Explaining Children*s Wh-In-Situ Questions: Against Economy

advertisement
An Information-Structural
Account of Children’s Wh-In
Situ Questions in French
Megan Gotowski, UCLA
mgotowski@ucla.edu
Misha Becker, UNC Chapel Hill
mbecker@email.unc.edu
2
Introduction
 Languages differ in whether they license fronted whquestions or wh-in situ
 English: fronted-wh questions required
(1) a. What did John buy?
b. *John bought what?
 Mandarin: wh-in situ required
(2) a. Lisi mai-le sheme (ne)
Lisi buy-ASP what
(Q)
“Lisi bought what?”
b. *Sheme Lisi mai-le (ne)
what Lisi buy-ASP (Q)
 * is acceptable only if echo-question
3
Introduction
 French: both fronted wh- and wh-in situ questions allowed as
information-seeking Qs
(3) a. Qu’est-ce que tu
Q
as
vu?
you.2SG have.PRES.2SG seen
“What did you see?”
b. Tu
as
vu quoi?
SCL.2SG have.PRES.2SG seen what?
“You saw what?”
 Previous research has shown that French children produce more
in-situ forms than adults; explain the asymmetry by Economy
(Hamann 2006, Zuckerman & Hulk 2001)
 We present new data supporting this observation but propose an
account based on the discourse properties of wh-phrases
4
Outline
1. Present data on production of wh-in situ in child French
and French child-directed speech (CDS)

Focus specifically on production of qu’est-ce que vs. quoi,
the most frequent wh-Q type in the input
2. Shortcomings of the Economy account
3. Propose a new analysis based on discourse properties of
wh-in situ, following Mathieu’s (2004) analysis of them as
“lower-order topics”
4. Finally, show that children overproduce wh-in situ only in
languages in which in situ can be information-seeking Q
(i.e. not in English)
5
Previous Research: Hamann (2006)
 Geneva corpus: Augustin, Louis, Marie (1;8-2;9)
 Compared production of fronted wh-questions to wh-in
situ questions in child French (no adult data)
 Children overwhelmingly produced in situ forms (all whwords/phrases)
Fronted Wh
Wh-In Situ
Augustin
9.3%
90.7%
Louis
26.7%
73.3%
Marie
23.9%
76.1%
 Al (1976): colloquial (adult) French: 33% wh-in situ
6
Previous Research: Z&H (2001)
 Elicitation study: children (4;0-5;9) and adults
 Compared production of fronted wh-questions to wh-in situ
AND compared production by wh-word
 Children produced more in situ forms
 Significant contrast even when comparing total production
(p<0.05)
 Children produce more in situ forms with quoi than adults
Wh-In Situ (Overall)
Wh-In Situ (Quoi)
Children
3%
Children
18%
Adults
1%
Adults
3%
7
Our Data
 Palasis corpus (Palasis 2010): last 50 files; 20 children
(2;9-3;10) and 1 adult
Corpus/Study
Age
Geneva
1;8-2;9
Palasis
2;9-3;10
Z&H (2001)
4;0-5;9
8
Our Data
 Focus on qu’est-ce que / quoi = the most frequent wh-Q type in
the input
Wh-Word
French CDS (Palasis 2010)
qu’est-ce que ‘what’
626
qui ‘who’
19
ou ‘where’
15
quand ‘when’
0
comment ‘how’
83
combien ‘how many’
3
 Qu’est-ce que lit. ‘what-is-it-that’ = KESK (Hulk 1996)
 We included object-wh questions with a subject & verb
 Bare questions were excluded:
(4) Quoi? (‘What?’)
