Evaluating morphosyntactic, semantic and pragmatic abilities in the search of persistent clinical markers of SLI in Italian children

Evaluating morphosyntactic, semantic and pragmatic abilities in

the search of persistent clinical markers of SLI in Italian children.

Fabrizio Arosio

Università degli Studi di Milano-Bicocca

Chiara Branchini

Università Ca’ Foscari Venezia

Lina Barbieri

Centro di Psicomotricità di Lodi

Maria Teresa Guasti

Università degli Studi di Milano-Bicocca

This paper was supported by the CLAD project (Crosslinguistic Language

Diagnosis - TRA-STUCOR 2007-1992/001-001), founded by the European

Union. The CLAD project is part of the European "Lifelong Learning Program"

under the "Leonardo da Vinci" section.

www.cladproject.eu

1

ABSTRACT

Purpose: to extend our understanding of clinical markers of SLI to school age

Italian speaking children and investigate different language abilities useful for

defining the origin of the deficit and for differential diagnosis.

Method: experiment 1 investigated morphosyntactic abilities by eliciting the

production of direct object clitics and reflexive clitics in 19 children with SLI

and in 57 typically developing children matched for age, grammar and

vocabulary. Experiment 2 investigated their semantic and pragmatic abilities by

testing the comprehension of quantified sentences.

Results: experiment 1 showed that children with SLI fail to produce direct object

clitics and have no problems in the production of reflexive clitics. Experiment 2

showed that children with SLI had no problems in the comprehension of

felicitously uttered quantified sentences and that they are mildly weak in the

comprehension of underinformative existential quantification.

Conclusions: failure to produce direct object clitics is a good clinical marker of

SLI after age 5 years in Italian. We argue that this failure is not determined by a

pragmatic weakness nor directly dependent on prosodic factors. The results are

compatible with theories that see SLI as a deficit in the processing of complex

aspects of linguistic relations.

2

1. Introduction

Intensive work has been recently carried out to recognize and describe

linguistic phenomena that are potential clinical markers for Specific Language

Impairment (SLI), namely phenomena that correctly identify SLI subjects as

language impaired (sensitivity) and non language impaired children as such

(specificity). Crosslinguistic studies have shown that accurate identification of

children with SLI is, at least partly, language specific. While it is true that

grammatical morphology is in general vulnerable in these children, specific

areas of grammar are particularly challenging, depending on the language. Initial

studies have been carried out with English speaking children and successively,

attention has turned to other languages (see Leonard, 1998 and references found

there). From studies with Italian speaking children with SLI, it has emerged that

(present tense) third-person plural inflections and the use of functional words

such as articles and clitic pronouns is particularly vulnerable for these children

(Bortolini & Leonard, 1996; Bortolini, Leonard & Caselli., 1998; Bottari,

Cipriani, Chilosi & Pfanner, 2001; Bottari, Cipriani & Chilosi, 2001; Cipriani,

Chilosi, Bottari, Pfanner, Poli & Sarno, 1991; Leonard & Bortolini, 1998;

Leonard, Bortolini, Caselli, McGregor, & Sabbadini, 1992; Leonard, Bortolini,

Caselli, & Sabbadini, 1993; Sabbadini, Volterra, Leonard, & Campagnoli,

1987). In particular, the production of direct object clitics has been found to be

an excellent clinical marker of SLI in preschool children (3;7 - 5;5 years) with

90% sensitivity and 100% specificity (Bortolini, Arfé, Caselli, Degasperi,

Deevy, & Leonard, 2006). Other clinical markers for Italian have also been

discovered, such as production of present 3rd person plural inflection and

repetition of non-words, but their sensitivity or specificity is lower. Children

with SLI, as typically developing children, improve their language, and a

clinical marker at 5 may no longer be so at later ages. Therefore, if the notion of

clinical marker has to be useful, we need to establish what happens to children

with SLI during development. 3

Clinical markers are generally found in the area of morphosyntax or

phonology, as children are particularly vulnerable in these domains, but other

areas may also be affected, such as lexical semantics or pragmatics (Rice &

Wexler, 1996; Aguilar-Mediavilla, Sanz-Torrent & Serra-Raventos, 2002;

