See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/7340267

Language Acquisition is Language Change

Article  in  Journal of Psycholinguistic Research · February 2006

DOI: 10.1007/s10936-005-9002-7 · Source: PubMed

CITATIONS READS
50 611

3 authors:

Stephen Crain Takuya Goro

Macquarie University Tsuda College
211 PUBLICATIONS   6,862 CITATIONS    12 PUBLICATIONS   177 CITATIONS   

SEE PROFILE SEE PROFILE

Rosalind Thornton
Macquarie University
84 PUBLICATIONS   2,177 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Sentence Acceptability View project

Mandarin-speaking children's knowledge of sentence structures View project

All content following this page was uploaded by Stephen Crain on 06 May 2016.

The user has requested enhancement of the downloaded file.

Journal of Psycholinguistic Research,Volume 35, Issue 1 (January 2006)

Language Acquisition is Language Change

Stephen Crain+, Takuya Goro* and Rosalind Thornton+

+
Macquarie University
*University of Maryland at College Park

Abstract
According to the theory of Universal Grammar, the primary linguistic data guides children
through an innately specified space of hypotheses. On this view, similarities between child-
English and adult-German are as unsurprising as similarities between cousins who have never
met. By contrast, experience-based approaches to language acquisition contend that child
language matches the input, with nonadult forms being simply less articulated versions of the
forms produced by adults. This paper reports three studies that provide support for the theory of
Universal grammar, and resist explanation on experience-based accounts. Two studies
investigate English-speaking children’s productions, and the third examines the interpretation of
sentences by Japanese speaking children. When considered against the input children are
exposed to, the findings of these studies are consistent with the continuity hypothesis, which
supposes that child language can differ from the language spoken by adults only in ways that
adult languages can differ from each other.
From the perspective of many psycholinguists working within the theory of Universal Grammar,
many aspects of adult grammar are innate and are expected to be in place at a very early age.
Children are born with a set of universal linguistic principles, and a set of parameters that
account for variation among languages (Chomsky 1981). These innate linguistic parameters
define a space of possible human languages -- a space the child explores, influenced by her
environment, until she stabilizes on a grammar that is equivalent to that of adults in her linguistic
community. If this is correct then, at least as an idealization, language acquisition is a process of
language change. At any given time children are speaking a possible human language, just not
the language spoken around them. Of course, the adult languages spoken around the globe today
constitute only a small subset of the possible human languages. Nevertheless, one expects to find
many "linguistic errors" by children to arise because children are trying out grammars with
features that can be found in adult languages elsewhere on the globe. This acquisition scenario is
called the continuity hypothesis (Crain 1991; Crain & Pietroski 2002; Goodluck 1991; Pinker
1984). If the continuity hypothesis is confirmed, this would provide dramatic support for the
theory of Universal Grammar.

Alternative models of language development posit a greater role for experience in

children’s grammar formation and, consequently, less innate linguistic knowledge. On such
experience-based accounts of language development, children’s non-adult productions and
comprehension take on a different character than on the Universal Grammar-based account.
Children’s “linguistic errors” are expected to reflect less articulated grammars than those of
adults in the same linguistic community. Nevertheless, the child’s productions are modeled on
the environmental input. A child who is in the process of learning language would not yet
display full linguistic competence in any human language; such a child would have an imperfect
grasp of the local language, which he or she would gradually modify in response to the
environmental input. The child would not be expected to produce utterances that do not reflect
the target language, but properties of some other natural language. Nor would the child be
expected to interpret utterances in a way that is incorrect for the target language, but correct for
some other language.

The present paper summarizes some recent evidence that appears to favor the
continuity hypothesis, and which appears to resist explanation on many learning-theoretic
accounts. We document four cases of children’s productions, and one case of their
interpretations, which are not compatible with the target language. In each case, there is little or
no positive evidence that could cause children to adopt the hypothesis they entertain.
Furthermore, the evidence from children’s productions suggests that children are not simply
producing less articulated versions of adult utterances; instead, children are seen to elaborate the
structures used by adults. In each case, we suggest that children have adopted an option that is
manifested in some other language. These ‘linguistic errors’ can be thought of as ‘misset’
parameters in the theory of Universal Grammar.

One interesting characteristic of parameters is that particular settings sometimes entail a

cluster of properties. For example, Baker (2001) points out that once a child has settled on the
fact that she is speaking a polysynthetic language, she should also allow noun incorporation,
dislocated subjects and objects, certain causative structures and so on. This is also true of the two

2
production cases we discuss. We find that children produce a cluster of non-adult question
structures, which are typical of some language, just not the one being spoken around them.
In addition to these three cases, we briefly review two findings from the literature where children
appear to assume a more restricted grammatical hypothesis than their input would suggest. In
both cases again, the language children speak resembles other languages, just not the one spoken
around then. In the last section, we turn from production to comprehension. There we describe
another case of a misset parameter. This example of continuity involves Japanese-speaking
children’s interpretation of disjunction. We find that Japanese children interpret disjunction in
the same way as child and adult speakers of English, despite the absence of decisive evidence for
this interpretation in simple sentences.