9
Results: Palasis Corpus
 Children produce significantly more wh-in situ questions
than the adult
Children
Adult
Ages
Fronted Wh
(KESK)
Wh-In Situ
(quoi)
2;9-3;10
11.8% (13)
88.2% (97)
83.4% (626)
16.6% (125)
10
Results
 Comparing data across all studies, children’s rate of whin situ drops over time, but is higher than adults’ until
around age 4
Corpus/Study
Age
In Situ
(Quoi)
Geneva
1;8-2;9
80%*
Palasis
2;9-3;10
88.2%
Z&H (2001)
4;0-5;9
18%
Palasis
adult
16.6%
Al (1976)
adult
33%
*rate of all wh-in situ questions
11
Weaknesses of Economy
 Crisma (1992) studied Philippe (Leveillé corpus)
 Philippe initially produced only fronted wh-questions,
only later produced in situ (after 2;6)
 Hamann (2006) and Z&H (2001) acknowledge this, claim
input is responsible
 However, Palasis children hear fronted forms 85% of the
time, but produce them only 12% of the time
 As we show later, rates of wh-in situ in input are not
directly related to children’s rates of in situ production
12
Weakenesses of Economy
 European Portuguese allows both fronted and in situ whquestions
 Soares (2004): corpus data from 3 children (1;2-4;5)
 These children produce fronted wh-questions before in
situ questions
 cross-linguistically, children do not uniformly prefer whin situ at the earliest stages of question production
13
Overt vs. Covert Movement
 Both fronted wh-questions and in situ involve movement
of wh-operator (cf. Aoun et al. 1981, Huang 1982, Lasnik & Saito
1992)
 fronted wh-questions = overt movement
 wh-in situ = covert movement (LF)
 Children are thus not avoiding movement with in situ!
 Instead we propose an alternate explanation that relies
on discourse properties of fronted and wh-in situ forms
14
Our Account
 Mathieu (2004): in situ phrases are a kind of split-DP
 In wh-in situ, the wh-operator moves, and is “split” from the
nominal/wh-word
 An overt split-DP:
(5) a. Combieni as-tu
[split]
how-many
split]
have-you
lu ti de livres?
read
of books
b. Combien de livresi as-tu
how-many of books
have-you
lus ti?
[non-
read
“How many books have you read?”
[Mathieu 2004, ex 7]
15
Properties of In Situ
 Both split-DPs and wh-in situ are licensed in restricted
contexts (intervention effects)
 Wh-in situ not allowed under negation
(6)
a. *Op Il ne voit pas qui?
[in situ]
he ne sees neg who
b. Op Quii est-ce qu’il ne voit pas ti?
who is-it
[fronted]
that-he ne sees neg
“Who does he not see?”
[Mathieu 2004, ex. 4]
16
Properties of In Situ
 Wh-in situ not allowed in embedded clauses
(7) a. *Op Il a dit [que qui avait eternué?]
he has said that who has
[in situ]
sneezed
b. Op Quii est-ce qu’il a dit [qui ti avait eternué?] [fronted]
who is-it
that-he has said that
“Who did he say sneezed?”
has
sneezed
[Mathieu 2004, ex. 67]
 Likewise, split-DPs are not possible in these contexts:
(8) a. *Combieni n’as-tu
pas lu ti de livres?
[split-DP]
how-many ne-have-you neg read of books
split]
b. Combien de livresi n’as-tu
pas lus ti?
[non-
how-many of books ne-have-you neg read
“How many books have you not read?”
(=6a)]
c. *Op Il ne voit pas qui?
[in situ
he ne sees neg who
(“He doesn’t see who?” )
 Connection between Split-DPs and in situ is important
because...
 Split-DPs introduce a discourse referent, and they are
considered “new topics” = lower order/backgrounded
 Fronted wh-questions are Focus phrases (=foregrounded)
17
18
Our Account
 Children are able to distinguish between Topic and Focus
(De Cat 2009, 2011) and given vs. new (Hickmann 2003, i.a.)
 However, children overattribute referents to the common
ground (CG) (cf. Schaeffer & Matthewson 2005)
 Fronted wh-phrase = foregrounded, salient (not in CG)
 In-situ = backgrounded referent (in CG)
 Children may believe referents are sufficiently salient 
need not be foregrounded via fronting = more in situ
19
Our Account
 Children often think that what is salient to them is also
salient to others
 They will often “exploit joint attention” in discourse (cf.
