith
these properties than to speech intended for adults )/ernald, +,0?, +,,=; 7irsh:Pasek, "elson,
Ausc(yk, *assidy, &russ, E Jennedy, +,0-2.
<n all speech, a number of prosodic properties of the speech ave, such as lengthening, intonation, and
pausing, are influenced by the syntactic structure of the sentence )*ooper E Paccia:*ooper, +,0C2.
Aust listen to ho you ould say the ord like in the sentence ;he boy < like slept compared to ;he
boy < sa likes sleds. <n the first sentence, the ord like is at the boundary of a relative clause and is
dran out, exaggerated in intonation, and folloed by a pause; in the second, it is in the middle of a
verb phrase and is pronounced more quickly, uniformly in intonation, and is run together ith the
folloing ord. 9ome psychologists )e.g., Dleitman E #anner, +,0?; Dleitman, +,,C2 have
suggested that children use this information in the reverse direction, and read the syntactic structure of
a sentence directly off its melody and timing. #e ill examine the hypothesis in 9ection .
+.) %onte't
*hildren do not hear sentences in isolation, but in a context. "o child has learned language from the
radio; indeed, children rarely if ever learn language from television. Ervin:;ripp )+,-12 studied
hearing children of deaf parents hose only access to English as from radio or television broadcasts.
;he children did not learn any speech from that input. 4ne reason is that ithout already knoing the
language, it ould be difficult for a child to figure out hat the characters in the unresponsive
+.
televised orlds are talking about. <n interacting ith live human speakers, ho tend to talk about the
here and no in the presence of children, the child can be more of a mind:reader, guessing hat the
speaker might have meant )%acnamara, +,-=, +,0=; 9chlesinger, +,-+2. ;hat is, before children have
learned syntax, they kno the meaning of many ords, and they might be able to make good guesses
as to hat their parents are saying based on their knoledge of ho the referents of these ords
typically act )for example, people tend to eat apples, but not vice:versa2. <n fact, parental speech to
young children is so redundant ith its context that a person ith no knoledge of the order in hich
parents! ords are spoken, only the ords themselves, can infer from transcripts, ith high accuracy,
hat as being said )9lobin, +,--2.
%any models of language acquisition assume that the input to the child consists of a sentence and a
representation of the meaning of that sentence, inferred from context and from the child!s knoledge
of the meanings of the ords )e.g. $nderson, +,--; 6erick, +,0>; Pinker, +,0=, +,0?; #exler E
*ulicover, +,0C2. 4f course, this can!t literally be true :: children don!t hear every ord of every
sentence, and surely don!t, to begin ith, perceive the entire meaning of a sentence from context.
6lind children, hose access to the nonlinguistic orld is obviously severely limited, learn language
ithout many problems )Landau E Dleitman, +,0.2. $nd hen children do succeed in guessing a
parent!s meaning, it can!t be by simple temporal contiguity. /or example, Dleitman )+,,C2 points out
that hen a mother arriving home from ork opens the door, she is likely to say, '#hat did you do
today3,' not <!m opening the door. 9imilarly, she is likely to say 'Eat your peas' hen her child is, say,
looking at the dog, and certainly not hen the child is already eating peas.
9till, the assumption of context:derived semantic input is a reasonable ideali(ation, if one considers
the abilities of the hole child. ;he child must keep an updated mental model of the current situation,
created by mental faculties for perceiving ob5ects and events and the states of mind and
communicative intentions of other humans. ;he child can use this knoledge, plus the meanings of
any familiar ords in the sentence, to infer hat the parent probably meant. <n 9ection e ill discuss
ho children might fill the important gaps in hat they can infer from context.
. *hat and *hen %hildren Learn
People do not reproduce their parents! language exactly. <f they did, e ould all still be speaking like
*haucer. 6ut in any generation, in most times, the differences beteen parents! language and the one
their children ultimately acquire is small. $nd remember that, 5udging by their spontaneous speech,
e can conclude that most children have mastered their mother tongue )alloing for performance
errors due to complexity or rarity of a construction2 some time in their threes. <t seems that the success
criterion for human language is something close to full mastery, and in a short period of time.
;o sho that young children really have grasped the design plan of language, rather than merely
approximating it ith outardly:convincing routines or rules of thumb hich ould have to be
supplanted later in life, e can!t 5ust rely on hat they say; e need to use clever experimental
techniques. Let!s look at to examples that illustrate ho even very young children seem to obey the
innate complex design of 8niversal Drammar.
Earlier < mentioned that in all languages, if there are derivational affixes that build ne ords out of
old ones, like :ism, :er, and :able, and inflectional affixes that modify a ord according to its role in
the sentence, like :s, :ed, and :ing, then the derivational affix appears inside the inflectional one:
&arinisms is possible, &arinsism is not. ;his and many other grammatical quirks ere nicely
explained in a theory of ord structure proposed by Paul Jiparsky )+,0=2.
Jiparsky shoed that ords are built in layers or 'levels.' ;o build a ord, you can start ith a root
)like &arin2. ;hen you can rules of a certain kind to it, called 'Level + Fules,' to yield a more
+>
complex ord. /or example, there is a rule adding the suffix :ian, turning the ord into &arinian.
Level + Fules, according to the theory, can affect the sound of the stem; in this case, the syllable
carrying the stress shifts from &ar to in. Level = rules apply to a ord after any Level + rules have
been applied. $n example of a Level = rule is the one that adds the suffix :ism, yielding, for example,
&arinism. Level = rules generally do not affect the pronunciation of the ords they apply to; they
5ust add material onto the ord, leaving the pronunciation intact. );he stress in &arinism is the same
as it as in &arin.2 /inally, Level 1 rules apply to a ord after any Level = rules have been applied.
;he regular rules of inflectional morphology are examples of Level 1 rules. $n example is the rule
that adds an :s to the end of a noun to form its plural :: for example, &arinians or &arinisms.
*rucially, the rules cannot apply out of order. ;he input to a Level + rules must be a ord root. ;he
input to a level = rule must be either a root or the output of Level + rules. ;he input to a Level 1 rule
must be a root, the output of Level + rules, or the output of Level = rules. ;hat constraint yields
predictions about hat kinds of ords are possible and hich are impossible. /or example, the
ordering makes it impossible to derive &arinianism and &arinianisms, but not &arinsian,
&arinsism, and &arinismian.
"o, irregular inflection, such as the pairing of mouse ith mice, belongs to Level +, hereas regular
inflectional rules, such as the one that relates rat to rats, belongs to Level 1. *ompounding, the rule
that ould produce &arin:lover and mousetrap, is a Level = rule, in beteen. ;his correctly predicts
that an irregular plural can easily appear inside a compound, but a regular plural cannot. *ompare the
folloing:
ice:infested )4J2; rats:infested )bad2
men:bashing )4J2; guys:bashing )bad2
teethmarks )4J2; clasmarks )bad2
feet:armer )4J2; hand:armer )bad2
purple people:eater )4J2; purple babies:eater )bad2
%ice:infested is a possible ord, because the process connecting mouse ith mice comes before the
rule combining the noun ith infested. 7oever, rats:infested, even though it is cognitively quite
similar to mice:infested, sounds strange; e can say only rat:infested )even though by definition one
rat does not make an infestation2.
Peter Dordon )+,0>2 had children beteen the ages of 1 and . participate in an elicited:production
experiment in hich he ould say, '7ere is a puppet ho likes to eat PPPPP. #hat ould you call
him3' 7e provided a response for several singular mass nouns, like mud, beforehand, so that the
children ere aare of the existence of the 'x:eater' compound form. *hildren behaved 5ust like
adults: a puppet ho likes to eat a mouse as called a mouse:eater, a puppet ho likes to eat a rat as
called a rat:eater, a puppet ho likes to eat mice as called either a mouse:eater or a mice:eater :: but
:: a puppet ho likes to eat rats as called a rat:eater, never a rats:eater. <nterestingly, children treated
their on overregulari(ations, such as mouses, exactly as they treated legitimate regular plurals: they
ould never call the puppet a mouses:eater, even if they used mouses in their on speech.
Even more interestingly, Dordon examined ho children could have acquired the constraint. Perhaps,
he reasoned, they had learned the fact that compounds can contain either singulars or irregular plurals,
never regular plurals, by paying keeping track of all the kinds of compounds that do and don!t occur in
their parents! speech. <t turns out that they ould have no ay of learning that fact. $lthough there is
no grammatical reason hy compounds ould not contain irregular plurals, the speech that most
children hear does not contain any. *ompounds like toothbrush abound; compounds containing
irregular plurals like teethmarks, people:eater, and men:bashing, though grammatically possible, are
statistically rare, according to the standardi(ed frequency data that Dordon examined, and he found
none that as likely to appear in the speech children hear. ;herefore children ere illing to say
+-
mice:eater and unilling to say rats:eater ith no good evidence from the input that that is the pattern
required in English. Dordon suggests that this shos that the constraints on level:ordering may be
innate.
Let!s no go from ords to sentences. 9entence are ordered strings of ords. "o child could fail to
notice ord order in learning and understanding language. 6ut most regularities of language govern
hierarchically:organi(ed structures :: ords grouped into phrases, phrases grouped into clauses,
clauses grouped into sentences )see the *hapters by Lasnik, by Larson, and by "eport E Dleitman2.
<f the structures of linguistic theory correspond to the hypotheses that children formulate hen they
analy(e parental speech and form rules, children should create rules defined over hierarchical
structures, not simple properties of linear order such as hich ord comes before hich other ord or
ho close to ords are in a sentence. ;he chapter by Dleitman and "eport discusses one nice
demonstration of ho adults )ho are, after all, 5ust gron:up children2 respect constituent structure,
not simple ord order, hen forming questions. 7ere is an example making a similar point that has
been tried out ith children.