1. The Medial-Wh in Child English

The parade case of children’s non-adult, but UG-compatible, productions is the "medial-wh"
phenomenon in child English. Using an elicited production task, Thornton (1990) found that
about one-third of the 3-4 year-old children (of English-speaking parents) she interviewed
consistently inserted an ‘extra’ wh-word in their long-distance questions, as illustrated in (1) and
(2) (Crain & Thornton, 1998; Thornton 1996).

(1) What do you think what pigs eat?

(2) Who did he say who is in the box?

This “linguistic error” by English-speaking children is presumably not a response to the

children's environment, since medial-wh constructions are not part of the primary linguistic data
for children in English-speaking environments. It is not an impoverished structure – there is an
extra word - but nor could these questions have been produced by merging the template for two
simple questions. Such a strategy could work for (2), but not for (1). If one were to merge “What
do you think?’ and ‘What do pigs eat?’ the result would be “What do you think what do pigs
eat?”. Questions like this with the dummy ‘do’ in each clause do not occur in these children’s
speech. Structures like (1) and (2) are attested in other languages, such as dialects of German and
Romani (McDaniel 1986, 1995). An example from German is given in (3).

(3) Weri glaubst du weri nach Hause geht?

‘Who do you think who goes home?’

The medial-wh phenomenon accords with the expectations of the theory of Universal Grammar.
And further investigation shows that the similarity of child English to a foreign language, adult
German, runs deep. Other similarities emerge. For both adult Germans and American children,
lexical (full, discourse-linked) wh-phrases cannot be repeated in the medial position. German-
speaking adults judge (4) to be unacceptable, and English-speaking children never produce
questions like (5), as indicated by the '#.' Instead, children either produce a question with a
reduced medial element like (6), or they produce the adult English structure with no medial-wh.

(4) *Wessen Buchi glaubst du wessen Buchi Hans liest?

‘Whose book do you think whose book Hans is reading?’

3
 (5) #Which Smurf do you think which Smurf is wearing roller skates
(6) Which Smurf do you think (who) is wearing roller skates?

Similarly, children never used a medial-wh when extracting from infinitival clauses. Nor is this
permissible in languages that permit the medial-wh. Indeed, insertion of medial-wh in infinitival
complements is universally ungrammatical. So in German, a question with no medial-wh would
be used, as in (8).

(7) #Who do you want who to win?

(8) Wen versucht Hans anzurufen

‘Whom is Hans trying to call?’

The wh-phrases that children consistently avoid in questions like (5) and (7) are well-formed
fragments of the local language; they appear in embedded questions: e.g., He asked me which
boy that is. I know who to play with. Therefore, these questions could be formed by the kinds of
‘cut-and-paste’ operations that experience-based approaches invoke to explain how complex
constructions are formed by combining simple constructions (e.g., Goldberg 2003; Tomasello
2000).

This complex pattern of linguistic behavior suggests that many children of English-
speakers go through a stage at which they speak a language that is like (adult) English in many
respects, but one that is like German (Romani etc.) in allowing for the medial-wh. There is
nothing wrong with such a language; it just so happens that adults in New York and London do
not speak it. But it is quite striking that many children of adults in New York and London don't
emulate their parents in this respect. Instead, these children speak like foreigners for a little
while. Nevertheless, even children who speak like foreigners in certain respects do not differ
from them in the acquisition of many other linguistic principles, including linguistic universals.
Learning-based approaches to language development owe an account of why children project
beyond the data in certain ways but not others.

2. Why-questions in Child English

We now turn to a second “linguistic error” by children that can be explained using the theory of
Universal grammar. The example is English-speaking children’s questions with the wh-word,
why. It has been observed repeatedly that why-questions in child English differ in form from
those produced by adults, and that these differences persist long after children have converged on
adult forms for other question words, e.g., what, who, where, how. The main observation in the
literature is that some English-speaking children’s why-questions lack inversion, as illustrated in
(9).

(9) a. Why that boy is looking at us? (2;4)

b. Why you gotta went to a conference? (2;5)
c. Why we can’t go in the upstairs bed? (2;5)
d. Why that’s not your sandwich? (2;5)
e. Why unicorns are pretend? (3;1)

4
 f. Why Sophie has a magic bed? (4;5)
g. Why Daddy’s never coming to watch? (4;3)

The source of children’s failure to invert auxiliary verbs in why-questions has been
extensively investigated (e.g., Berk 2003; de Villiers 1990; Labov & Labov 1978; Rowland &
Pine 2000; Stromswold 1990). As de Villiers (1990) and Berk (2003) note, the tendency for why
to lack inversion in child English is mirrored in other languages – notably in French questions
with pourquoi, and with Italian questions with perché. The cross-linguistic parallelism suggests
that English-speaking children may be drawing on an option made available by Universal
Grammar. If so, the expectation is that children’s why-questions should exhibit the cluster of
properties found in, say, adult Italian perché-questions. Such a pattern would constitute further
support for a UG-based approach to acquisition, and against a learning-theoretic account.