Wexler 1998; de Cat 2009)
 French-speaking children may rely on visual saliency to
avoid linguistic mechanisms of foregrounding (e.g. overt
wh-movement)
 We predict that children will produce in situ questions
regardless of the discourse context
20
 In (9) the adult (KAT) produces a fronted wh-question when
introducing the referent; the referent is placed in a salient
position.
 After the referent has become part of the common ground, she
uses an in situ question.
(9) KAT: Qu’est ce-qu’ il fait
Q
le renard?
[adult]
he do.PRES.2SG the fox
“What is the fox doing?”
LSN: Le renard il mange.
the fox
[child]
he eat.PRES.3SG
“The fox is eating.”
KAT: Oui il mange
quoi?
yes he eat.PRES.3SG what
“Yes, he is eating what?”
[adult]
21
 Children may not be making the same distinction:
(10) KAT : Tu lui demandes si i(l) sait
you him ask
ce que c’est.
if he knows that
[adult]
that-is
“Ask him if he knows what it is.”
LSN : C’est quoi ça ?
[child]
that-is what that
“What is that?” (Lit. That is what that?)
MAS : Euh ça c’est
à moi ça.
[child]
uh that that-is to me that
“Uh that’s mine, that.”
KAT : Et qu’est-ce que c’est ?
and Q
[adult]
that-is
“And what is it?”
 Children produce wh-in situ in contexts where adults require
fronting: children permit a wider CG = more backgrounded
referents
22
Comparing French and English
 We compared rates of wh-in situ (with quoi/what) in
the input to French and English-speaking children
 Rate of echo-Qs in English CDS closely parallels rate of
wh-in situ in French CDS
Language
Corpus
Input- In Situ (Adults)
English
Brown (Eve 1-20)
16.0%
English
Brown (Adam 1-19)
22.0%
Language
Corpus
Input- In Situ
(Adult)
French
Palasis
16.6%
23
Comparing French and English
 Recall that English permits wh-in situ only in echo
questions
 Echo Qs are used when speaker is surprised, but also to
request repetition/clarification:
(11)
MOT: no more celery?
MOT: alright.
CHI: man have it.
MOT: man have what?
(Eve 01)
24
Comparing French and English
 How often do English-speaking children produce wh-in
situ?
Language
Corpus
Input- In Situ
(Adults)
Output- In
Situ (Eve)
English
Brown (Eve 120)
16.0%
(201/1257)
0.4%
(1/247)
Language
Corpus
Input- In Situ
(Adults)
Output- In
Situ (Adam)
English
Brown (Adam
1-19)
22.0%
(257/911)
0.2%
(1/744)
Language
Corpus
Input- In Situ
(Adult)
Output
(Children)
French
Palasis
16.6%
88.2%
25
Comparing French and English
 This suggests that input rates are not obviously related
to children’s own rates of in situ production
 Results also indicate that children know whether or
not in situ is a grammatical option (Takahashi 1991, Maxfield 1991)
 They overproduce in situ ONLY if in situ is a grammatical
option in the grammar
 How do children know whether wh-in situ can only be
echo Q?
26
Echo vs. non-Echo
 How do children know if wh-in situ can only be echoic in
their language?
 Information-seeking Q: what can only be a DP
(12) a. *What did you buy a?
b. *What did you see a very house?
c. *What did you?
• Echo question what can be (almost) any category
(13) a. You bought a what? (NP)
b. You saw a very what house? (AP)
c. You what? (VP)
27
Echo vs. non-Echo
 In English CDS, ~50% of wh-in situ questions involve
what replacing a non-DP (Eve: 39%; Adam: 55%)
(14)
a. Eve’s what kind? (AP; eve02)
b. The tap what? (VP; eve08)
c. Because what? (IP; eve13)
d. big what? (N; adam10)
e. All around the what? (NP; eve19)
f. cock-a-doodle fresh what? (unclear; adam09)
28
Summary
 French children produce considerably more in situ
questions than adults until at least around age 4
 This results from children’s overattribution of referents to
the common ground
 They permit more backgrounded NPs = higher rate of in
situ than adults
 Input rates of wh-in situ in CDS does not match children’s
rates
29
Future Research: Elicitation
 Elicit wh-questions controlling for what is in the CG
(foregrounded vs. backgrounded referents)
 Younger (3;0-4;5) and older children (4;6-5;11)
 Present videos of animals doing different actions
 Across conditions vary whether a referent is introduced into
the discourse (backgrounded) or not.