Languages often have embedded clauses missing a sub5ect, such as Aohn told %ary to leave, here the
embedded 'donstairs' clause to leave has no sub5ect. ;he phenomenon of control governs ho the
missing sub5ect is interpreted. <n this sentence it is %ary ho is understood as having the embedded
sub5ect!s role, that is, the person doing the leaving. #e say that the phrase %ary 'controls' the missing
sub5ect position of the loer clause. /or most verbs, there is a simple principle defining control. <f the
upstairs verb has no ob5ect, then the sub5ect of the upstairs verb controls the missing sub5ect of the
donstairs verb. /or example, in Aohn tried to leave, Aohn is interpreted as the sub5ect of both try and
leave. <f the upstairs verb has a sub5ect and an ob5ect, then it is the ob5ect that controls the missing
sub5ect of the donstairs verb, as e sa in Aohn told %ary to leave.
<n +,>,, *arol *homsky published a set of classic experiments in developmental psycholinguistics.
9he shoed that children apply this principle quite extensively, even for the handful of verbs that are
exceptions to it. <n act:out comprehension experiments on children beteen the ages of . and +C, she
shoed that even relatively old children ere prone to this kind of mistake. #hen told '%ickey
promised &onald to 5ump; %ake him 5ump,' the children made &onald, the ob5ect of the first verb, do
the 5umping, in accord ith the general principle. ;he 'right anser' in this case ould have been
%ickey, because promise is an exception to the principle, calling for an unusual kind of control here
the sub5ect of the upstairs verb, not the ob5ect of the upstairs verb, should act as controller.
6ut hat, exactly, is the principle that children are over:applying3 4ne possibility can be called the
%inimal &istance Principle: the controller of the donstairs verb is the noun phrase nearest to it in the
linear string of ords in the sentence. <f children analy(e sentences in terms of linear order, this
should be a natural generali(ation. 7oever, it isn!t right for the adult language. *onsider the passive
sentence %ary as told by Aohn to leave. ;he phrase Aohn is closest to the sub5ect position for leave,
but adult English speakers understand the sentence as meaning that %ary is the one leaving. ;he
%inimal &istance Principle gives the rong anser here. <nstead, for the adult language, e need a
principle sensitive to grammatical structure, such as the 'c:control' structural relation discussed in the
*hapter by Lasnik Q3R. Let!s consider a simplified version, hich e can call the 9tructural Principle.
<t might say that the controller of a missing sub5ect is the grammatical ob5ect of the upstairs verb if it
has one; otherise it is the grammatical sub5ect of the upstairs verb )both of them c:command the
missing sub5ect2. ;he ob5ect of a preposition in the higher clause, hoever, is never alloed to be a
controller, basically because it is embedded 'too deeply' in the sentence!s tree structure to c:command
the missing sub5ect. ;hat!s hy %ary as told by Aohn to leave has %ary as the controller. )<t is also
hy, incidentally, the sentence %ary as promised by Aohn to leave is unintelligible :: it ould
require a prepositional phrase to be the controller, hich is ruled out by the 9tructural Principle.2
+0
<t ould certainly be understandable if children ere to follo the %inimal &istance Principle. "ot
only is it easily stated in terms of surface properties that children can easily perceive, but sentences
that ould disconfirm it like %ary as told by Aohn to leave are extremely rare in parents! speech.
%ichael %aratsos )+,-?2 did the crucial experiment. 7e gave children such sentences and asked them
ho as leaving. 4f course, on either account children ould have to be able to understand the
passive construction to interpret these sentences, and %aratsos gave them a separate test of
comprehension of simple passive sentences to select out only those children ho could do so. $nd
indeed, he found that those children interpreted passive sentences ith missing embedded sub5ects 5ust
as adults ould. ;hat is, in accord ith the 9tructural Principle and in violation of the %inimal
&istance Principle, they interpreted %ary as told by Aohn to leave as having the sub5ect, %ary, do
the leaving; that is, as the controller. ;he experiment shos ho young children have grasped the
abstract structural relations in sentences, and have acquired a grammar of the same design as that
spoken by their parents.
/ The %hild0s Language1Learning Algorith$
7ere is the most basic problem in understanding ho children learn a language: ;he input to language
acquisition consists of sounds and situations; the output is a grammar specifying, for that language,
the order and arrangement of abstract entities like nouns, verbs, sub5ects, phrase structures, control,
and c:command )see the *hapters by Lasnik and Larson, and the demonstrations in this chapter and
the one by Dleitman and "eport2. 9omeho the child must discover these entities to learn the
language. #e kno that even preschool children have an extensive unconscious grasp of grammatical
structure, to the experiments on discussed in the previous section, but ho has the child managed to
go from sounds and situations to syntactic structure3
<nnate knoledge of grammar itself is not sufficient. <t does no good for the child to have ritten
don in his brain ';here exist nouns'; children need some ay of finding them in parents! speech, so
that they can determine, among other things, hether the nouns come before the verb, as in English,
or after, as in <rish. 4nce the child finds nouns and verbs, any innate knoledge ould immediately
be helpful, because the child could then deduce all kinds of implications about ho they can be used.
6ut finding them is the crucial first step, and it is not an easy one.
<n English, nouns can be identified as those things that come after articles, get suffixed ith :s in the
plural, and so on. 6ut the infant obviously doesn!t kno that yet. "ouns don!t occur in any constant
position in a sentence across the languages of the orld, and they aren!t said ith any particular tone
of voice. "or do nouns have a constant meaning :: they often refer to physical things, like dogs, but
don!t have to, as in ;he days of our lives and ;he armth of the sun. ;he same is true for other
linguistic entities, such as verbs, sub5ects, ob5ects, auxiliaries, and tense. 9ince the child must
someho 'lift himself up by his bootstraps' to get started in formulating a grammar for the language,
this is called the 'bootstrapping problem' )see Pinker, +,0=, +,0?, +,0-b, +,0,, +,,?; %organ, +,0>;
Dleitman, +,,C; and the contributors to %organ and &emuth, +,,.2. 9everal solutions can be
envisioned.
/.1 E'tracting #i$(le %orrelations
4ne possibility is that the child sets up a massive correlation matrix, and tallies hich ords appear in
hich positions, hich ords appear next to hich other ords, hich ords get hich prefixes and
suffixes in hich circumstances, and so on. 9yntactic categories ould arise implicitly as the child
discovered that certain sets of properties are mutually intercorrelated in large sets of ords. /or
example, many ords tend to occur beteen a sub5ect and an ob5ect, are inflected ith :s hen the
sub5ect is singular and in the third person and the tense is present, and often appear after the ord to.
+,
;his set of ords ould be grouped together as the equivalent of the 'verb' category )%aratsos E
*halkley, +,0+2.
;here are to problems ith this proposal. ;he main one is that the features that the prelinguistic
child is supposed to be cross:referencing are not audibly marked in parental speech. Father, they are
perceptible only to child ho has already analy(ed the grammar of the language :: 5ust hat the
proposal is trying to explain in the first placeK 7o is a prelinguistic child supposed to find the
'sub5ect' of the sentence in order to correlate it ith the ending on the ords he or she is focusing on3
$ sub5ect is not the same thing as the first ord or to of the sentence )e.g., ;he big bad olf huffed
and puffed2 or even the first phrase )e.g., #hat did the big bad olf do32. #e have a dilemma. <f the
features defining the ros and columns of the correlation matrix are things that are perceptible to the
child, like 'first ord in a sentence,' then grammatical categories ill never emerge, because they
have no consistent correlation ith these features. 6ut if the features are the things that do define
grammatical categories, like agreement and phrase structure position, the proposal assumes 5ust hat
it sets out to explain, namely that the child has analy(ed the input into its correct grammatical
structures. 9omeho, the child must break into this circle. <t is a general danger that pops up in
cognitive psychology henever anyone proposes a model that depends on correlations among
features: there is alays a temptation to glibly endo the features ith the complex, abstract
representations hose acquisition one is trying to explain.
;he second problem is that, ithout prior constraints on the design of the feature:correlator, there are
an astronomical number of possible intercorrelations among linguistic properties for the child to test.
;o take 5ust to, the child ould have to determine hether a sentence containing the ord cat in
third position must have a plural ord at the end, and hether sentences ending in ords ending in d
are invariably preceded by ords referring to plural entities. %ost of these correlations never occur in
any natural language. <t ould be mystery, then, hy children are built ith complex machinery
designed to test for them :: though another ay of putting it is that it ould be a mystery hy there
are no languages exhibiting certain kinds of correlations given that children are capable of finding
them.
/.2 2sing ,rosody
$ second ay in hich the child could begin syntax learning ould be to attend to the prosody of
sentences, and to posit phrase boundaries at points in the acoustic stream marked by lengthening,
pausing, and drops in fundamental frequency. ;he proposal seems attractive, because prosodic
properties are perceptible in advance of knoing any syntax, so at first glance prosody seems like a
straightforard ay for a child to break into the language system.
6ut on closer examination, the proposal does not seem to ork )Pinker, +,0-, +,,?b; /ernald and
%cFoberts, in press; 9teedman, in press2. Aust as gold glitters, but all that glitters is not gold, syntactic
structure affects aspects of prosody, but aspects of prosody are affected by many things besides
syntax. ;he effects of emotional state of the speaker, intent of the speaker, ord frequency, contrastive
stress, and syllabic structure of individual ords, are all mixed together, and there is no ay for a
child to disentangle them from the sound ave alone. /or example, in the sentence ;he baby ate the
slug, the main pause coincides ith the ma5or syntactic boundary beteen the sub5ect and the
predicate. 6ut a child cannot ork backards and assume that the main pause in an input sentence
marks the boundary beteen the sub5ect and the predicate. <n the similar sentence 7e ate the slug, the
main pause is at the more embedded boundary beteen the verb and its ob5ect.