So, let’s take a brief look at Italian wh-questions. In our description of the facts, we lean
heavily on theoretical work by Rizzi (1997, 1999) and a longitudinal case study by Thornton
(2004). In Italian, wh-questions exhibit an operation much like subject-aux inversion in English,
except that the main verb raises to a position adjacent to the wh-element in Italian, whereas an
auxiliary verb assumes this position in English. For most wh-words in Italian, if the main verb is
not adjacent to the wh-word, the wh-question is ungrammatical, as shown in (10). There are two
exceptions to this, perché (why) and come mai (how come). For these wh-phrases, questions with
and without inversion are both grammatical, as (11a) and (11b) illustrate for perché.

(10) a. Che cosa ha fatto Gianni?

What has done Gianni
‘What did Gianni do?’

b. *Che cosa Gianni ha fatto?

What Gianni has done
‘What did Gianni do?’

(11) a. Perché Gianni è venuto?

Why Gianni has come
‘Why did Gianni come?’

b. Perché è venuto Gianni?

Why has come Gianni
‘Why did Gianni come?’

There are several other differences between perché on the one hand, and ‘ordinary’ wh-words,
such as che cosa (what), on the other. The observed differences between perché and ‘ordinary’
wh-words led Rizzi (1999) to propose that perché sits ‘higher’ in the left periphery than ordinary
wh-phrases, which move to CP (Rizzi’s FocusP; cf. Rizzi 1997). Because perché is positioned
higher in the phrase marker, there is considerable latitude in the elements it can be combined
with. With perché, other linguistic material can intrude between the wh-phrase and the main
verb. As one case in point, subordinate clauses may intrude, as illustrated in (12). Example (13)
shows that the intrusion of a subordinate clause is not permitted for che cosa.

5
 (12) Perché quando va a Milano Gianni compra il panettone?
Why when (he) goes to Milan Gianni buys the panettone
‘Why does Gianni buy panettone when he goes to Milan?’

(13) *Che cosa quando va a Milano compra Gianni?

What when (he) goes to Milan buys Gianni
‘What does Gianni buy when he goes to Milan?’

In addition, Italian perché-questions can be followed by phrases expressing contrastive focus, as

in (14) whereas ordinary wh-questions cannot be, as (15) indicates.

(14) Perché QUESTO avremmo dovuto dirgli, non qualcos’altro?

Why THIS (we) have should said, not something else
‘Why should we have said THIS to him and not something else?’

(15) *Che cosa A GIANNI hanno detto (non a Piero)?

What TO GIANNI (they) have said (not to Piero)
‘What have they said TO GIANNI (and not to Piero)?’

Italian perché can also be preceded (and followed by) Topic phrases, as illustrated in (16).

(16) Il mio libro, perché, a Gianni, non glielo avete ancora dato?
My book, why, to Gianni, not to-him-it (you) have still given
‘Why have you still not given my book to Gianni?’

Another interesting fact about Italian perché-questions concerns the interpretation of

questions with two clauses. If there is no inversion in the main clause, then only one reading of
the question is available. Thus, example (17) can only be asking why Gianni said he would
resign; it cannot be asking about the reason for his resignation. With inversion, however, the
question becomes ambiguous, as illustrated in (18). That is, (18) again has the interpretation we
saw for (17); but, in addition, it can be interpreted as asking about Gianni’s reason for resigning.
This ‘long distance’ interpretation of (18) is analyzed as movement of perché from the embedded
clause, by successive cyclic movement, to a landing site (the SpecCP position of the matrix
clause) where ordinary wh-phrases reside. For this interpretation to arise, inversion is required,
as it is for ordinary wh-phrases.

(17) Perché Gianni ha detto che si dimetterà?

Why Gianni has said that self will resign
‘Why did Gianni say that he will resign?’

(18) Perché ha detto che si dimetterà?

Why has (he) said that self will resign
‘Why did he say that he will resign?’

6
This cluster of properties of Italian wh-questions will suffice to examine English-speaking
children’s why questions, to see if they exhibit properties of Italian.

In a longitudinal study of one child, A.L., Thornton (2004) recorded over 900 why-
questions, when A.L. was between the ages of 2 and 6 years. The first prediction is that A.L.’s
why-questions should be compatible with a focus phrase immediately following why. The first
prediction is borne out, although the empirical support is limited. The diary corpus offers just
one example in which the phrase immediately following why expresses clear contrastive focus.

(19) Why SOME OF YOUR MAKE-UP I can’t use (and some I can) (5;2)

The second prediction is that A.L.’s why questions should be compatible with topic phrases
(preceding and) following why. There is abundant evidence for the second prediction. As in adult
Italian, A.L. allows topics in the position following why; time adverbials are particularly
plentiful. Topic phrases are italicized in the examples in (20).

(20) a. Why every winter there’s a snowstorm? (4;3)

b. Why every day when I wake up the hall light isn’t on? (5;1)
c. Why last time in New Zealand you didn’t give me a bath? (5;2)
d. Why everyday we’re still in the black car? (5;3)
e. Why at Falling Water kids can’t come? (6;3)
f. Why today we got a sticker instead of a sticker and a picture? (6;4)

The third prediction is that A.L.’s why-questions should allow subordinate clauses with if
and when preceding the main clause. Examples are provided in (21).