30
Future Research: Elicitation
 Experimenter cannot see the scene with the animals, but the
puppet can;
 The child is then prompted with either (15) or (16) matching
one of two conditions:
(15) Condition #1: Backgrounding
« Il y a des animaux. Je peux entendre un chien! Il fait quelque
chose, mais je sais pas quoi. Demande à la poupée ce qu’il fait. »
(Eng.: There are some animals. I can hear a dog! He is doing something,
but I don’t know what. Ask the puppet what he’s doing.)
31
Future Research: Elicitation
(16) Condition #2: No Backgrounding
« Il y a des animaux. Ils font quelque chose, mais je sais
pas quoi. Demande à la poupée ce qu’ils font. »
(Eng.: There are some animals. They are doing something,
but I don’t know what. Ask the puppet what they’re
doing.)
Hypotheses:
 If children assume that any referent is in the CG, they will
produce in situ questions regardless of condition.
 Older children will produce fewer in situ questions than
younger children in condition (16) =>
They will be more adult-like.
32
Future Research: Prosody
 Explore connection between prosody and wh-in situ in
French-speaking children (cf. Déprez et al. 2013a,b,
Richards 2006)
 Specific pitch contour associated with in situ Qs
 Record children’s elicited fronted and wh-in situ Qs and
submit to phonetic analysis
 Compare to adult’s pattern of pitch compression...
 Do children make the same prosodic distinctions?
 Do they have the same pitch contour for fronted and in situ?
33
Thank you!
Special thanks to Daniel Blackey for help coding the English data.
34
Selected References
Crisma, P. 1992. On the acquisition of wh-questions in French. GenGenP 0: 115-122.
De Cat, C. 2009. Experimental evidence for preschoolers’ mastery of Topic.
Language Acquisition 16: 224-239.
De Cat, C. 2011. Information tracking and encoding in early L1: Linguistic competence
vs. cognitive limitations. Journal of Child Language 38: 828-860.
Hamann, C. 2006. Speculations about early syntax: The production of wh-questions by
normally developing French children and French children with SLI. Catalan
J. of Linguistics 5:143-189.
Mathieu, E. 2004. The mapping of form and interpretation: The case of optional whmovement in French. Lingua 114: 1090-1132.
Soares, C. 2004. Computational complexity and the acquisition of the CP field in
European Portuguese. In S. Blaho, L. Vincente, and M. de Vos (Eds.)
Proceedings of Console XII.
Wexler, K. 1998. Very early parameter setting and the unique checking constraint: A
new explanation of the optional infinitive stage. Lingua 106: 23-79.
Zuckerman, S. & A. Hulk. 2001. Acquiring optionality in French wh-questions: An
experimental study. Revue Queb. de Ling. 30:71-97.
AppendixData from Older Children
35
 Montréal corpus (Feider & Saint-Pierre 1987): first 11
files (5;6-6;8) (no adult; we compared these children to
adult from Palasis corpus)
Ages
Fronted Wh
(Qu’est-ce que)
Wh-In Situ
(Quoi)
ChildrenPalasis Corpus
2;9-3;10
11.8%
88.2%
ChildrenMontreal Corpus
5;6-6;8
77.8%
22.2%
83.4%
16.6%
Adult (Palasis)
36
English Echo Questions
 Takahashi (1991):





3 different animals eating 3 different foods
horse eats carrot, giraffe eats leaves, elephant eats banana
statement: “The animals are eating food.”
“What are the animals eating?” -> carrot, leaves, banana
“The animals are eating what?” -> food (superordinate)
 Fronted question -> 38% superordinate
95% superordinate
Download