#orse, the mapping beteen syntax and prosody, even hen it is consistent, is consistent in different
ays in different languages. 9o a young child cannot use any such consistency, at least not at the very
=C
beginning of language acquisition, to decipher the syntax of the sentence, because it itself is one of the
things that has to be learned.
/.! 2sing %onte't and #e$antics
$ third possibility )see Pinker, +,0=, +,0?, +,0,; %acnamara, +,0=; Drimsha +,0+; #exler E
*ulicover, +,0C; 6loom, in press2 exploits the fact that there is a one:ay contingency beteen syntax
and semantics in the basic sentences of most of the orld!s languages. ;hough not all nouns are
physical ob5ects, all physical ob5ects are named by nouns. 9imilarly, if a verb has an argument playing
the semantic role of !agent!, then that argument ill be expressed as the sub5ect of basic sentences in
language after language. )$gain, this does not ork in reverse: the sub5ect is not necessarily an agent.
<n Aohn liked %ary the sub5ect is an 'experiencer'; in Aohn pleased %ary it is an ob5ect of experience;
in Aohn received a package it is a goal or recipient; in Aohn underent an operation it is a patient.2
9imilarly, entities directly affected by an action are expressed as ob5ects )but not all ob5ects are entities
affected by an action2; actions themselves are expressed as verbs )though not all verbs express
actions2. Even phrase structure configurations have semantic correlates: arguments of verbs reliably
appear as 'sisters' to them inside the verb phrase in phrase structure trees )see the chapter by Lasnik2.
<f children assume that semantic and syntactic categories are related in restricted ays in the early
input, they could use semantic properties of ords and phrases )inferred from context; see 9ection 2 as
evidence that they belong to certain syntactic categories. /or example, a child can infer that a ord
that designated a person, place or thing is a noun, that a ord designating an action is a verb, that a
ord expressing the agent argument of an action predicate is the sub5ect of its sentence, and so on. /or
example, upon hearing the sentence ;he cat chased the rat, the child can deduce that in English the
sub5ect comes before the verb, that the ob5ect comes after the verb, and so on. ;his ould give the
child the basis for creating the phrase structure trees that allo him or her to analy(e the rules of the
language.
4f course, a child cannot literally create a grammar that contains rules like '$gent ords come before
action ords.' ;his ould leave the child no ay of knoing ho to order the ords in sentences
such as $pples appeal to %ary or Aohn received a package. 6ut once an initial set of rules is learned,
items that are more abstract or that don!t follo the usual patterns relating syntax and semantic could
be learned through their distribution in already:learned structures. ;hat is, the child could no infer
that $pples is the sub5ect of appeal, and that Aohn is the sub5ect of receive, because they are in sub5ect
position, a fact the child no knos thanks to the earlier cat:chased:rat sentences. 9imilarly, the child
could infer that appeal is a verb to begin ith because it is in the 'verb' position.
3 Acquisition in Action
#hat do all these arguments mean for hat goes on in a child!s mind moment by moment as he or she
is acquiring rules from parental speech3 Let!s look at the process as concretely as possible.
3.1 Bootstra((ing the 4irst 5ules
/irst imagine a hypothetical child trying to extract patterns from the folloing sentences, ithout any
innate guidance as to ho human grammar orks.
%yron eats lamb.
%yron eats fish.
%yron likes fish.
$t first glance, one might think that the child could analy(e the input as follos. 9entences consist of
three ords: the first must be %yron, the second either eats or likes, the third lamb or fish. #ith these
=+
micro:rules, the child can already generali(e beyond the input, to the brand ne sentence %yron likes
chicken.
6ut let!s say the next to sentences are
%yron eats loudly.
%yron might fish.
;he ord might gets added to the list of ords that can appear in second position, and the ord loudly
is added to the list that can appear in third position. 6ut look at the generali(ations this ould allo:
%yron might loudly.
%yron likes loudly.
%yron might lamb.
;his is not orking. ;he child must couch rules in grammatical categories like noun, verb, and
auxiliary, not in actual ords. ;hat ay, fish as a noun and fish as a verb can be kept separate, and the
child ould not adulterate the noun rule ith instances of verbs and vice:versa. <f children are illing
to guess that ords for ob5ects are nouns, ords for actions are verbs, and so on, they ould have a
leg up on the rule:learning problem.
6ut ords are not enough; they must be ordered. <magine the child trying to figure out hat kind of
ord can occur before the verb bother. <t can!t be done:
;hat dog bothers me. Qdog, a nounR
#hat she ears bothers me. Qears, a verbR
%usic that is too loud bothers me. Qloud, an ad5ectiveR
*heering too loudly bothers me. Qloudly, an adverbR
;he guy she hangs out ith bothers me. Qith, a prepositionR
;he problem is obvious. ;here is a certain something that must come before the verb bother, but that
something is not a kind of ord; it is a kind of phrase, a noun phrase. $ noun phrase alays contains a
head noun, but that noun can be folloed by many other phrases. 9o it is useless of try to learn a
language by analy(ing sentences ord by ord. ;he child must look for phrases :: and the
experiments on grammatical control discussed earlier shos that they do.
#hat does it mean to look for phrases3 $ phrase is a group of ords. %ost of the logically possible
groups of ords in a sentence are useless for constructing ne sentences, such as ears bothers and
cheering too, but the child, unable to rely on parental feedback, has no ay of knoing this. 9o once
again, children cannot attack the language learning task like some logician free of preconceptions;
they need prior constraints. #e have already seen here such constraints could come. /irst, the child
could assume that parents! speech respects the basic design of human phrase structure: phrases contain
heads )e.g., a noun phrase is built around a head noun2; arguments are grouped ith heads in small
phrases, sometimes called @:bars )see the chapter by Lasnik2; @:bars are grouped ith their modifiers
inside large phrases )"oun Phrase, Berb Phrase, and so on2; phrases can have sub5ects. 9econd, since
the meanings of parents! sentences are guessable in context, the child could use the meanings to help
set up the right phrase structure. <magine that a parent says ;he big dog ate ice cream. <f the child
already knos the ords big, dog, ate, and ice cream, he or she can guess their categories and gro
the first branches of a tree: <n turn, nouns and verbs must belong to noun phrases and verb phrases, so
the child can posit one for each of these ords. $nd if there is a big dog around, the child can guess
that the and big modify dog, and connect them properly inside the noun phrase: <f the child knos that
the dog 5ust ate ice cream, he or she can also guess that ice cream and dog are arguments of the verb
eat. &og is a special kind of argument, because it is the causal agent of the action and the topic of the
sentence, and hence it is likely to be the sub5ect of the sentence, and therefore attaches to the '9.' $
tree for the sentence has been completed: ;he rules and dictionary entries can be peeled off the tree:
==
9 ::S "P BP
"P ::S )det2 )$2 "
BP ::S B "P
dog: "
ice cream: "
ate: B; eater T sub5ect, thing eaten T ob5ect
the: det
big: $
;his hypothetical example shos ho a child, if suitably equipped, could learn three rules and five
ords from a single sentence in context.
;he use of part:of:speech categories, phrase structure, and meaning guessed from context are
poerful tools that can help the child in the daunting task of learning grammar quickly and ithout
systematic parental feedback )Pinker, +,0?2. <n particular, there are many benefits to using a small
number of categories like " and B to organi(e incoming speech. 6y calling both the sub5ect and ob5ect
phrases '"P,' rather than, say PhraseU+ and PhraseU=, the child automatically can apply knoledge
about nouns in sub5ect position to nouns in ob5ect position, and vice:versa. /or example, our model
child can already generali(e, and use dog as an ob5ect, ithout having heard an adult do so, and the
child tacitly knos that ad5ectives precede nouns not 5ust in sub5ects but in ob5ects, again ithout
direct evidence. ;he child knos that if more than one dog is dogs in sub5ect position, more than one
dog is dogs in ob5ect position.
%ore generally, English allos at least eight possible phrasemates of a head noun inside a noun
phrase, such as Aohn!s dog; dogs in the park; big dogs; dogs that < like, and so on. <n turn, there are
about eight places in a sentence here the hole noun phrase can go, such as &og bites man; %an
bites dog; $ dog!s life; Dive the boy a dog; ;alk to the dog; and so on. ;here are three ays to inflect a
noun: dog, dogs, dog!s. $nd a typical child by the time he or she is in high school has learned
something like =C,CCC different nouns )%iller, +,,+; Pinker, +,,?a2. <f children had to learn all the
combinations separately, they ould need to listen to about +?C million different sentences. $t a rate
of a sentence every ten seconds, ten hours a day, it ould take over a century. 6ut by unconsciously
labeling all nouns as '"' and all noun phrases as '"P,' the child has only to hear about tenty:five
different kinds of noun phrase and learn the nouns one by one, and the millions of possible
combinations fall out automatically.
<ndeed, if children are constrained to look for only a small number of phrase types, they automatically
gain the ability to produce an infinite number of sentences, one of the hallmarks of human language.
;ake the phrase the tree in the park. <f the child mentally labels the park as an "P, and also labels the
tree in the park as an "P, the resulting rules generate an "P inside a PP inside an "P :: a loop that can
be iterated indefinitely, as in the tree near the ledge by the lake in the park in the city in the east of the
state .... <n contrast, a child ho as free to to label in the park as one kind of phrase, and the tree in
the park another, ould be deprived of the insight that the phrase contains an example of itself. ;he
child ould be limited to reproducing that phrase structure alone.