(21) a. Why when you was a kid people called you “Rozzy”? (3;5)
b. Why when I was a baby I loved Boomer’s dog food? (3;6)
c. Why last year when I invited Emily and Julie to my party Julie didn’t
play? (4;4)
d. Why if he goes to jail she can have his room? (5;0)
e. Why when I went through security it didn’t beep? (6;4)

According to the fourth prediction, an inverted auxiliary in A.L.’s two-clause questions should
express an interpretation that is not possible if inversion is absent. To investigate this prediction,
all wh-questions by A.L. with the matrix verbs think and say were counted from age 3 years to 5
and a half years. With wh-phrases other than why, inversion occurred 100% of the time. For
questions with why, only 4 of the 21 questions produced by A.L. lacked inversion. The four
questions at issue are given in (22), and a sample of A.L.s adult-like two clause questions with
why is given in (23). The expectation is that the four non-adult two clause why-questions in (22)
had only a ‘local’ reading questioning the ‘thinking’ or the ‘saying’; (22d) clearly supports this
interpretation. At any rate, 81% of A.L.’s two-clause why-questions with the verbs think/say
have do-support, despite the fact that A.L. continued to fail to invert the majority of the time in
one-clause why-questions.

7
 (22) a. Why you just think Boomer’s [the dog] cute? I’m cute too. (3;7)
b. Why you said there’s no trunk in this car? (4;3)
c. Why he thinks a back brace protects his leg (4;4)
d. Why they said they might be going to a movie? Why did they say that?
(4;11)

(23) a. Why do you think Santa’s not coming this year? (3;10)
b. Why do you think that Boomer came in with us? (4;2)
c. Why do you think that Mommy would not wanna watch the show? (4;6)

To summarize, the properties of English-speaking children’s why-questions have been found to

match, in considerable detail, the properties of why-questions in adult Italian. The cross-
linguistic parallel is not limited to the possibility of non-inversion in matrix clauses, but extends
to (i) compatibility of focus phrases immediately following why, (ii) use of topic phrases, (iii) the
possibility of a preceding subordinate clause, and (iv) obligatory inversion in two clause
questions. In sum, an entire cluster of properties emerges along with the non-inverted matrix why
questions. These observations lend strong support to the continuity hypothesis, which maintains
that child language can differ from the language spoken by adults in the same linguistic
community, but only in ways that adult languages can differ from each other (cf. Crain, 1991
Crain & Pietroski 2002).

Experience-based accounts of language learning would be hard-pressed to explain these

data. First, this group of properties is not manifested in the environmental input to children
learning English. Children learning English are not exposed to large quantities of non-inverted
why-questions, let alone ones with intruding focus or topic phrases, or preceding subordinate
clauses and so on. It would also be surprising, on experience-based accounts, to find all of these
properties of why-questions to emerge in tandem. On a data-driven account, these properties
would be expected to emerge one by one, as the child encountered evidence for each individual
‘construction’ (e.g., Tomasello 2003). The empirical evidence from A.L. runs counter to this
expectation. This error, too, can be construed as a mis-set parameter.

3. The Syntactic Subset Principle

Whether the issue is form or meaning, if language learners admit into their language only
expressions (and interpretations) they have actually encountered in the linguistic environment, then
they will undergenerate; their grammars will generate only a subset of the expressions of the adult
language. Such learners would not achieve the stable state that allows for production and
comprehension of sentences never before encountered. As Pinker (1990; p. 6) remarks, “… children
cannot simply stick with the exact sentences they hear, because they must generalize to the infinite
language of their community.” By the time children achieve the stable state, they are able to project
beyond their data so as to produce and comprehend (an unbounded number of) novel sentences.
Undergeneration does not always favor the data-driven account of language development. In fact,
undergeneration can constitute counter-examples to the model. It would be problematic for the
experience-based learning accounts, for example, if young children systematically disregard the
primary linguistic data and, instead, restrict their use of verbal morphology in specific ways. This
pattern of children’s behavior is expected, however, if their grammars are constrained by the

8
Syntactic Subset Principle (Berwick 1985).

According to the Syntactic Subset Principle, children are expected to initially adopt the
grammatical hypothesis that corresponds to the most restricted setting of the parameters of the theory
of Universal Grammar, regardless of the properties of the language spoken by adults in the linguistic
community. One example of undergeneration comes from a study of children’s use of past tense
morphology by Broman Olsen and Weinberg (1999). These authors adopt Broman Olsen's (1997)
definitions of aspectual classes, which derive different classes using features that are semantically
privative (+ versus ‘unmarked’), rather than equipollent (+ versus -). For example, verbs that specify
certain kinds of events are designated as [+telic], whereas 'atelic' verbs are simply unmarked for
telicity. The aspectual feature [+telic] is specified on verbs that refer to completed events with clear
results, such as made, closed, broke, came, fell, found, gave, got, lost, spilled, fixed. Atelic verbs lack
this aspectual feature; these verbs include saw, seed, had, bit, bumped, licked, choked, did, stuck. As
Broman Olsen and Weinberg point out, the perfective is restricted to [+telic] verbs in Korean; cross-
linguistically this is “the most restrictive mapping between grammatical morpheme and verbal
class.” (p. 533). Therefore, children learning less restrictive languages, such as English, are expected
to mirror adult-Korean in this aspect of verbal morphology: This prediction is based on a particular
variant of the continuity assumption, as the following quote makes clear:

“Our model's assumptions (innateness and the restriction of relevant input to positive
examples only) require that the child's initial hypothesis be the most restrictive. "Most
restrictive" is defined with respect to the range of possible adult languages; environmental
input therefore serve to relax initial hypotheses. For example, the mapping between
imperfective affixes and verbal forms is restricted to the lexical aspectual class of events in
Chinese. Since Chinese is a possible final state in the child's linguistic development, the
Syntactic Subset Principle requires that the child assume initially that the restrictions that
apply in Chinese also apply to English (Berwick 1985).” (p. 533)

The critical contrast concerns telic verbs versus atelic verbs. In a search of the CHILDES
database (see MacWhinney and Snow 1985), Broman Olsen and Weinberg found that young
children's productive use of regular past tense (-ed) and irregular past tense endings were
significantly correlated with telic verbs; only 17% of past tense endings occurred with atelic verbs.
By contrast, past tense endings appeared significantly more often with atelic verbs (56%) than with
telic verbs in the speech of caretakers to these children. As a descriptive generalization, young
children are undergenerating in the use of past tense morphology. Children are obviously not tying
their language to the input. Instead, it appears that children initially establish a one-to-one mapping
between form and meaning, disregarding much of what they encounter in the input. The mapping is
not one-to-one for adults speaking to adults; nor is it one-to-one in caretaker speech to children.
Children who show this pattern of behavior are largely unresponsive to the input. Consequently,
patterns that are robustly attested in adult data, such as the relationship between -ed and [+durative]
verbs, are not replicated in early child speech. To explain this, Broman Olsen and Weinberg turn to
the continuity hypothesis.

Another example of undergeneration is problematic for the experience-based model. The

finding comes from a study of the acquisition of Hebrew by Armon-Lotem and Crain (1998). In
adult Hebrew, the genitive construction is marked by the preposition shel ‘of’. This construction

9
occurs with both mass and count nouns in adult Hebrew. The occurrence of shel is limited to
inalienable relations, however, such as the part/whole relations in the (a) examples of (24). The
genitive construction cannot be used to refer to alienable relations, however, as illustrated by the
ungrammaticality of the (b) examples.

(24) Inalienable Alienable

a) ha-yad shel ha-ish. b) *ha-ec shel ha-ish.
'the-hand of the-man' 'the-tree of the-man'
“the man's hand.” “the man's hand.”
a) ha-galgal shel ha-traktor. b) *ha-even shel ha-traktor.
'the-wheel of the-tractor' 'the-rock of the-tractor'
“the tractor's wheel.” “the tractor's rock.”
a) ha-gargerim shel ha-orez. b) *ha-egozim shel ha-orez.
'the-grains of the-rice' 'the-nuts of the-rice'
“the grains of rice.” “the nuts of the rice.”

One kind of genitive construction in English is formed by the possessive marker, 's, which relates
two noun phrases. As in Hebrew, the English genitive construction cannot be used to express certain
alienable relations (e.g., *the truck's rocks), but it can be used to express inalienable possession (e.g.,
the truck's wheels). English is more restrictive than Hebrew in its use of the genitive, however. The
genitive in English cannot be used for mass nouns (e.g., *the rice's grains).

Based on the Syntactic Subset Principle, Hebrew speaking children are expected to initially
restrict the use of shel in a way that mirrors English, but not adult Hebrew. This is exactly what
happens. Hebrew-speaking children consistently avoid shel with mass nouns like rice in (24), despite
the widespread use of shel by adult speakers in referring to substances. Young Hebrew-speaking
children distinguish between inalienable and alienable relations for count nouns only. Hebrew-
speaking children’s more restrictive mapping of sound and meaning provides further evidence that
young children initially draw distinctions that are not motivated by the primary linguistic data, but
stem from Universal Grammar.

4. Disjunction in Child Japanese

Another source of evidence for the continuity hypothesis comes from a recent study of children's
comprehension of disjunction in Japanese. The study revealed that the interpretation given by
Japanese children to disjunction in negative sentences is similar to the interpretation of such
sentences in English, but quite unlike the interpretation that adult speakers of Japanese give these
sentences. Again, the point is that children apparently ignore the input, in accord with the
Syntactic Subset Principle. This case is different from all of the preceding examples, however,
because this is an instance of a comprehension.

According to one of De Morgan's laws of propositional logic, disjunctive statements that

occur within the scope of negation logically entail two negative conjunctive statements, as in
(25):

(25) ¬ (P!Q) " ¬P#¬Q

10
In English, there are sentences with truth conditions that closely resemble the inference in (24).
Consider (26), for example, in which the disjunction operator, or, appears under negation. The
truth conditions can be recast using the conjunction, as in (26). In what follows, we will call this
interpretation of a disjunctive statement its 'conjunctive interpretation'. The conjunctive
interpretation contrasts with the truth conditions that are associated with exclusive-or.

(26) John didn’t eat ice cream or cake.

John didn’t eat ice cream and didn’t eat cake.

In contrast to English, the Japanese disjunction operator ka appears to be less faithful to the laws
in logic. Sentence (27) is a word-by-word translation of the English sentence in (26), but it does
not exhibit the same pattern of inference, i.e., ka does not have the conjunctive interpretation:

(27) John-wa aisu ka keki-wo tabe-nakat-ta.

