Content uploaded by Anna Gavarró
Author content
All content in this area was uploaded by Anna Gavarró on May 31, 2016
Content may be subject to copyright.
THE ACQUISITION OF CATALAN RELATIVES:
STRUCTURE AND PROCESSING
ANNA GAVARRÓ, ARNAU CUNILL, MÍRIAM MUNTANÉ,
MARC REGUANT1
Abstract. Research in various languages indicates that children interpret subject
relatives in an adult-like manner substantially earlier than they interpret object relatives.
This asymmetry may be attributed to processing of the corresponding syntactic
structures, as in Gibson (1998), Morrill (2000). We address this basic asymmetry by
considering the acquisition of Catalan relatives, both for comprehension and
production. The interpretation of Catalan relatives was investigated by Gavarró et al.
(forthcoming) and the results show the well-known asymmetry in comprehension
between subject and object relatives. The elicitation experiment, designed by
Friedmann et al. (in preparation), was run with twenty Catalan-speaking 5-year-olds.
Children produced 98% of subjects as adults do, while they produced fewer object
relatives. Briefly, for our results the parallelism between comprehension and production
holds. We propose to account for the findings by adopting an analysis based on
Morrill’s (2000) metric of syntactic complexity, an implementation of Gibson’s (1998)
insight that processing difficulties increase as a function of the number of unresolved
dependencies that the speaker must keep in memory. Gibson’s and Morrill’s proposals
are neutral with respect to whether linguistic knowledge is put to use in production or
comprehension: here we claim that, in fact, for the empirical domain considered,
production and comprehension are equally taxed.
Keywords: Catalan, relative clauses, acquisition, production, comprehension,
complexity metric, categorial grammar.
1. INTRODUCTION
Research in various languages indicates that children interpret subject
relatives in an adult-like manner substantially earlier than they interpret object
1 Universitat Autònoma de Barcelona. Address for correspondence: Departament de Filologia
Catalana, Universitat Autònoma de Barcelona, Edifici B; 08193 Bellaterra (Cerdanyola del Vallès);
anna.gavarro@uab.cat
*An earlier version of this paper was presented in Galana 2010 in Toronto, and appeared in its
selected proceedings (Gavarró et al. 2011). Our thanks to the audience of Galana, to Glyn Morrill for
his help with the analyses of Catalan, Chinese and Basque, and to an anonymous reviewer of RRL for
comments to an earlier version of the paper. We are grateful to the children and teachers of the
schools Lloriana in Sant Vicenç de Torelló and Maria Borés in La Pobla de Claramunt for their
collaboration. The first author acknowledges the financial support of project HUM2006–13295–
C02–01.
RRL, LVII, 2, p. 183–201, Bucureşti, 2012
Anna Gavarró, Arnau Cunill, Míriam Muntané and Marc Reguant 2
184
relatives. This asymmetry, while grounded in a grammatical contrast, may be
attributed to processing of the corresponding syntactic structures, as in Gibson
(1998), Johnson (1998), and Morrill (2000). One question that emerges is: if
processing can be argued to be the source of poor performance in the interpretation
of object relatives, does this carry over to production? Here we address this issue
with the acquisition of Catalan; we present original results for relative clause
elicitation and compare them with those of a relative clause interpretation experiment.
The paper proceeds as follows: first we provide the background to the present
study, and we give an analysis for the subject/object relative asymmetry found in
comprehension (Section 2). Then we describe the experiment for the elicitation of
relative clauses carried out and provide the results (Section 3); finally, we compare
production and comprehension and draw conclusions (Section 4).
2. BACKGROUND: RELATIVE CLAUSE COMPREHENSION
IN CATALAN
2.1. An experiment on relative clause comprehension
The acquisition literature reports on work in numerous languages showing a
contrast between the interpretation of subject and object relative clauses. This is
attested for English (Brown 1971, de Villiers et al. 1979 and references therein),
French (Frauenfelder et al. 1980), German (Schriefers et al. 1995), Italian (Arosio
et al. 2009), Greek (Guasti et al. 2008), and Hebrew (Friedmann et al. 2009),
among others. This asymmetry is, however, not universal, since languages with
prenominal relative clauses such as Chinese and Basque are known to behave
differently: see for example the work of Hsiao and Gibson (2003) on Chinese and
that of Carreiras et al. (2010) on Basque. Nonetheless we will centre our attention
here on the type of language mentioned first, namely, that of head-initial,
postnominal relative clauses, and consider one particular language, Catalan, for
which comprehension results are available, and then revisit prenominal relatives at
the end of the paper.