#ith a rudimentary but roughly accurate analysis of sentence structure set up, the other parts of
language can be acquired systematically. $bstract ords, such as nouns that do not refer to ob5ects and
people, :: can be learned by paying attention to here they sit inside a sentence. 9ince situation in ;he
situation 5ustifies drastic measures occurs inside a phrase in "P position, it must be a noun. <f the
language allos phrases to be scrambled around the sentence, like Latin or the $ustralian aboriginal
language #arlpiri, the child can discover this feature upon coming across a ord that cannot be
connected to a tree in the expected place ithout crossing branches )in 9ection , e ill see that
children do seem to proceed in this order2. ;he child!s mind can also kno hat to focus on in
decoding case and agreement inflections: a noun!s inflection can be checked to see if it appears
=1
henever the noun appears in sub5ect position, in ob5ect position, and so on; a verb!s inflection might
can be checked for tense, aspect, and the number, person, and gender of its sub5ect and ob5ect. ;he
child need not bother checking hether the third ord in the sentence referred to a reddish or a bluish
ob5ect, hether the last ord as long or short, hether the sentence as being uttered indoors or
outdoors, and billions of other fruitless possibilities that a purely correlational learner ould have to
check.
3.2 The 6rgani7ation of ra$$ar as a uide to Acquisition
$ grammar is not a bag of rules; there are principles that link the various parts together into a
functioning hole. ;he child can use such principles of 8niversal Drammar to allo one bit of
knoledge about language to affect another. ;his helps solve the problem of ho the child can avoid
generali(ing to too large a language, hich in the absence of negative evidence ould be incorrigible.
<n cases ere children do overgenerali(e, these principles can help the child recover: if there is a
principle that says that $ and 6 cannot coexist in a language, a child acquiring 6 can use it to catapult
$ out of the grammar.
3.2.1 Bloc8ing and Inflectional 6verregulari7ation
;he next chapter presents a good example. ;he 6locking principle in morphology dictates that an
irregular form listed in the mental dictionary as corresponding to a particular inflectional category
)say, past tense2, blocks the application of the corresponding general rule. /or example, adults kno
the irregular form broke, and that prevents them from applying the regular 'add :ed' rule to break and
saying breaked. *hildren, ho have not heard broke enough times to remember it reliably on demand,
thus fail to block the rule and occasionally say breaked. $s they hear broke enough times to recall it
reliably, 6locking ould suppress the regular rule, and they ould gradually recover from these
overgenerali(ation errors )%arcus, et al., +,,=2.
3.2.2 Interactions between *ord "eaning and #ynta'
7ere is another example in hich a general principle rules out a form in the adult grammar, but in the
child!s grammar, the crucial information alloing the principle to apply is missing. $s the child!s
knoledge increases, the relevance of the principle to the errant form manifests itself, and the form
can be ruled out so as to make the grammar as a hole consistent ith the principle.
Every verb has an 'argument structure': a specification of hat kinds of phrases it can appear ith
)Pinker, +,0,2. $ familiar example is the distinction beteen a transitive verb like devour, hich
requires a direct ob5ect )you can say 7e devoured the steak but not 5ust 7e devoured2 and an
intransitive verb like dine, hich does not )you can say 7e dined but not 7e dined the steak2. *hildren
sometimes make errors ith the argument structures of verbs that refer to the act of moving something
to a specified location )6oerman, +,0=b; Dropen, Pinker, 7ollander, and Doldberg, +,,+a2:
< didn!t fill ater up to drink it; < filled it up for the floers to drink it.
*an < fill some salt into the bear3 Qa bear:shaped salt shakerR
<!m going to cover a screen over me.
/eel your hand to that.
;erri said if this Qa rhinestone on a shirtR ere a diamond then people ould be trying to rob the shirt.
$ general principle of argument structure is that the argument that is affected in some ay specified
by the verb gets mapped onto the syntactic ob5ect. ;his is an example of a 'linking rule,' hich links
semantics ith syntax )and hich is an example of the contingency a young child ould have
employed to use semantic information to bootstrap into the syntax2. /or example, for adults, the
'container' argument )here the ater goes2 is the direct ob5ect of fill :: fill the glass ith ater, not
=?
fill ater into the glass :: because the mental definition of the verb fill says that the glass becomes
full, but says nothing about ho that happens )one can fill a glass by pouring ater into it, by dripping
ater into it, by dipping it into a pond, and so on2. <n contrast, for a verb like pour, it is the 'content'
argument )the ater2 that is the ob5ect :: pour ater into the glass, not pour the glass ith ater ::
because the mental definition of the verb pour says that the ater must move in a certain manner
)donard, in a stream2 but does not specify hat happens to the container )the ater might fill the
glass, merely et it, end up beside it, and so on2. <n both cases, the entity specified as 'affected' ends
up as the ob5ect, but for fill, it is the ob5ect hose state is affected )going from not full to full2,
hereas for pour, it is the ob5ect hose location is affected )going from one place to a loer one2.
"o, let!s say children mistakenly think that fill refers to a manner of motion )presumably, some kind
of tipping or pouring2, instead of an end state of fullness. )*hildren commonly use end state verbs as
manner verbs: for example, they think that mix 5ust means stir, regardless of hether the stirred
ingredients end up mixed together; Dentner, +,-02. <f so, the linking rule for direct ob5ects ould
cause them to make the error e observe: fill x into y. 7o could they recover3 #hen children
observe the verb fill in enough contexts to reali(e that it actually encodes the end state of fullness, not
a manner of pouring or any other particular manner )for example eventually they may hear someone
talking about filling a glass by leaving it on a indo sill during a storm2, they can change their
mental dictionary entry for fill. $s a result, they ould ithdra it from eligibility to take the
argument structure ith the contents as direct ob5ect, on the grounds that it violates the constraint that
'direct ob5ect T specifically affected entity.' ;he principle could have existed all along, but only been
deemed relevant to the verb fill hen more information about its definition had been accumulated
)Dropen, et al., +,,+a, b; Pinker, +,0,2.
;here is evidence that the process orks in 5ust that ay. Dropen et al. )+,,+a2 asked preschool
children to select hich picture corresponded to the sentence 9he filled the glass ith ater. %ost
children indiscriminately chose any picture shoing ater pouring; they did not care hether the
glass ended up full. ;his shos that they do misconstrue the meaning of fill. <n a separate task, the
children ere asked to describe in their on ords hat as happening in a picture shoing a glass
being filled. %any of these children used incorrect sentences like 7e!s filling ater into the glass.
4lder children tended to make feer errors of both verb meaning and verb syntax, and children ho
got the verb meaning right ere less likely to make syntax errors and vice:versa. <n an even more
direct demonstration, Dropen, et al. )+,,+b2 taught children ne verbs like to pilk, referring to actions
like moving a sponge over to a cloth. /or some children, the motion had a distinctive (ig(ag manner,
but the cloth remained unchanged. /or others, the motion as nondescript, but the cloth changed color
in a litmus:like reaction hen the sponge ended up on it. ;hough none of the children heard the verb
used in a sentence, hen asked to describe the event, the first group said that the experimenter as
pilking the sponge, hereas the second group said that he as pilking the cloth. ;his is 5ust the kind
of inference that ould cause a child ho finally figured out hat fill means to stop using it ith the
rong direct ob5ect.
<nterestingly, the connections beteen verbs! syntax and semantics go both ays. Dleitman )+,,C2
points out that there are some aspects of a verb!s meaning that are difficult, if not impossible, for a
child to learn by observing only the situations in hich the verb is used. /or example, verb pairs like
push and move, give and receive, in and beat, buy and sell, chase and flee, and drop and fall often
can be used to describe the same event; only the perspective assumed by the verb differs. $lso, mental
verbs like see, kno, and ant, are difficult to infer by merely observing their contexts. Dleitman
suggests that the crucial missing information comes from the syntax of the sentence. /or example, fall
is intransitive )it fell, not Aohn fell the ball2; drop can be transitive )7e dropped the ball2. ;his reflects
the fact that the meaning of fall involves the mere act of plummeting, independent of ho if anyone
caused it, hereas the extra argument of drop refers to an agent ho is causing the descent. $ child
could figure out the meaning difference beteen the to by paying attention to the transitive and
=.
intransitive syntax :: an example of using syntax to learn semantics, rather than vice:versa. )4f course,
it can only ork if the child has acquired some syntax to begin ith.2 9imilarly, a verb that appears
ith a clause as its complement )as in < think that ...2 must refer to a state involving a proposition, and
not, say, of motion )there is no verb like 7e 5umped that he as in the room2. ;herefore a child
hearing a verb appearing ith a clausal complement can infer that it might be a mental verb.
"aigles )+,,C2 conducted an experiment that suggest that children indeed can learn some of a verb!s
meaning from the syntax of a sentence it is used in. ;enty:four:month:olds first sa a video of a
rabbit pushing a duck up and don, hile both made large circles ith one arm. 4ne group of
children heard a voice saying ';he rabbit is gorping the duck'; another heard ';he rabbit and the duck
are gorping.' ;hen both groups sa a pair of screens, one shoing the rabbit pushing the duck up and
don, neither making arm circles, the other shoing the to characters making arm circles, neither
pushing don the other. <n response to the command '#here!s gorping no3 /ind gorpingK', the
children ho heard the transitive sentence looked at the screen shoing the up:and:don action, and
the children ho heard the intransitive sentence looked at the screen shoing the making:circles
action. /or a general discussion of ho children could use verb syntax to learn verb semantics, and
vice:versa, see Pinker )+,,?b2.