John-TOP ice cream or cake-ACC eat-neg-past
John didn’t eat ice cream or didn’t eat cake

In (27), the negated disjunction ka can only have the 'disjunctive' interpretation, associated with
exclusive-or. To convey the meaning in (26), Japanese speakers use the form "…mo…mo",
which literally is "…also…also", as in (28).

(28) John-wa aisu mo keki mo tabe-nakat-ta

John-TOP ice cream also cake also eat-neg-past
John didn’t eat ice cream and didn’t eat cake

At this point, one might suspect that Japanese ka is not logically equivalent to English or.
However, Goro (2004) points out that Japanese ka and English or yield identical conjunctive
interpretations when they are embedded in a subordinate clause, and when negation appears in
the main clause. The following examples illustrate the similarity between ka and or when they
appear in a sentential complement (29), and in a relative clause (30).

(29) a. John didn’t say that Mary could speak English or Japanese.
John didn’t say that Mary could speak English and he didn’t say that Mary could
speak Japanese

b. John-wa Mary-ga eigo ka nihongo-wo hanaseru-to iwanakatta

John-TOP Mary-NOM English or Japanese-ACC speak-can-COMP say-neg-past
John didn’t say that Mary could speak English and didn’t say that Mary could speak
Japanese

(30) a. John didn’t see a student who can speak English or Japanese
John didn’t see a student who can speak English and didn’t see a student who can
speak Japanese

11
 b. John-wa eigo ka nihongo-wo hanaseru gakusei-wo minakatta
Taro-TOP English or Japanese-ACC speak-can student-ACC see-neg-past
John didn’t see a student who can speak English and he didn’t see a student who can
speak Japanese

This parallelism suggests that Japanese ka does have the same semantics as English or, but that
this meaning fails to surface for some reason in simple negative sentences, such as (27). Passing
over details (see, Szabolcsi 2002; Goro 2004), we conclude that the disjunction operators in both
English and Japanese disjunction receive the conjunctive interpretation, as illustrated in (29) and
(30), but this fact is obscured in simple Japanese sentences due to some linguistic property
specific to Japanese.

These considerations invite us to ask how Japanese children interpret disjunction in

simple sentences with negation disjunctions. If children’s behavior is tied to experience, their
interpretation of sentences like (27) should conform to that of adults, in apparent violation of De
Morgan’s laws. At some point, however, they will learn to assign the conjunctive interpretation
of ka in more complex sentences, such as (29) and (30). On this scenario, Japanese children
acquire the conjunctive interpretation of disjunction in later stages of acquisition, since this
interpretation requires more complex data.

In contrast, the UG-based approach predicts just the opposite pattern of behavior by
children. This approach supposes that the inference schema in (25) and the semantic properties
of disjunction are innate (e.g., Crain & Pietroski 2002). If so, the conjunctive interpretation of
disjunction under negation “comes for free", and does not have to be learned from experience,
irrespective of the child’s target language. What Japanese children have to learn is the language-
specific property of Japanese that prevents the schema in (25) from applying in simple negative
sentences. Until Japanese children learn this, their interpretations of negated disjunctions in
simple (and complex) sentences should essentially be the same as the interpretations assigned by
English-speaking children and adults. In other words, the UG-based approach predicts that there
is a stage in language acquisition in which Japanese children interpret negated disjunctions
conjunctively, in apparent disregard to the vast majority of input they encounter. This makes the
acquisition of Japanese disjunction in negative sentences an interesting research question. As we
have seen, experienced-based approaches and the UG-based approach make opposite predictions
about the course of acquisition.

An experiment by Goro and Akiba (2004) was designed to examine Japanese children’s
interpretations of negated disjunctions. The experiment used a Truth Value Judgment task (Crain
& Thornton 1998). One experimenter acted out a short story about an "eating-game". In the
game, there were 12 animals who are each asked to eat vegetables that they don’t like: a carrot
and a green pepper. Four of those animals eat both the carrot and the green pepper; 4 of them eat
only one; 4 of them eat none. Each animal gets a different prize depending on how well they did:
those who ate both vegetables get a gold medal; those who ate only one of the vegetables get a
blue medal; and those who ate none get a black cross (for more detailed description of the
protocol and the logic behind this design, see Goro & Akiba (2004)). After the story, Kermit the
Frog puppet tries to guess how well each of the animals did in the game, using the color of the
prizes they were awarded as a clue. The crucial test cases are the puppet’s guess about those

12
animals with a blue medal, that is, those who ate only one of the vegetables. For example, the
puppet utters the following test sentence for one of those animals, the pig, who ate the carrot but
not the green pepper:

(31) Butasan-wa ninjin ka piiman-wo tabe-nakat-ta

Pig-TOP carrot or pepper-ACC eat-neg-past
Literally: 'The pig didn’t eat the carrot or the pepper'

Under the adult Japanese disjunctive interpretation of ka, the sentence means "The pig didn’t eat
the carrot or didn’t eat the pepper" and therefore matches the situation. However under the
conjunctive interpretation of ka, the sentence means "The pig didn’t eat the carrot and didn’t eat
the pepper" and is false, since he did eat one of the vegetables and got a blue medal, not a black
cross. If children have an adult-like interpretation of negated disjunction in Japanese, they should
accept the sentence. However, if children assign the conjunctive interpretation to ka, then they
should reject the puppet's statement.