Catalan relative clauses are postnominal and headed by the relative pronoun
que that, unlike e.g. the Spanish relative pronoun, is not identical to a wh- word.
Que introduces both subject and object relatives – only prepositional relatives
present the alternative pronoun qui (La nena a qui he enviat el llibre ‘The girl to
whom I sent the book’).
Gavarró et al. (forthcoming) carried out an experiment on the comprehension
of Catalan relative clauses, replicating the experiment of Arosio et al. (2009) and
Adani (2010). The experiment was an agent identification task with picture
support. Children had to identify a character in a picture when the experimenter
requested it, as illustrated in (1).
3 The acquisition of Catalan relatives
185
(1) a. Assenyala el camell que segueix els elefants!
point to the.SG camel that follows the.PL elephants
‘Point to the camel that is following the elephants!’
b. Assenyala el camell que els elefants segueixen!
point to the.SG camel that the.PL elephants follow
‘Point to the camel that the elephants are following!’
c. Assenyala el camell que segueixen els elefants!
point to the.SG camel that follow the.PL elephants
‘Point to the camel that the elephants are following!’
(1a) includes a subject relative, (1b) an object relative with a preverbal subject in
the embedded clause, and (1c) an object relative with a postverbal subject in the
embedded clause. Object relatives of this last type are potentially ambiguous as
Catalan has no overt Case marking and therefore postverbal subjects may be
interpreted as subjects or objects, unless subject–verb agreement disambiguates the
sentence. (This ambiguity is in fact found in all the Romance null subject languages.)
All sentences in the experiment were unambiguous and reversible, so that the
interpretation rested on the linguistic input only. The materials were designed so
that the questions would be pragmatically felicitous. The task was run with 33
children, 12 of whom were younger than 4;6 (mean age: 3;11,26), 11 of ages
between 4;6 and 5;6 (mean age: 4;11,6), and 10 older than 5;6 (mean age: 6;0,12);
the age range was 3;5,9–6;2,30 and the mean age 4;11,4. The results appear in
Table 1.
Table 1
Results for relative clause comprehension: target comprehension by age
SR OR
<4;6 57/72 79% 47/132 35%
4;6–5;6 64/66 97% 53/121 43%
5;6< 60/60 100% 63/110 57%
adults 131/132 99% 243/252 96%
SR: subject relative; OR: object relative
These results indicate how subject relatives are interpreted in an adult-like
manner from early on (79% of the time in the younger group), while object
relatives develop at a slower pace; in fact object relatives with postverbal subjects
are misinterpreted for the whole period investigated. Very similar results are found
in a related language, Italian (see Arosio et al. 2009, Adani 2010).
2.2. A categorial analysis
In Morrill and Gavarró (2010) we analyse these comprehension contrasts as
the result of the relative processing load of the structures involved. We propose to
Anna Gavarró, Arnau Cunill, Míriam Muntané and Marc Reguant 4
186
account for the findings by adopting an analysis based on Morrill’s (2000, 2010)
metric of syntactic complexity, an implementation of Gibson’s (1998) and
Johnson’s (1998) insight that processing difficulties increase as a function of the
number of unresolved dependencies that the speaker must keep in memory. Morrill
(2000, 2010) proposes a metric of processing cost that can account for the relative
difficulty of Catalan object relatives as opposed to subject relatives (notice that
both are low in processing cost if compared to centre embedding, e.g. the cheese
that the rat that the cat saw ate stank, for which adults are found to have
difficulties).
Categorial grammar (Morrill 2010) classifies words and expressions by
means of fractional types built over basic types such as sentence (S) and nominal
(which we parameterise here with number singular, N(sg), plural, N(pl), or
unspecified, N(_)). An expression of type A\B is one which concatenates with any
expression of type A to the left to form an expression of type B. An expression of
type B/A is one which concatenates with any expression of type A to the right to
form an expression of type B. Formally:
(2) A\B = {s| for all s’∈A, s’+s∈B}
B/A = {s| for all s’∈A, s+s’∈B}
Morrill (2000, 2010) describes a complexity metric founded on incremental
categorial processing in terms of proof nets. In this view of processing, types are
marked with input polarity (•), meaning that a resource is given, or output polarity
(◦), meaning that a resource is wanted. Polar types are unfolded upwards into polar
type trees as follows:
(3)
A
◦ B
• B
◦ A
•
A\ B
• A\B◦
B
• A
◦ A
• B◦
B/A
• B/A◦
We refer the reader to the references above for the details, which are quite
involved, but we illustrate the basic idea here with the processing of the sentence in
(4):
(4) John loves Mary.