3.! ,ara$eter1#etting and the #ubset ,rinci(le
$ striking discovery of modern generative grammar is that natural languages seem to be built on the
same basic plan. %any differences among languages represent not separate designs but different
settings of a fe 'parameters' that allo languages to vary, or different choices of rule types from a
fairly small inventory of possibilities. ;he notion of a 'parameter' is borroed from mathematics. /or
example, all of the equations of the form 'y T 1x N b,' hen graphed, correspond to a family of
parallel lines ith a slope of 1; the parameter b takes on a different value for each line, and
corresponds to ho high or lo it is on the graph. 9imilarly, languages may have parameters )see the
chapter by Lasnik2.
/or example, all languages in some sense have sub5ects, but there is a parameter corresponding to
hether a language allos the speaker to omit the sub5ect in a tensed sentence ith an inflected verb.
;his 'null sub5ect' parameter )sometimes called 'PF4:drop'2 is set to 'off' in English and 'on' in
9panish and <talian )*homsky, +,0+2. <n English, one can!t say Does to the store, but in 9panish, one
can say the equivalent. ;he reason this difference is a 'parameter' rather than an isolated fact is that it
predicts a variety of more subtle linguistic facts. /or example, in null sub5ect languages, one can also
use sentences like #ho do you think that left3 and $te Aohn the apple, hich are ungrammatical in
English. ;his is because the rules of a grammar interact tightly; if one thing changes, it ill have
series of cascading effects throughout the grammar. )/or example, #ho do you think that left3 is
ungrammatical in English because the surface sub5ect of left is an inaudible 'trace' left behind hen
the underlying sub5ect, ho, as moved to the front of the sentence. /or reasons e need not cover
here, a trace cannot appear after a ord like that, so its presence taints the sentence. Fecall that in
9panish, one can delete sub5ects. ;herefore, one can delete the trace sub5ect of left, 5ust like any other
sub5ect )yes, one can 'delete' a mental symbol even it ould have made no sound to begin ith2. ;he
is trace no longer there, so the principle that disallos a trace in that position is no longer violated,
and the sentence sounds fine in 9panish.
4n this vie, the child ould set parameters on the basis of a fe examples from the parental input,
and the full complexity of a language ill ensue hen those parameteri(ed rules interact ith one
another and ith universal principles. ;he parameter:setting vie can help explain the universality
and rapidity of the acquisition of language, despite the arcane complexity of hat is and is not
grammatical )e.g., the ungrammaticality of #ho do you think that left32. #hen children learn one fact
=>
about a language, they can deduce that other facts are also true of it ithout having to learn them one
by one.
;his raises the question of ho the child sets the parameters. 4ne suggestion is that parameter settings
are ordered, ith children assuming a particular setting as the default case, moving to other settings as
the input evidence forces them to )*homsky, +,0+2. 6ut ho ould the parameter settings be
ordered3 4ne very general rationale comes from the fact that children have no systematic access to
negative evidence. ;hus for every case in hich parameter setting $ generates a subset of the
sentences generated by setting 6 )as in diagrams )c2 and )d2 of /igure +2, the child must first
hypothesi(e $, then abandon it for 6 only if a sentence generated by 6 but not by $ as encountered
in the input )Pinker, +,0?; 6erick, +,0.; 4sherson, et al, +,0.2. ;he child ould then have no need
for negative evidence; he or she ould never guess too large a language. )/or settings that generate
languages that intersect or are dis5oint, as in diagrams )a2 and )b2 of /igure +, either setting can be
discarded if incorrect, because the child ill eventually encounter a sentence that one grammar
generates but the other does not2.
%uch interesting research in language acquisition hinges on hether children!s first guess from among
a set of nested possible languages really is the smallest subset. /or example, some languages, like
English, mandate strict ord orders; others, such as Fussian or Aapanese, list a small set of admissible
orders; still others, such as the $ustralian aborigine language #arlpiri, allo almost total scrambling
of ord order ithin a clause. #ord order freedom thus seems to be a parameter of variation, and the
setting generating the smallest language ould obviously be the one dictating fixed ord order. <f
children follo the 9ubset Principle, they should assume, by default, that languages have a fixed
constituent order. ;hey ould back off from that prediction if and only if they hear alternative ord
orders, hich indicate that the language does permit constituent order freedom. ;he alternative is that
the child could assume that the default case as constituent order freedom.
<f fixed:order is indeed the default, children should make fe ord order errors for a fixed:order
language like English, and might be conservative in learning freer:ord order languages, sticking ith
a subset of the sanctioned orders )hether they in fact are conservative ould depend on ho much
evidence of multiple orders they need before leaping to the conclusion that multiple orders are
permissible, and on ho frequent in parental speech the various orders are2. <f, on the other hand, free:
order is the default, children acquiring fixed:ord:order languages might go through a stage of
overgenerating )saying, give doggie paper; give paper doggie, paper doggie give; doggie paper give,
and so on2, hile children acquiring free ord:order languages ould immediately be able to use all
the orders. <n fact, as < have mentioned, children learning English never leap to the conclusion that it
is a free:ord order language and speak in all orders )6ron, +,-1; 6raine, +,->; Pinker, +,0?;
6loom, Lightbon, E 7ood, +,-.2. Logically speaking, though, that ould be consistent ith hat
they hear if they ere illing to entertain the possibility that their parents ere 5ust conservative
speakers of Jorean, Fussian or 9edish, here several orders are possible. 6ut children learning
Jorean, Fussian, and 9edish do sometimes )though not alays2 err on the side of caution, and use
only one of the orders alloed in the language, pending further evidence )6ron, +,-12. <t looks like
fixed:order is the default, 5ust as the 9ubset Principle ould predict.
#exler E %an(ini )+,0-2 present a particularly nice example concerning the difference beteen
'anaphors' like herself and 'pronouns' like her. $n anaphor has to be have its antecedent lie a small
distance aay )measured in terms of phrase si(e, of course, not number of ords2; the antecedent is
said to be inside the anaphor!s 'governing category.' ;hat is hy the sentence Aohn liked himself is
fine, but Aohn thought that %ary liked himself is ungrammatical: himself needs an antecedent )like
Aohn2 ithin the same clause as itself, hich it has in the first example but not the second. &ifferent
languages permit different:si(e governing categories for the equivalents of anaphors like himself; in
some languages, the translations of both sentences are grammatical. ;he 9ubset Principle predicts that
=-
children should start off assuming that their language requires the tiniest possible governing category
for anaphors, and then to expand the possibilities outard as they hear the telltale sentences.
<nterestingly, for pronouns like 'her,' the ordering is predicted to be the opposite. Pronouns may not
have an antecedent ithin their governing categories: Aohn liked him )meaning Aohn liked himselfR is
ungrammatical, because the antecedent of him is too close, but Aohn thought that %ary liked him is
fine. 9ets of languages ith bigger and bigger governing categories for pronouns allo feer and
feer grammatical possibilities, because they define larger ranges in hich a pronoun prohibits its
antecedent from appearing :: an effect of category si(e on language si(e that is in the opposite
direction to hat happens for anaphors. #exler and %an(ini thus predict that for pronouns, children
should start off assuming that their language requires the largest possible governing category, and then
to shrink the possibilities inard as they hear the telltale sentences. ;hey revie experiments and
spontaneous speech studies that provide some support for this subtle pattern of predictions.
19 %onclusion
;he topic of language acquisition implicate the most profound questions about our understanding of
the human mind, and its sub5ect matter, the speech of children, is endlessly fascinating. 6ut the
attempt to understand it scientifically is guaranteed to bring on a certain degree of frustration.
Languages are complex combinations of elegant principles and historical accidents. #e cannot design
ne ones ith independent properties; e are stuck ith the confounded ones entrenched in
communities. *hildren, too, ere not designed for the benefit of psychologists: their cognitive, social,
perceptual, and motor skills are all developing at the same time as their linguistic systems are
maturing and their knoledge of a particular language is increasing, and none of their behavior
reflects one of these components acting in isolation.
Diven these problems, it may be surprising that e have learned anything about language acquisition
at all, but e have. #hen e have, < believe, it is only because a diverse set of conceptual and
methodological tools has been used to trap the elusive ansers to our questions: neurobiology,
ethology, linguistic theory, naturalistic and experimental child psychology, cognitive psychology,
philosophy of induction, theoretical and applied computer science. Language acquisition, then, is one
of the best examples of the indispensability of the multidisciplinary approach called cognitive science.
11 4urther 5eading
$ general introduction to language can be found in my book ;he Language <nstinct )Pinker, +,,?2,
from hich several portions of this chapter ere adapted. ;here is a chapter on language acquisition,
and chapters on syntactic structure, ord structure, universals and change, prescriptive grammar,
neurology and genetics, and other topics.
;he logical problem of language acquisition is discussed in detail by #exler and *ulicover )+,0C2,
Pinker )+,-,, +,0?, +,0-, +,0,2, 4sherson, 9tob, E #einstein )+,0.2, 6erick )+,0.2, and %organ
)+,0>2. Pinker )+,-,2 is a nontechnical introduction. ;he study of learnability ithin theoretical
computer science has recently taken on interesting ne turns, revieed in Jearns E Ba(irani )+,,?2,
though ith little discussion of the special case e are interested in, language acquisition. 6rent
)+,,.2 contains state:of:the:art ork on computer models of language acquisition.