The results are as follows. First, the adult control group (N=10, Age 29-32, Japanese
monolingual non-linguists) accepted the crucial test sentences 100% of the time (20/20). This
result clearly shows that the experiment is properly designed so that the crucial test sentences are
in fact judged to be true by those who have the disjunctive interpretation of negated ka. In
contrast, children (N=30, Age 3;7-6;3, Mean: 5;3, Japanese monolingual) accepted the crucial
test sentences only 25% of the time (15/60). When they were asked to explain the reason for their
negative judgment, most children said either "Because the pig did eat the carrot" or "Because it is
only the pepper that the pig didn’t eat". These results, combined with children’s explanation for
their negative judgment, strongly suggest that Japanese children are assigning the conjunctive
interpretation to ka in simple negative sentences. Consequently, it is concluded that Japanese
children's interpretation of negated disjunctions is different from the interpretation of Japanese
adults, but is more similar to the interpretation of English-speaking children/adults (see Gualmini
& Crain 2002 for the experimental study on the English-speaking children's interpretation of
negated disjunctions). A experience-based account would not anticipate such an outcome.

An account based on learning could object that children’s interpretation of ka in simple

negative sentences was an instance of ‘overgeneralizing.’ Recall that Japanese ka and English or
both yield conjunctive interpretations when they appear in a subordinate clause, with negation in
the main clause. It is conceivable that children extended the interpretation of ka from embedded
sentences to simple negative sentences. While this explanation of children’s errors is logically
coherent, the study by Goro and Akiba (2004) included several control conditions which render
this kind of learning scenario unlikely. In one control condition, children were asked to judge
sentences containing …mo…mo. Recall that expressions of the form A mo B mo in simple
negative sentences yields the conjunctive ("not A and not B") interpretation in adult Japanese:

(32) Butasan-wa ninjin mo piiman mo tabe-naka-ta

Pig-TOP carrot also pepper also eat-neg-past
"The pig didn’t eat the carrot and didn’t eat the pepper"

13
Children performed almost without error in this control condition. For those animals who got a
blue medal (i.e., those who ate only one of the vegetables), children correctly rejected sentences
containing …mo…mo 95% of the time (57/60); for those animals who got a black cross (i.e.,
those who ate none of the vegetables), they correctly accepted the sentences 100% of the time
(60/60). In short, Japanese children showed adult-like performance in interpreting …mo…mo in
simple negative sentences. It is important to point out that …mo…mo has a different
interpretation when it appears in embedded contexts, just as ka does. When …mo…mo appears in
a subordinate clause and when negation appears in the main clause, it typically yields a "not
both" interpretation, as in (33).

(33) John-wa Mary-ga eigo mo nihongo mo hanaseru-to

John-TOP Mary-NOM English also Japanese also speak-can-COMP
Iwanakatta say-neg-past
"John didn’t say that Mary could speak both English and Japanese"

This cannot mean "John didn’t say that Mary could speak English and didn’t say that
Mary could speak Japanese"

(34) John-wa eigo mo nihongo mo hanaseru gakusei-wo minakatta

Taro-TOP English also Japanese also speak-can student-ACC see-neg-past
"John didn’t see a student who can speak both English and Japanese"

This cannot mean "John didn’t see a student who can speak English and didn’t see a
student who can speak Japanese"

Thus, children's adult-like performance with the ...mo…mo construction in simple sentences
makes it unlikely that they overgeneralize from embedded contexts to simple clauses in negative
sentences with ka.

The results of this study provide further support for a UG-based approach to language
acquisition. The non-adult linguistic behavior that Japanese children show with respect to the
interpretation of negated disjunctions conforms to the pattern that is observed in another natural
language, namely, English. Given this, the UG-based approach concludes that children’s
nonadult interpretation is presumably another instance of the realization of innate linguistic
knowledge. In contrast, the data poses a serious challenge for data-driven learning approaches.
Japanese children’s behavior does not pattern with adults’ behavior, and there is no obvious
source of confusion within the positive input that is available to children.

Conclusion and Future Directions

The pattern of children’s adult and non-adult linguistic behaviors is the key to
distinguishing between the competing accounts of child language development. The UG-based
approach embraces the assumption that child languages can differ from the local adult language
only in ways that adult languages can differ from each other (Crain 1991; Crain & Pietroski
2002; Goodluck 1991; Pinker 1984). Children who adhere to this assumption are expected to “try
out” constructions that are unattested in the local language. At any given time, children will be

14
speaking a possible human language, just not the language spoken around them. Eventually, of
course, children converge on a grammar that is sufficiently like that of other speakers of the local
language; at that point, language change is no longer initiated by the input. The nature of the
difference between child and adult language is a trademark of the Universal Grammar approach
to child language.