5 The acquisition of Catalan relatives
187
Initially, a sentence is sought and after hearing the first word its type is given:
(5)
S
◦ N(sg)
•
John
When the second word is heard, its unfolded type is connected by two
dependencies: the subject sought is given by the first word John and the sentence
projected is matched by the initial expectation of a sentence. We represent this as
follows:
(6)
N(sg)◦ S•
N(sg)\S• N(_)◦
S◦ N(sg)• (N(sg)\S)/N(_)•
John loves
When the final word is heard, the parse is completed thus (the unspecified object
number on the verb type becomes instantiated by unification with the type with
which it is matched):
(7)
N(sg)◦ S•
N(sg)\S• N(_)◦
S◦ N(sg)• (N(sg)\S)/N(_)• N(sg)•
John loves Mary
The derivations proposed for a subject relative and an object relative in
Catalan appear in (8) and (9) respectively.
Anna Gavarró, Arnau Cunill, Míriam Muntané and Marc Reguant 6
188
(8) Derivation of sentence (1a), from Morrill and Gavarró (2010)
(9) Derivation of sentence (1b), from Morrill and Gavarró (2010)
The relative pronoun que seeks to the right a category in turn seeking a DP, and
this is satisfied earlier in the subject relative clause (in subject position) than in the
object relative clause. The processing load at each point in the sentence is
mechanically worked out by the metric. The complexity profile of a sentence
describes the incremental load at each word boundary by counting the syntactic
dependencies that are unresolved at each point (where syntactic dependencies
include major categories and feature values, both counting for 1). When the last
word of the sentence has been heard, all dependencies are resolved and the profile
reaches 0. We can observe the differences in the number of dependencies to be
resolved in the subject and object relatives exemplified above. The complexity
profiles are read as follows: the Y-axis represents the load on memory at a point;
the character a represents the load level in the example after you hear one word and
before you hear the next. The load level corresponds to the number of lines
overarching across in the derivation of the sentence. The complexity profile of (1b)
in (11) is higher than that of (1a) in (10), thus predicting lower acceptability, which
by hypothesis results in higher comprehension problems.
7 The acquisition of Catalan relatives
189
(10) Complexity profile of sentence (1a)
3 a
2 a
1 a
0 a
Assenyala el camell que segueix els elefants.
point-to the camel that follow-3s the-pl elephants
(11) Complexity profile of sentence (1b)
7 a
6
5 a
4
3 a
2
1
0 a
Assenyala el camell que els elefants segueixen.
point-to the camel that the-pl elephants follow-3pl
To summarise, we have shown how a categorial metric of complexity can be put to
use to account for differences between subject and object relative comprehension.
What we set out to do next is consider the production of relative clauses under the
same analysis.
3. AN EXPERIMENT OF RELATIVE CLAUSE PRODUCTION
3.1. Experimental design
In order to test the production of relative clauses in child Catalan, we ran an
elicitation experiment, our version for Catalan of the elicitation task designed in the
context of COST Action A33, in turn based on Novogrodsky and Friedmann
(2006). Children were asked for their preferences, given a choice: ‘A boy drinks
milk, and a boy drinks water. Which child would you rather be?’ and were
requested to start their answers with ‘I’d rather be the boy…’. The prompts
crucially induced the production of relative clauses, and the method proved to be
very effective. This elicitation method is exemplified for subject relatives and
object relatives for Catalan in (12).
(12) a. Hi ha dos nens. Un nen beu llet i un altre nen beu aigua. Quin nen t’agradaria ser?
CL have two children A child drinks milk and another child drinks water which
child CL like to-be
‘There are two children. One child is drinking milk and the other child is drinking
water. Which child would you rather be?’
Anna Gavarró, Arnau Cunill, Míriam Muntané and Marc Reguant 8
190
b. Hi ha dos nens. Un pare abraça un nen i un pare gronxa un nen. Quin nen t’agradaria
ser?
CL have two children A father hugs a child and a father swings a child. Which child
CL like to be
‘There are two children. A father is hugging a child and a father is swinging a child.
Which child would you rather be?’
There were a total of twenty items, of which ten corresponded to subject
relatives and ten to object relatives. The verbs in the embedded sentence were all
transitive. Of each ten subject and ten object relatives, six were reversible (both
referents involved could fulfil the Agent and Theme theta role, as in 12b), while
four were irreversible (like 12a above). Reversibility will not be considered in the
analysis.