;he most comprehensive recent textbook on language development is <ngram )+,0,2. $mong other
recent textbooks, Dleason )+,,12 has a focus on children!s and mothers! behavior, hereas $tkinson
)+,,=2, Doodluck )+,,+2, and *rain and Lillo:%artin )in press2, have more of a focus on linguistic
theory. 6loom )+,,12 is an excellent collection of reprinted articles, organi(ed around the acquisition
of ords and grammar. 7oekstra and 9chart( )+,,?2 is a collection of recent papers more closely
=0
tied to theories of generative grammar. /letcher E %ac#hinney!s ;he 7andbook of *hild Language
)+,,.2, has many useful survey chapters; see also the surveys by Paul 6loom in Dernsbacher!s
7andbook of Psycholinguistics )+,,?2 and by %ichael %aratsos in %ussen!s *armichael!s %anual of
*hild Psychology )?th edition +,01; .th edition in preparation at the time of this riting2.
Earlier collections of important articles include Jrasnegor, et al., )+,,+2, %ac#hinney )+,0-2, Foeper
E #illiams )+,0-2, #anner E Dleitman )+,0=2, 6aker E %c*arthy )+,0+2, /letcher and Darman
)+,-,2, /erguson E 9lobin )+,-12, 7ayes )+,-C2, 6ron E 6ellugi )+,>?2, and Lenneberg )+,>?2.
9lobin )+,0.aM+,,12 is a large collection of ma5or revies on the acquisition of particular languages.
;he most ambitious attempts to synthesi(e large amounts of data on language development into a
cohesive frameork are 6ron )+,-12, Pinker )+,0?2, and 9lobin )+,0.b2. *lark )+,,12 revies the
acquisition of ords. Locke )+,,12 covers the earliest stages of acquisition, ith a focus on speech
input and output. %organ E &emuth )in press2 contains papers on children!s perception of input
speech and its interaction ith their language development.
12 ,roble$s
+. '"egative evidence' is reliable information available to a language learner about hich strings
of ords are ungrammatical in the language to be acquired. #hich of the folloing ould, and
ould not, count as negative evidence. Austify your ansers.
a. %other expresses disapproval every time Aunior speaks ungrammatically.
b. /ather often reards Aunior hen he speaks grammatically, and often punishes him hen he
speaks ungrammatically.
c. %other rinkles her nose every time Aunior speaks ungrammatically, and never rinkles her
nose any other time.
d. /ather repeats all of Aunior!s grammatical sentences verbatim, and converts all of his
ungrammatical sentences into grammatical ones.
e. %other blathers incessantly, uttering all the grammatical sentences of English in order of
length :: all the to ord sentences, then all the three:ord sentences, and so on.
f. /ather corrects Aunior henever he produces an overregulari(ation like breaked, but never
corrects him hen he produces a correct past tense form like broke.
g. #henever Aunior speaks ungrammatically, %other responds by correcting the sentence to
the grammatical version. #hen he speaks grammatically, %other responds ith a follo:up
that merely recasts the sentence in different ords.
h. #henever Aunior speaks ungrammatically, /ather changes the sub5ect.
i. %other never repeats Aunior!s ungrammatical sentences verbatim, but sometimes repeats his
grammatical sentences verbatim.
5. /ather blathers incessantly, producing all possible strings of English ords, furroing his
bros after every ungrammatical string and pursing his lips after every grammatical sentence.
=,
=. *onsider three languages. Language $ is is English, in hich sentence must contain a
grammatical sub5ect: 7e ate the apple is good; $te the apple is ungrammatical. <n Language 6,
the sub5ect is optional, but the verb alays has a suffix hich agrees ith the sub5ect )hether
it is present or absent2 in person, number, and gender. ;hus 7e ate:1%9 the apple is good
)assume that '1%9' is a suffix, like :o or :ik, that is used only hen the sub5ect is 1rd person
masculine singular2, as is $te:1%9 the apple. );hose of you ho speak 9panish or <talian ill
see that this hypothetical language is similar to them.2 Language * has no inflection on the
verb, but allos the sub5ect to be omitted: 7e ate the apple and $te the apple are both good.
$ssuming a child has no access to negative evidence, but knos that the language to be
learned is one of these three. &oes the child have to entertain these hypotheses in any fixed
order3 <f so, hat is it3 #hat learning strategy ould guarantee that the child ould arrive at
the correct language3 9ho hy.
1. <magine a verb pilk that means 'to have both of one!s elbos grabbed by someone else,' so
Aohn pilked 6ill meant that 6ill grabbed Aohn!s elbos.
a. #hy is this verb unlikely to occur in English3
b. <f children use semantic context and semantic:syntax linking rules to bootstrap their ay
into a language, hat ould a languageless child infer about English upon hearing ';his is
pilking' and seeing 6ill grab Aohn!s elbos3
c. <f children use semantic context and semantics:syntax linking rules to bootstrap their ay
into a language, hat ould a languageless child infer about English upon hearing 'Aohn
pilked 6ill' and seeing 6ill grab Aohn!s elbos3
d. <f children use semantic context and semantics:syntax linking rules to bootstrap their ay
into a language, hat ould a child have to experience in order to learn English syntax and the
correct use of the ord pilk3
1! Answers to ,roble$s
+. a. "o. Presumably %other also expresses disapproval for other reasons, such as Aunior uttering
a rude or false :: but grammatical :: sentence. <f Aunior ere to assume that disapproved:of
sentences ere ungrammatical, he ould spuriously eliminate many grammatical sentences
from his language.
b. "o, because /ather may also reard him hen he speaks ungrammatically and punish him
hen he speaks grammatically.
c. Les, because Aunior can deduce that any nose:rinkle:eliciting sentence is grammatical.
d. Les, because Aunior can deduce that any sentence that is not repeated verbatim is
ungrammatical.
e. Les, because for any sentence that Aunior is unsure about, he can keep listening to mother
until she begins to utter sentences longer than that one. <f, by that time, %other has uttered his
sentence, it is grammatical; if she hasn!t, it!s ungrammatical.
f. "o, because e don!t kno hat /ather does for the rest of the language.
g. "o, because hile e kno hether the changeover in Aunior!s sentence is a 'correction' or
a 'recasting,' because e kno hat!s ungrammatical )hence corrected2 or grammatical
1C
)hence recast2, Aunior has no ay of knoing that from his point of vie, %other 5ust changes
everything he says into different ords.
h. "o, because presumably /ather changes the sub5ect on some occasions hen Aunior!s
sentence as grammatical but /ather as 5ust getting bored ith the topic.
i. "o, because many of his grammatical sentences might never be repeated verbatim, either.
5. Les, because sooner or later /ather ill utter Aunior!s last ord string, and Aunior can see
hether /ather!s bro as furroed.
=. English )Language $2 has to be hypothesi(ed before Language *, and re5ected only if a
sub5ectless and suffixless sentence turns up in the input. ;hat is because Language * is a
superset of English; if the learner tries * first, nothing in the input ill ever tell him he!s
rong. Language 6 can be hypothesi(ed at any point, and confirmed henever the child hears
a sentence ith an agreement in it or disconfirmed hen the child hears a sentence ithout
agreement.
1. a. <n English )and almost every other language2, the agent of the action is the sub5ect of an
active sentence, and the entity affected by the action is the ob5ect.
b. 7e ould infer, incorrectly, that pilk means 'to hold someone!s elbos.'
c. 7e ould infer, incorrectly, that English ord order as 4b5ect:Berb
9ub5ect. ;hat ould cause him subsequently to apply universals about sub5ects to ob5ects, and
vice:versa.
d. 7e ould have to have heard enough ordinary English verbs )ith agents as sub5ects and
affected entities as ob5ects2 to have inferred that the sub5ect comes before the verb, hich in
turn comes before the ob5ect. ;hen he ould have to hear Aohn pilked 6ill and see 6ill grab
Aohn!s elbos, and use the verb!s syntax to infer its unusual semantics.
5eferences
$nderson, A. )+,--2 <nduction of augmented transition netorks. *ognitive 9cience, +, +=.:+.-.
6ates, E. )+,0,2 /unctionalism and the competition model. <n 6. %ac#hinney and E. 6ates )Eds.2,
;he crosslinguistic study of sentence processing. "e Lork: *ambridge 8niversity Press.
6ates, E., ;hal, &., E Aanosky, A. 9. )+,,=2 Early language development and its neural correlates. <n
<. Fapin E 9. 9egaloit( )Eds.2, 7andbook of "europsychology. Bol. >, *hild "eurology.
$msterdam: Elsevier.
6ellugi, 8., 6ihrle, $., Aernigan, ;., ;rauner, &., E &oherty, 9. )+,,C2 "europsychological,
neurological, and neuroanatomical profile of #illiams 9yndrome. $merican Aournal of %edical
Denetics 9upplement, >, ++.:+=..
6erick, F. *. )+,0.2 ;he acquisition of syntactic knoledge. *ambridge, %$: %<; Press.
6ickerton, &. )+,0?2 ;he language bioprogram hypothesis. 6ehavioral and 6rain 9ciences, -, +-1:
==+.
1+
6loom, L. )+,-C2 Language &evelopment: /orm and /unction in Emerging Drammars.
*ambridge, %$: %<; Press.
6loom, L., Lightbon, P., E 7ood, %. )+,-.2 9tructure and variation in child language. %onographs
of the 9ociety for Fesearch in *hild &evelopment, vol. ?C.
6loom, P. )in press2 6ohannon, A. "., and 9tanoic(, L. )+,002 ;he issue of negative evidence: $dult
responses to children!s language errors. &evelopmental Psychology =?: >0?:>0,.
6oerman, %. )+,0=2. Evaluating competing linguistic models ith language acquisition data:
<mplications of developmental errors ith causative verbs. Huaderni di 9emantica 1: .:>>.