Many questions remain unanswered. For one thing, we would like to know if the entire
pattern of why-questions observed for the child A.L. is manifested by other English-speaking
children. It will be important to investigate other clusters of properties that are tied together
within linguistic theory, as well as other cases of children projecting beyond their experience. In
any event, the findings of the studies described in the present paper illustrate the kind of
evidence that can be used to adjudicate between a UG-based approach to child language and an
experience-based approach.

References

Armon-Loten, S. and S. Crain. (1998) Interface Conditions on Child Language: A Crosslinguistic

Look at Genitives. Proceedings of the 22nd Annual Penn Linguistics Colloquium, 13-23.
Philadelphia, PA: University of Pennsylvania.
Baker, Mark (2001) The Atoms of Language. New York: Basic Books.
Berk, Stephanie (2003) Why “why” is different. In B. Beachley, A. Brown and F. Conlin (eds.)
BUCLD 27, 127-137. Somerville, MA: Cascadilla Press.
Berwick, R. (1985) The acquisition of syntactic knowledge. Cambridge, MA: The MIT Press.
Broman Olsen, M. and A. Weinberg. (1999) Innateness and the Acquisition of Grammatical Aspect
via Lexical Aspect. Proceedings of the Boston University Conference on Language
Development, 23, 529-540. Sommerville, MA: Cascadilla Press.
Chomsky, Noam (1981) Lectures on Government and Binding. Dordrecht: Foris.
Crain, Stephen (1991) Language acquisition in the absence of experience. Behavioral and Brain
Sciences 14, 597-650.
Crain, Stephen and Paul Pietroski (2002) Why language acquisition is a snap. The Linguistic
Review 19, 163-183.
Crain, Stephen and Rosalind Thornton (1998) Investigations in Universal Grammar: A Guide to
Experiments on the Acquisition of Syntax and Semantics. Cambridge, MA: MIT Press.
De Villiers, Jill (1991) Why questions? In T. Maxfield and B. Plunkett (eds.) UMOP Special
Edition: Papers in the Acquisition of Wh, pp. 155-175. Amherst, MA: GLSA.
Goodluck, Helen (1991) Language Acquisition: A Linguistic Introduction. Cambridge, MA:
Blackwell.
Goro, Takuya (2004) Japanese disjunction and the locality of positive polarity. Paper to be
presented at the Georgetown University Round Table, Washington D.C.
Goro, Takuya and Sachie Akiba (2004) The acquisition of disjunction and positive polarity in
Japanese. Paper to be presented at the West Coast Conference on Formal Linguistics
(WCCFL), University of Davis, California.
Labov, William and Teresa Labov (1978) Learning the syntax of questions. In R. Campbell and
P. Smith (eds). Recent Advances in the Psychology of Language III: 4b, pp. 1-44. New
York: Plenum Press.

15
 McDaniel, Dana (1986) Conditions on Wh-Chains. Unpublished PhD dissertation, City
University of New York.
McDaniel, Dana, Bonnie Chiu and Thomas Maxfield (1995) Parameters for wh-movement types.
Natural Language and Linguistic Theory 13, 709-753.
MacWhinney, B. and C. Snow. (1985) The Child Language Exchange Data System. Journal of
Child Language 12, 271-296.
Pinker, Steven (1984) Language Learnability and Language Development. Cambridge, MA:
Harvard University Press.
Pinker, S. (1990) Language acquisition. In D.N. Osherson and H. Lasnik (eds) An Invitation to
Cognitive Science: Volume 1: Language (199-241). Cambridge, MA: The MIT Press.
Rizzi, Luigi (1990) Relativized Minimality. Cambridge, MA: MIT Press.
Rizzi, Luigi (1997) The fine structure of the left periphery. In L. Haegeman. (ed.) Elements of
Grammar. Dordrecht, The Netherlands: Kluwer Academic Press.
Rizzi, Luigi (1999) On the position “Int(errogative) in the left periphery of the clause. Ms.
University of Siena.
Rowland, Caroline and Julian Pine (2000) Subject-auxiliary inversion errors and wh-question
acquisition: ‘what do children know?’ Journal of Child Language 27, 157-181.
Stromswold, Karin (1990) Learnability and the Acquisition of Auxiliaries. Doctoral dissertation,
MIT.
Szabolcsi, Anna (2002) Hungarian disjunctions and positive polarity. In Kenesei and Siptár
(eds.) Approaches to Hungarian 8. Budapest: Akadémiai Kiadó.
Thornton, Rosalind (1990) Adventures in Long-Distance Moving: The Acquisition of Complex
Wh-Questions. Doctoral dissertation, University of Connecticut.
Thornton, R. (1995) Referentiality and wh-movement in child English: Juvenile D-Linkuency.
Language Acquisition 4, 139-175. (Special Issue on the Acquisition of Wh-Questions.)
Thornton, Rosalind (2004) Why continuity. BUCLD 28. Somerville, MA: Cascadilla Press.
Tomasello, Michael (2000) Do children have adult syntactic competence? Cognition 74, 209-
253.
Tomasello, Michael (2003) Constructing a Language: A Usage-Based Theory of Language
Acquisition. Cambridge, MA: Harvard University Press.
Wexler, Kenneth (1998) Very early parameter setting and the unique checking constraint: A new
explanation of the optional infinitive stage. Lingua 106, 23-79.

16

View publication stats