The children were all native speakers of Catalan (in particular Central
Catalan) recruited at the Lloriana primary school in Sant Vicenç de Torelló and at
the Maria Borés primary school in La Pobla de Claramunt; adult controls also came
from the same area. The relevant age information about child and controls subjects
appears in Table 2.
Table 2
Subjects
subjects # age range mean age
5-year-olds 20 5;0,11–5;11,24 5;5,15
Adults 10
Children were tested individually in a quiet room in their schools. The input
was not recorded, but rather produced by the experimenters, who also transcribed
the answers. Only sentences with a relative pronoun were considered to be
relatives, and the authors judged whether a relative clause was well-formed or not.
3.2. Results
For subject relatives, the response types relevant for Catalan were (i) adult-
like subject relatives with a gap (13a), headless relatives (13b), and fragments
without a relative clause (13c).
(13) a. M’agradaria ser el nen que beu llet.
CL would-like to-be the child that drinks milk
‘I would like to be the child who is drinking milk.’
9 The acquisition of Catalan relatives
191
b. M’agradaria ser el que beu llet.
CL would-like to-be the that drinks milk
‘I would like to be the one who is drinking milk.’
c. M’agradaria ser el de la llet.
CL would-like to-be the of the milk
‘I would like to be the one of the milk.’
In the case of object relatives, the strategies that Catalan speakers can adopt
are more diverse, and likewise the errors found are also diverse. Answers included:
(i) adult-like object relatives with a gap (14a), (ii) relatives with a postverbal
argument, possibly the subject (object relative) or the object (subject relative)
(14b), (iii) object relatives with a resumptive pronoun (14c), (iv) object relatives
with a reflexive resumptive (14d), (v) object relatives with a resumptive full DP
(14e), and fragments and inappropriate subject relatives.
(14) a. M’agradaria ser el nen que el pare gronxa.
CL would-like to-be the child that the father swings
‘I would like to be the child that the father is swinging.’
b. M’agradaria ser el nen que gronxa el pare.
CL would-like to-be the boy that swings the father
‘I would like to be the boy who is swinging the father’
or ‘I would like to be the child who the father is swinging.’
c. M’agradaria ser el nen que el desperta la música.
CL would-like to-be the boy that him wakes-up the music
‘I would like to be the child that the music is waking up.’
d. M’agradaria ser el nen que em desperten.
CL would-like to-be the child that REFL wake up-3pl
‘I would like to be the child they are waking up.’
e. M’agradaria ser el nen que el veí pentina el nen.
CL would-like to-be the boy that the neighbour combs the boy
‘I would like to be the child that the neighbour is combing.’
While all the options encountered amongst the productions of subject
relatives were well-formed, not all the productions of object relatives were so;
resumptive pronouns are grammatical in colloquial Catalan, but first person
resumptives (14d) and resumptive full DPs (14e) are not well-formed.
The total number of answers by the children was 400, of which only 13 were
not relative clauses. The results for 5-year-olds and adults for subject and object
relatives appear in Tables 3 and 4:
Anna Gavarró, Arnau Cunill, Míriam Muntané and Marc Reguant 10
192
Table 3
Results for subject relative production
5-year-olds adults
Target relative 173/200 86.5% 97.5%
headless 23/200 11.5% 2.5%
fragment 3/200 1.5% 0%
other 1/200 0.5% 0%
Table 4
Results for object relative production
5-year-olds adults
target unambig 15/200 7.5% 16%
ambiguous 76/200 38% 19%
resumptive pron 34/200 17% 53%
reflexive resumpt 6/200 3% 2%
null subject 9/200 4.5% 0%
subject relative 26/200 13% 0%
DP filled gap 25/200 12.5% 0%
passive 0% 10%
fragment & other 9/200 4.5% 0%
The first table shows that children produce subject relatives straightforwardly
and, although they produce more headless relatives than adults, the pattern of
production is very adult-like. While at age 5 children produced 98% of subject
relatives, just like adults (including headed and headless relatives), they produced
fewer object relatives, as shown in the second table. Only 7.5% of relatives were
object relatives with a gap, and 38% corresponded to ambiguous relatives
(compared to the 19% ambiguous responses produced by adults). The difference
between subject and object relatives is striking.