6oerman, %. )+,0=b2 Feorgani(ational processes in lexical and syntactic development. <n E.
#anner E L. Dleitman )eds.2, Language $cquisition: ;he 9tate of the $rt. "e Lork: *ambridge
8niversity Press.
6raine, %. &. 9. )+,->2 *hildren!s first ord combinations. %onographs of the 9ociety for Fesearch
in *hild &evelopment, ?+.
6rent, %. )Ed.2 )in press2 9pecial issue of *ognition on *omputational %odels of Language
$cquisition.
6ron, F. E 6ellugi, 8. )Eds.2 )+,>?2 9pecial issue of 7arvard Educational Fevie.
6ron, F., E 7anlon, *. )+,-C2. &erivational complexity and order of acquisition in child speech. <n
A. F. 7ayes )Ed.2, *ognition and the &evelopment of Language. "e Lork: #iley.
6ron, F. )+,-12 $ /irst Language: the Early 9tages. *ambridge, %ass.: 7arvard 8niversity Press.
*homsky, *. )+,>,2 $cquisition of 9yntax in *hildren from . : +C. *ambridge, %$: %<; Press.
*homsky, ". )+,.,2 $ Fevie of 6. /. 9kinner!s 'Berbal 6ehavior.' Language, 1., =>:.0.
*homsky, ", )+,-.2 Feflections on Language. "e Lork: Fandom 7ouse.
*homsky, ". )+,0+2 Lectures on Dovernment and 6inding. &ordrecht, "etherlands: /oris
Publications.
*homsky, ". )+,,+2 Linguistics and cognitive science: Problems and mysteries. <n $. Jasher, )Ed.2,
;he *homskyan turn. *ambridge, %$: 6lackell.
*lark, E. B. )+,,12 ;he lexicon in acquisition. "e Lork: *ambridge 8niversity Press.
*omrie, 6. )+,0+2 Language universals and linguistic typology. *hicago: 8niversity of *hicago Press.
*ooper, #. E. and Paccia:*ooper, A. )+,0C2 9yntax and 9peech. *ambridge, %ass.: 7arvard
8niversity Press.
*rain, 9. )+,,=2 Language acquisition in the absence of experience. 6ehavioral and 6rain 9ciences.
*rain, 9., E Lillo:%artin, &. )in press2 Language acquisition. *ambridge, %$: 6lackell.
1=
*romer, F. /. )+,,=2 Language and thought in normal and handicapped children.
*ambridge, %$: 6lackell.
*urtiss, 9. )+,0,2 ;he independence and task:specificity of language. <n $. 6ornstein E A. 6runer
)Eds.2, <nteraction in human development. 7illsdale, "A: Erlbaum.
&emetras, %. A., Post, J. ", E 9no, *. E. )+,0>2 /eedback to first language learners: ;he role of
repetitions and clarification questions. Aournal of *hild Language, +1, =-.:=,=.
Ervin:;ripp, 9. )+,-12 9ome strategies for the first to years. <n ;. <n ;. E. %oore. )Ed.2, *ognitive
&evelopment and the $cquisition of Language. "e Lork: $cademic Press.
/erguson, *. and 9lobin, &. <., )Eds.2 )+,-12, 9tudies of *hild Language &evelopment. "e Lork:
7olt, Finehart and #inston.
/ernald, $. )+,0?2 the perceptual and affective salience of mothers! speech to infants. <n L. /eagans,
*. Darvey, E F. Dolinkoff )Eds.2, ;he origins and groth of communication. "orood, "A: $blex.
/ernald, $. )+,,=2 7uman maternal vocali(ations to infants as biologically relevant signals: $n
evolutionary perspective. <n 6arko, et al. <n A. 7. 6arko, L. *osmides, E A. ;ooby )Eds.2, ;he
$dapted %ind: Evolutionary Psychology and the Deneration of *ulture. "e Lork: 4xford
8niversity Press.
/ernald, $., E %cFoberts, D. )in press2 Prosodic bootstrapping: $ critical analysis of the argument
and the evidence. <n A. L. %organ E J. &emuth, Eds. 9ignal to syntax. 7illsdale, "A: Erlbaum.
/isher, *., 7all, D., Fakoit(, 9., E Dleitman, L. F. )+,,+2 #hen it is better to receive than to give:
9yntactic and conceptual constraints on vocabulary groth. 8npublished manuscript, &epartment of
Psychology, 8niversity of <llinois, 8rbana.
/letcher and %. Darman, )Eds.2 )+,-,2 Language $cquisition. "e Lork: *ambridge 8niversity
Press.
/letcher, P. and %ac#hinney, 6. )Eds.2 )+,,.2 ;he handbook of child language. *ambridge, %$:
6lackell.
/odor, A. $. )+,-.2 ;he Language of ;hought. "e Lork: ;. L. *roell.
/odor, A. $. )+,012 %odularity of mind. *ambridge, %$: 6radford 6ooksM%<; Press.
Dardner, F. $. E Dardner, 6. ;. )+,>,2 ;eaching sign language to a chimpan(ee. 9cience, +>., >>?:
>-=.
Dentner, &. )+,-02 4n relational meaning: ;he acquisition of verb meaning. *hild &evelopment ?,:
,00:,,0.
Dernsbacher, %. $. )Ed.2 )+,,?2 7andbook of psycholinguistics. 9an &iego: $cademic Press.
Dleason, A. 6erko )Ed.2 )+,,12 ;he development of language, 1rd edition. "e Lork: %acmillan.
11
Dleitman, L. F. and #anner, E. )+,0?2 Fichly specified input to language learning. <n 4. 9elfridge, E.
L. Fissland, E %. $rbib )Eds.2, $daptive control of ill:defined systems. "e Lork: Plenum.
Dold, E. )+,>-2 Language identification in the limit. <nformation and *ontrol, +C, ??-:?-?.
Doodluck, 7. )+,,+2 Language $cquisition: a linguistic introduction. *ambridge, %$: 6lackell.
Dreenberg, A. )Ed.2 )+,-02 8niversals of human language. Bol. ?: 9yntax. 9tanford, *$: 9tanford
8niversity Press.
Drimsha, A. )+,0+2 /orm, function, and the language acquisition device. <n *. L. 6aker and A.
%c*arthy )eds.2, ;he Logical Problem of Language $cquisition. *ambridge, %$: %<; Press.
Dropen, A. Pinker, 9., 7ollander, %., E Doldberg, F. )+,,+a2 9yntax and semantics in the acquisition
of locative verbs. Aournal of *hild Language, +0, ++.:+.+.
Dropen, A., Pinker, 9., 7ollander, %., E Doldberg, F. )+,,+b2 $ffectedness and direct ob5ects: ;he
role of lexical semantics in the acquisition of verb argument structure. *ognition, ?+, +.1:+,..
7ayes, A. F. )Ed.2, *ognition and the &evelopment of Language. "e Lork: #iley.
7eath, 9. 6. )+,012 #ays ith ords: Language, life, and ork in communities and classrooms. "e
Lork: *ambridge 8niversity Press.
7irsh:Pasek, J., "elson, &. D. "., Ausc(yk, P. #., *assidy, J. #., &russ, 6., E Jennedy, L. )+,0-2
*lauses are perceptual units for young infants. *ognition, =>, =>,:=0>.
7irsh:Pasek, J., ;reiman, F., E 9chneiderman, %. )+,0?2. 6ron and 7anlon revisited: %others!
sensitivity to ungrammatical forms. Aournal of *hild Language, ++, 0+:00.
7irsh:Pasek, J., E Dolinkoff, F. %. )+,,+2 Language comprehension: $ ne look at some old
themes. <n ". Jrasnegor, &. %. Fumbaugh, F. L. 9chiefelbusch, E %. 9tuddert:Jennedy )Eds.2
)+,,+2 6iological and behavioral determinants of language development. 7<llsdale, "A: Erlbaum.
7oekstra, ;. E 9chart(, 6. )Eds.2 )+,,?2 Language acquisition studies in generative grammar.
Philadelphia: Aohn 6en5amins.
7uttenlocher, P. F. )+,,C2 %orphometric study of human cerebral cortex development.
"europsychologia, =0, .+-:.=-.
<ngram., &. )+,0,2 /irst language acquisition: %ethod, description, and explanation. "e Lork:
*ambridge 8niversity Press.
Jearns, %. A., E Ba(irani, 8. B. )+,,?2. $n <ntroduction to *omputational Learning ;heory.
*ambridge, %$: %<; Press.
Jegl, A. )+,,?2 ;he "icaraguan 9ign Language Pro5ect: an overvie. 9ignpost, -, 1=:1,.
Jiparsky, P. )+,0=2 Lexical phonology and morphology. <n <. 9. Lang )Ed.2, Linguistics in the
morning calm. 9eoul: 7ansin, pp. 1:,+.
1?
Jrasnegor, ". $., Fumbaugh, &. %., 9chiefelbusch, F. L., E 9tuddert:Jennedy, %. )eds.2 )+,,+2
6iological and behavioral determinants of language development. 7<llsdale, "A: Erlbaum.
Juhl, P., #illiams, J. $., Lacerda, /., 9tevens, J. "., E Lindblom, 6. )+,,=2 Linguistic experience
alters phonetic perception in infants by six months of age. 9cience, =.., >C>:>C0.
Labov, #. )+,>,2 ;he logic of nonstandard English. Deorgeton %onographs on Language and
Linguistics, ==, +:1+.
Landau, 6. E Dleitman, L. F. )+,0.2 Language and experience. *ambridge, %$: 7arvard 8niversity
Press.
Lenneberg, E. 7. )Ed.2 )+,0?2 "e directions in the study of language. *ambridge, %$: %<; Press.