Object relatives with resumptive pronouns deserve special mention; they
occurred in 17% of cases in the children’s production, but constituted 53% of adult
11 The acquisition of Catalan relatives
193
production (although these resumptives are considered substandard in Catalan, they
are widely used and identified in the literature; see Solà 2002). Children, unlike
adults, produced relatives with a full DP copy of the relativised element. Child
production of resumptives has also been noted for languages which disallow it in
the adult grammar, such as English and French (data from Pérez-Leroux 1995 and
Labelle 1990 respectively):
(15) the one that he lifted it (Lia; 4;5)
(16) sur la balle qu’il l’attrappe.
over the ball that he it catches (LE, 3;8)
Pérez-Leroux (1995) ran a relative clause elicitation experiment with
Spanish- and English- speaking children (26 Spanish-speaking children, aged 3;5
to 6;8, mean age 5;3; eleven English-speaking children aged 3;5 to 5;5; mean age
4;10); based on her results, she argued that resorting to resumptive pronouns
should be considered a strategy analogous to DP filled gaps in relatives, as in
(17)–(18), and was found across languages in a systematic way.
(17) the one that the cowboy is pulling the horse (Thomas, 3;7)
(18) sur la balle qu’il lance la balle (M, 5;0, from Labelle 1990)
on the ball that he catches the ball
In Catalan we found resumptives with object relatives only, consistent with
Pérez-Leroux’s findings for English and Spanish, and produced 17% of the time;
DP filled gaps were found in 12.5% of cases; taken together, these two sets thus
represent 29.5% of answers. Pérez-Leroux found that the percentage of
resumptives in a broad sense (including resumptive pronouns and DP filled gaps)
was 36.2% in Spanish, 25% in English and 40.9% in French (French data from
Labelle’s 1990 study); she also found the difference in the production of
resumptives in the different languages not to be significant. Our results are clearly
in line with those of Pérez-Leroux, and this is particularly relevant given that
Catalan adults produce more resumptives than English adults (other than with
relatives involving such that, English speakers appear not to commonly resort to
resumptive pronouns). So we can conclude that our results are consistent with the
claim by Pérez-Leroux that, in acquisition, the proportion of resumptives does not
differ substantially across languages.2
2 To our knowledge there is no published work in the type logical categorial framework here
on the grammar of resumptives or their processing.
Anna Gavarró, Arnau Cunill, Míriam Muntané and Marc Reguant 12
194
As for the general outcome of the experiment reported, there is a sharp
contrast between subject and object relatives in their production by 5-year-olds. It
remains to be seen at what age the behaviour of children attains adult levels.
4. COMPREHENSION AND PRODUCTION COMPARED
In some domains of language acquisition, the observation has been made that
discrepancies emerge between production and comprehension of grammatical
constructions. In some of these cases the discrepancy may be just apparent, so that
children appear to comprehend a construction, and fail to produce it, yet under
closer scrutiny it is found that their comprehension is also delayed. In other cases,
such as the one of agreement production and comprehension, to which we turn
later, the discrepancy still needs to be accounted for. Here we deal with an instance
of this general issue. In particular, the research question which we address here is:
Does the production of relative clauses parallel comprehension in acquisition? As a
first step, let us compare the results available involving comprehension to those for
production obtained in the present study. It should be noted that the age range of
the two groups of children compared does not coincide in its span, since in the
comprehension experiment the age groups were children younger than 4;6, children
4;6 to 5;6, and children older than 5;6. Still, given that the pace of development in
relative clause comprehension is gradual, it seems legitimate to compare a large
group of 5-year-olds with two groups of children spanning 6 months in the older
and younger range.
Table 5
Subject and object relative clause comprehension and production
comprehension
SR OR
4;6–5;6 64/66 97% 53/121 43%
5;6< 60/60 100% 63/110 57%
total 124/126 98% 116/231 50.2%
production
5-y-o 98% 62.5%
Target object relatives in production here include: object relatives with a gap,
relatives with a resumptive pronoun and ambiguous relatives; inclusion of this last
type may overestimate the percentage of correct answers (since it may include
some or many disguised subject relatives). With this caveat, comprehension and
production are graphically represented in (19).
13 The acquisition of Catalan relatives
195
(19) Percentage of correct production and comprehension of relative clauses
The similarity between comprehension and production is notable, even more
so if we take into account that the production of object relatives may have been
overestimated. Here we will argue that, far from being accidental, this parallelism
is to be expected under our approach.
There are some differences which the graph does not portray, namely the way
in which miscomprehension of object relatives requires that children sometimes
ignore the disambiguating morphology in the embedded verb (as found also in
Arosio et al. 2009 in Italian), while children never produce errors in agreement in
relative production; this contrast actually extends beyond relatives: subject-verb
agreement is sometimes disregarded by children in comprehension, but not in
production: see Johnson et al. (2005), Pérez-Leroux (2005) for main clauses.
Gibson’s and Morrill’s proposals are neutral with respect to whether
linguistic knowledge is put to use in production or comprehension, but, as pointed
out by an anonymous reviewer, the parallelism between production and
comprehension is expected, given that the speaker has one underlying grammar.
For the empirical domain considered, relative clauses, production and
comprehension are equally taxed. It makes sense to think that if comprehension of
a sentence is more taxing to the speaker when there is a higher number of syntactic
dependencies that s/he must keep in mind, building such a structure would also be
more taxing.3
The fact that we are analysing the children’s difficulties in comprehending
and producing object relatives in Catalan as a processing effect implies that the
grammar that we are attributing to them is fully adult-like: we do not claim that
children have any problem with relativisation per se, with wh- movement, or any
3 The discrepancy in the production and comprehension of number agreement documented by
Johnson et al. (2005), Pérez-Leroux (2005) is one of those cases in which the expected parallelism
between production and comprehension does not seem to hold, and remains a puzzle.
Anna Gavarró, Arnau Cunill, Míriam Muntané and Marc Reguant 14
196
other basic syntactic operation. Indeed, they do not always fail with object
relatives, which is what we would expect if their grammars were immature to
handle them. If we analyse the individual performance of the children tested, we
see that all of them produced adult-like relatives at least some of the time. No child
produced less than 7 adult-like subject relatives out of 10, and for object relatives
one child produced only one adult-like relative, another child produced 3, and the
rest of children produced 5 or more.
Briefly, we claim that processing resources may be more limited in children
than in adults, as shown by the results here, and such limitations have a gradual
impact on performance. Equally, in adult populations the same gradual effect of
processing load can be found: Catalan-speaking adults also produced more errors
with object relatives than with subject relatives, and under pressure we would
expect that to become more visible.
As we have argued, the processing load of subject relatives and object
relatives is structure dependent, and therefore may vary across languages, and the
analysis here extends to those cases. As mentioned above, prenominal relatives
have been shown to behave quite differently, with subject relatives being more
costly than object relatives in terms of processing: Hsiao and Gibson (2003) show
in a self-paced reading task that object relatives are processed faster than subject
relatives. Guasti (2002) references the work by Lee (1992) on the acquisition of
Chinese relatives and reports that both object and subject relatives were well
comprehended by Chinese speaking children at age 4 under certain circumstances.
The sentences in (21), taken from Hsiao and Gibson (2003), exemplify subject
and object relatives:
(20) a. Yaoching fuhao de guanyuan shinhuaibugui danshi shanyu yintsang.
invite tycoon gen official have bad intentions but good at hiding
‘The official who invited the tycoon had bad intentions but is good at hiding them.’
b. Fuhao yaoching de guanyuan shinhuaibugui danshi shanyu yintsang.
tycoon invite gen official have bad intentions but good at hiding
‘The official who the tycoon invited had bad intentions but is good at hiding.’
The derivation of the two sentences appears in (21) on the next page; the
complexity profiles of the two sentences are those in (22) and (23).
(22) Complexity profile for (20a)
4 a
3 a a
2 a
1 a
0
Yaoching fuhao de guanyuan shinhuaibugui…
15 The acquisition of Catalan relatives
197
(21) Derivations of (20a) and (20b)
Anna Gavarró, Arnau Cunill, Míriam Muntané and Marc Reguant 16
198
(23) Complexity profile for (20b)
3 a a
2 a a
1 a
0
Fuhao yaoching de guanyuan shinhuaibugui…
In terms of the complexity metric here, the subject relative in (20a) represents a
higher processing load than the object relative in (20b): in (20a) the complexity
reaches 4, and 3 at two points; in (20b) it reaches 3 at two points, but never 4. The
results of the psycholinguistic experimental work by Hsiao and Gibson can be thus
accounted for.
In Basque, a head-final language with prenominal relative clauses, Carreiras
et al. (2010) ran a self-paced reading task and an ERP experiment, and showed that
Basque subject relative clauses are not easier to process than object relatives. Let
us compare a subject relative (24a) with an object relative (24b) (examples from
Carreiras et al. 2010):
(24) a. Irakasle-ak aipatu ditu-en ikasle-a-k lagun-ak ditu.
teacher-pl mentioned has-rel student-sg-S friend-pl has
‘The student that mentioned the teachers has friends.’
b. Irakasle-a-k aipatu ditu-en ikasle-ak lagun-ak dira.
teacher-sg-S mentioned has-rel student-pl friend-pl are
‘The students that the teacher mentioned are friends.’