Lenneberg, E. 7. )+,>-2 6iological /oundations of Language. "e Lork: #iley.
Levy, L. )+,012 <t!s frogs all the ay don. *ognition, +., -.:,1.
Lieberman, P. )+,0?2 ;he biology and evolution of language. *ambridge, %$: 7arvard 8niversity
Press.
Limber, A. )+,-12 ;he genesis of complex sentences. <n ;. E. %oore., *ognitive &evelopment and the
$cquisition of Language. "e Lork: $cademic Press.
Locke, A. L. )+,,=2 9tructure and stimulation in the ontogeny of spoken language. &evelopmental
Psychobiology, =0 , ?1C:??C.
Locke, A. L. )+,,12 ;he child!s path to spoken language. *ambridge, %$: 7arvard 8niversity Press.
%ac#hinney, 6. E 9no, *. )+,0.2 ;he *hild Language &ata Exchange 9ystem. Aournal of *hild
Language, +=, =-+:=,>.
%ac#hinney, 6. E 9no, *.)+,,C2 ;he *hild Language &ata Exchange 9ystem: $n update. Aournal
of *hild Language, +-, ?.-:?-=.
%ac#hinney, 6. )Ed.2 )+,0-2 %echanisms of language acquisition. 7illsdale, "A: Erlbaum.
%ac#hinney, 6. )+,,+2 ;he *7<L&E9 Pro5ect: *omputational tools for analy(ing talk. 7illsdale, "A:
Erlbaum.
%acnamara, A. )+,-=2 *ognitive basis of language learning in infants. Psychological Fevie -,, +:+1.
%acnamara, A. )+,0=2 "ames for ;hings: a 9tudy of *hild Language. *ambridge, %ass.: 6radford
6ooksM%<; Press.
%aratsos, %. P. )+,-?a2 7o preschool children understand missing complement sub5ects. *hild
&evelopment, ?., -CC:-C>.
%aratsos, %. P., and *halkley, %. )+,0+2 ;he internal language of children!s syntax: the ontogenesis
and representation of syntactic categories. <n J. "elson , ed., *hildren!s Language vol.=. "e Lork:
Dardner Press.
1.
%arcus, D. /. )+,,12 "egative evidence in language acquisition. *ognition, ?>R, .1:0..
%arcus, D. /., Pinker, 9., 8llman, %., 7ollander, %., Fosen, ;. A., E @u, /. )+,,=2 4verregulari(ation
in language acquisition. %onographs of the 9ociety for Fesearch in *hild &evelopment, .-.
%iller, D. $. )+,,+2 ;he science of ords. "e Lork: #. 7. /reeman.
%organ, A. L. )+,0>2 /rom simple input to complex grammar. *ambridge, %$: 6radford 6ooksM%<;
Press.
%organ, A. L., E &emuth, J., )Eds.2 )in press2. 9ignal to syntax. 7illsdale, "A: Erlbaum.
%ussen, P. ed., *armichael!s %anual of *hild Psychology, ?th ed. "e Lork: #iley.
"eport, E., Dleitman, 7. E Dleitman, E. )+,--2 %other <!d rather do it myself: 9ome effects and
non:effects of maternal speech style. <n *. E. 9no and *. $. /erguson )eds.2, ;alking to *hildren:
Language <nput and $cquisition. *ambridge: *ambridge 8niversity Press.
4sherson, &. "., 9tob, %. E #einstein, 9. )+,0.2. 9ystems that learn. *ambridge, %$: 6radford
6ooksM%<; Press.
Penner, 9. )+,0-2 Parental responses to grammatical and ungrammatical child utterances. *hild
&evelopment .0: 1->:10?.
Piaget, A. )+,=>2 ;he language and thought of the child. "e Lork: Foutledge E Jegan Paul.
Pinker, 9. )+,-,2 /ormal models of language learning. *ognition, -, =+-:=01.
Pinker, 9. )+,0?2. Language learnability and language development. *ambridge, %$: 7arvard
8niversity Press. )a2
Pinker, 9. )+,0-2 ;he bootstrapping problem in language acquisition. <n 6. %ac#hinney )Ed.2,
%echanisms of language acquisition. 7illsdale, "A: Erlbaum.
Pinker, 9. )+,0,2 Learnability and cognition: ;he acquisition of argument structure. *ambridge, %$:
%<; Press.
Pinker, 9. )+,,?a2 ;he language instinct. "e Lork: %orro.
Pinker, 9. )+,,?b2 7o could a child use verb syntax to learn verb semantics3 Lingua, ,=, 1--:?+C.
;o be reprinted in L. Dleitman and 6. Landau )Eds.2, Lexical acquisition. *ambridge, %$: %<;
Press.
Pinker, 9., E 6loom, P. )+,,C2 "atural language and natural selection. 6ehavioral and 6rain 9ciences,
+1, -C-:-0?.
Premack, &. E Premack, $. A. )+,012 ;he mind of an ape. "e Lork: "orton.
Putnam, 7. )+,-+2 ;he 'innateness hypothesis' and explanatory models in linguistics. <n A. 9earle
)Ed.2, ;he philosophy of language. "e Lork: 4xford 8niversity Press.
;. Foeper E E. #illiams )Eds.2, Parameter:setting and language acquisition )Feidel, +,0-2.
1>
9avage:Fumbaugh, E. 9. )+,,+2 Language learning in the bonobo: 7o and hy they learn. <n ". $.
Jrasnegor, &. %. Fumbaugh, F. L. 9chiefelbusch, F. L., E %. 9tuddert:Jennedy, )Eds.2 6iological
and behavioral determinants of language development. 7<llsdale, "A: Erlbaum.
9chlesinger, <. %. )+,-+2 Production of utterances and language acquisition. <n &. <. 9lobin , ed., ;he
4ntogenesis of Drammar. "e Lork: $cademic Press.
9chieffelin, 6. E Eisenberg, $. F. )+,0+2 *ultural variation in children!s conversations. <n F. L.
9chiefelbusch E &. &. 6ricker )Eds.2, Early language: acquisition and intervention. 6altimore:
8niversity Park Press.
9eidenberg, %. 9. )+,0>2 Evidence from great apes concerning the biological bases of language. <n #.
&emopoulos E $. %arras )Eds.2, Language learning and concept acquisition. "orood, "A: $blex.
9eidenberg, %. 9. E Petitto, L. $. )+,-,2 9igning behavior in apes: $ critical revie. *ognition, -,
+--:=+..
9eidenberg, %. 9. E Petitto, L. $. )+,0-2 *ommunication, symbolic communication, and language:
*omment on 9avage:Fumbaugh, %c&onald, 9evcik, 7opkins, and Fupert )+,0>2 Aournal of
Experimental Psychology: Deneral, ++>, =-,:=0-.
9enghas, $. )+,,?2. "icaragua!s lessons for language acquisition. 9ignpost, -, 1=:1,.
9hopen, ;. )Ed.2 )+,0.2 Language typology and syntactic description. Bol. <<: *omplex *onstructions.
"e Lork: *ambridge 8niversity Press.
9lobin, &. )+,-12 *ognitive prerequisites for the development of grammar. <n. *. /erguson and &. <.
9lobin )ed.2, 9tudies in *hild Language &evelopment.
"e Lork: 7olt, Finehart and #inston.
9lobin, &. <. )+,--2 Language change in childhood and in history. <n A. %acnamara, ed., Language
Learning and ;hought. "e Lork: $cademic Press.
9lobin, &. <. )Ed.2 )+,0.aM+,,=2 ;he crosslinguistic study of language acquisition. Bols. + E =, +,0.;
Bol. 1, +,,=V. 7illsdale, "A: Erlbaum.
9lobin, &. <. )+,0.b2 *rosslinguistic evidence for the language:making capacity. <n &. <. 9lobin )Ed.2,
;he crosslinguistic study of language acquisition. Bol. <<: ;heoretical issues. 7illsdale, "A: Erlbaum.
9no, *. E., and /erguson, *. $. )+,--2 ;alking to *hildren: Language <nput and $cquisition.
*ambridge: *ambridge 8niversity Press.
9teedman, %. )in press2 Phrasal intonation and the acquisition of syntax. <n A. %organ E J. &emuth,
Eds. 9ignal to syntax. 7illsdale, "A: Erlbaum.
9tromsold, J. )+,,?2 #hat a mute child tells us about language. 8npublished manuscript, Futgers
8niversity.
;allal, P., Foss, F., *urtiss, 9.. Aournal of 9peech and 7earing &isorders, Q.?, +>-: .
1-
;errace, 7., Petitto, L. $., 9anders, F. A. E 6ever, ;. D. )+,-,2 *an an ape create a sentence3 9cience,
=C>, 0,+:,C=.
#anner, E. E Dleitman L. )Eds.2, Language $cquisition: ;he 9tate of the $rt. "e Lork: *ambridge
8niversity Press.
#exler, J. E *ulicover, P. )+,0C2 /ormal Principles of Language $cquisition. *ambridge, %$: %<;
Press.
#exler, J. E %an(ini, F. )+,0-2 Parameters and learnability in binding theory. <n ;. Foeper E E.
#illiams )Eds.2, Parameters and linguistic theory. &ordrecht: Feidel.
#horf, 6. )+,.>2 Language, thought, and reality. *ambridge, %$: %<; Press.
/igure *aption
/our situations that a child could be in hile learning a language. Each circle represents the set of
sentences constituting a language. '7' stands for 'hypothesi(ed language'; ';' stands for 'target
language.' 'N' indicates a grammatical sentence in the language; ':' indicates an ungrammatical
sentence.
10