The derivations of (24a,b) appear in (25).
(25)
N◦ S• S◦ N• N• N◦ N◦ S•
N◦ N◦
S◦ N• N• N•
Irakasle-ak aipatu ditu-en ikasle-a-k lagun-ak ditu.
N◦ S• S◦ N• N• N◦ N◦ S•
N◦ N◦
S◦ N• N• N•
Irakasle-a-k aipatu ditu-en ikasle-ak lagun-ak dira.
17 The acquisition of Catalan relatives
199
The complexity profile in (26) corresponds to the subject relative in (24a), that in
(27) to the object relative in (24b); as can be observed, the number of dependencies
to be resolved is the same, and therefore the prediction here is that, based on
complexity, both subject and object relatives represent the same processing load.
(26) Complexity profile for (24a)
3 a a a
2 a a
1 a
0
a
Irakasle-ak aipatu ditu-en ikasle-a-k lagun-ak ditu.
(27) Complexity profile for (24b)
3 a a a
2 a a
1 a
0 a
Irakasle-a-k aipatu ditu-en ikasle-ak lagun-ak dira.
Carreiras et al. (2010) reported longer reading times for subject relatives than
for object relatives; this implies that reading time is not only limited by the
processing constraints considered here, but by other factors as well.
Recently, Gutierrez-Malgado (2011) carried out an experiment on the
comprehension of relatives with Basque-speaking children and adults. Her results
for the control SOV condition and for subject and object relatives appear in Table 6.
Table 6
Mean percentage correct by groups and sentence type
SOV SR OR
4-year-olds (# 14) 92% 58% 79.3%
5-year-olds (# 14) 95.2% 75.7% 87.85%
Adults (# 20) 98.4% 90% 97.25%
The statistical analysis of these results revealed better performance for SR
than for OR for all age groups, in line with the results of Carreiras et al. (2010).
Their experimental findings are consistent with our predictions, and not with
analyses buiding solely on the contrast between subject and object relatives
irrespective of their specific linguistic structure.
To summarise, we have presented new results on the production of relative
clauses by Catalan-speaking children, and shown that the asymmetry previously
found in comprehension between subject and object relatives also holds in
production. We have argued that a categorial metric of processing can account for
this asymmetry in a precise, non-stipulative way, and that the analysis extends to
other typologically different languages.
Anna Gavarró, Arnau Cunill, Míriam Muntané and Marc Reguant 18
200
REFERENCES
Adani, F. 2010, “Rethinking the acquisition of relative clauses in Italian: towards a grammaticality
based account”, Journal of Child Language, 38(1), 141–165.
Arosio, F., F. Adani, M. T. Guasti, 2009, “Grammatical features in the comprehension of Italian
relative clauses by children”, in: José M. Brucart, A. Gavarró, J. Solà (eds.), Merging
Features: Computation, Interpretation, and Acquisition, Oxford/New York, Oxford University
Press, 138–155.
Brown, H. D., 1971, “Children’s comprehension of relativized English sentences”, Child Development,
42, 1923–1936.
Carreiras, M., J. A. Duñabeitia, M. Vergara, I. de la Cruz-Pavía, I. Laka, 2010, “Subject relative
clauses are not universally easier to process: evidence from Basque”, Cognition, 115, 79–92.
de Villiers, J., H. B. Tager Flusberg, K. Hakuta, M. Cohen, 1979, “Children’s comprehension of
relative clauses”, Journal of Psycholinguistic Research, 17, 57–64.
Frauenfelder, U., J. Seguí, J. Mehler, 1980, “Monitoring around the relative clause”, Journal of Verbal
Learning and Verbal Behaviour, 19, 328–337.
Friedmann, N., A. Belletti, L. Rizzi, 2009, “Relativized relatives: Types of intervention in the
acquisition of A-bar dependencias”, Lingua, 119, 67–88.
Friedmann, N., A. Belletti, L. Tuller, J. Costa, M. T. Guasti, M. Lobo, E. Hamann, A. Gavarró, A. van
Hout, G. Håkansson, M. J. Ezeizabarrena, K. Jensen de López, H. K. J. van der Lely, L.
Avram, A. Sevcenco, J. Weston, I. Fattal, M. Yachini, H. Delage, A. Cunill, M. Muntané, M.
Reguant, K. Yatsushiro, U. Sauerland, J. Kainhofer, P. Hummer, H. Haider, A. Pych, A.
Miękisz, C. Contemori, K. Grohmann, S. Varlokosta, in preparation, “The child who prefers
not to produce object relatives: The acquisition of relative clause production in 16 languages”. Ms.
Gavarró, A., F. Adani, M. Ramon, S. Rusiñol, R. Sànchez, forthcoming, “La comprensió de les
clàusules de relatiu en català infantil”, Caplletra. Available as “On the comprehension of
relative clauses in child Catalan”, http://webs2002.uab.es/clt/publicacions/reports/pdf/GGT-09-11,
Universitat Autònoma de Barcelona.
Gavarró, A., A. Cunill, M. Muntané, M. Reguant, 2011, “Catalan child relative contrasts as a
processing effect”, in: M. Pîrvulescu, M. C. Cuervo, A. T. Pérez-Leroux, J. Steele, N. Strik
(eds.), Selected Proceedings of the 4th Conference on Generative Approaches to Language
Acquisition North America (GALANA 2010), Somerville, Mass., Cascadilla Press, 37–50.
Gibson, E., 1998, “Linguistic complexity: locality of syntactic dependencies”, Cognition, 68, 1–76.
Guasti, M. T., 2002, Language Acquisition. The Growth of Grammar, Cambridge, Mass., The MIT Press.
Guasti, M. T., S. Stavrakaki, F. Arosio, 2008, “Number and case in the comprehension of relative
clauses: evidence from Italian and Greek”, in: A. Gavarró, M. J. Freitas (eds.), Language
Acquisition and Development. Proceedings of GALA 2007, Newcastle, Cambridge Scholars
Publishing, 230–240.
Gutierrez-Mangado, M. J., 2011, “Children’s comprehension of relative clauses in an ergative
language: the case of Basque”, Language Acquisition, 18, 176–201.
Hsiao, F., E. Gibson, 2003, “Processing relative clauses in Chinese”, Cognition, 90, 3–27.
Johnson, M. E., 1998, “Proof nets and the complexity of processing center-embedded constructions”,
Journal of Logic, Language, and Information, 4(7), 433–447.
Johnson, V., J. de Villiers, H. Seymour, 2005, “Agreement without undestanding? The case of third
person singular /s/”, First Language, 25, 317–330.
Labelle, M., 1990, “Predication, wh-movement and the development of relative clauses”, Language
Acquisition, 1, 95–119.
Lee, T. H.-T., 1992, “The inadequacy of processing heuristics: Evidence from relative clause
acquisition in Mandarin Chinese”, in: T. H.-T. Lee (ed.), Research on Chinese Linguistics in
Hong Kong, Hong Kong, The Linguistic Society of Hong Kong, 48–86.
Morrill, G., 2000, “Incremental processing and acceptability”, Computational Linguistics, 26(3), 319–338.
19 The acquisition of Catalan relatives
201
Morrill, G., 2010, Categorial Grammar. Logical Syntax, Semantics, and Processing, Oxford/New
York, Oxford University Press.
Morrill, G., A. Gavarró, 2010, “Categorial complexity of relativisation and child Romance
performance”, in: P. Guijarro-Fuentes, L. Domínguez (eds.), New Directions in Language
Acquisition: Romance Languages in the Generative Perspective, Newcastle, Cambridge
Scholars Publishing, 365–387.
Novogrodsky, R., N. Friedmann, 2006, “The production of relative clauses in syntactic SLI: a window
to the nature of the impairment”, Advances in Speech-Language Pathology, 8, 364–375.
Pérez-Leroux, A.-T., 1995, “Resumptives in the acquisition of relative clauses”, Language
Acquisition, 4(1&2), 105–138.
Pérez-Leroux, A.-T., 2005, “Number problems in children”, in: C. Gurski (ed.), Proceedings of the
2005 Canadian Linguistics Association Annual Conference, 12 pages.
http://ling.uwo.ca/publications/CLA-ACL/CLA-ACL2005.htm
Schriefers, H., A. Friederici, K. Kühn, 1995, “The processing of locally ambiguous relative clauses in
German”, Journal of Memory and Language, 34, 499–520.
Solà, J., 2002, “Les subordinades de relatiu”, in: J. Solà, M. R. Lloret, J. Mascaró, M. Pérez-Saldanya
(eds.), Gramàtica del Català Contemporani (Vol. 3), Barcelona: Editorial Empúries, 2455–2565.