McGregor, Newman, Reilly & Capone, 2002; see Leonard 1998 and Cummings

2008 for an overview). In fact, it is known that language impairment is a broad

term and different SLI sub-groups have been identified on the basis of the

described dissociations (van der Lely, 1996, 2005; Conti-Ramsden, Crutchlet &

Botting, 1997; Conti-Ramsden & Botting, 1999; Rapin & Allen, 1983;

Friedmann & Novogrodsky, 2008; Dockrell, Messer & Murphy, 2005; Schaffer,

2003a, 2003b). A number of studies has found SLI subgroups with pragmatic

deficits, such as failures in logical inferences and inappropriate conversational

responses, that have been claimed not to be a consequence of deficits in core

grammatical skills (Bishop, 1998, 2000; Bishop, Chan, Hartley, Adams, & Weir,

2000; Botting and Adams, 2005). Thus, while looking for clinical markers in the

domain of syntax or morphosyntax, it is important to investigate also other areas

to establish how broad the deficit is. This is particularly important, as most of

the studies on different subtype of SLI are on English and not much research has

been carried out with other languages to define different linguistic profiles.

Starting from this background, our study has two goals: (1) extend our

understanding of clinical markers to school age Italian speaking children; (2)

investigate different language abilities in these children useful for defining the

origin of the deficit and for differential diagnosis. We will achieve these goals

through two tasks, one involving production of direct object and reflexive clitics

and one involving comprehension of quantified sentences tapping semantic and

pragmatic knowledge. Our choice to extend the investigation of clinical markers

to school aged children is motivated by the fact that many children in Italy fail

to be identified as SLI in the pre-school years. Their language difficulties start to

be emerge in school when they start to read and write, because SLI is often 4

associated with dyslexia (Bishop & Snowling, 2004; Catts, 1993; Catts, Adlof,

Hogan & Weismer, 2005; Cantiani, Lorusso, Perego, Molteni & Guasti in press;

McArthur, Hogben, Edwards, Heath, & Mengler, 2000; Flax, Realpe-Bonilla,

Hirsch, Brzustowicz, Bartlett & Tallal, 2003). Since a diagnosis of dyslexia does

not include language tests in Italy, it is important to find out diagnostic markers

of SLI in school years. Our choice of the tests finds its rationale in the

observation that some difficulties in children with SLI are persistent (Weiner,

1974; Kerschensteiner & Huber, 1975; Clegg, Hollis, Mawhood, & Rutter,

2005; Records, Tomblin, & Freese, 1992; Tomblin, Zhang, Buckwalter, &

O’Brien, 2003; Stothard, Snowling, Bishop, Chipchase, & Kaplan, 1998) and

clitic production is likely to be one of these. Since clitics production requires a

range of different linguistic abilities (phonological, morphosyntactic and

semantic-pragmatic), we designed our tests in specific way to examine each of

these abilities.

Our article is organized as follows. In section 2, we illustrate the properties of

clitics, what is known about their acquisition in TD children and in children with

SLI. Then, we turn to theories that account for the failure to use clitics by

children with SLI. Next, in section 3 we discuss quantified sentences and the

domain of language their comprehension involves. Finally, in section 4, we

present our studies with school aged children affected by SLI and then discuss

the implications of our findings in the light of our previous literature.

2. Italian clitic pronouns: description, acquisition and challenges.

In this section, we are going to present the properties of clitics in Italian and

to illustrate which domains of language are involved in their production. Next,

we will discuss how clitics are acquired in typically developing children and in

children with SLI. Finally, we present a previous account of why clitics are

challenging for children with SLI.

5

2.1. Description: some levels of complexity

The production of direct object (DO) clitics involves a number of highly

sophisticated linguistic pieces of competence in Italian. They are realised as

weak phonological monosyllabic morphemes and, at the syntactic level, their

placement depends on the finiteness of the verb. DO clitics appear in a preverbal

position giving rise to the uncanonical S(ubject) O(bject) V(erb) word order

when the verb of the sentence is finite, as in (1):

(1) Il bambino lo lava

The child DO-CLIT3SgMasc washes

The child is washing him

When the verb of the sentence is not finite, direct object clitics cliticize on the

verb in a post verbal position as shown in (2).

(2) Il bambino ha detto di lavarlo

The child has said to wash_DO-CLIT3SgMasc

The child said to wash him

Morphosyntactically, DO clitics are marked for person, number and gender.

When the tense of the sentence is the Present Perfect, the past participle on the

verb must agree in number and gender with them, as in (3) and (4).

(3) Il bambino lo ha lavato

The child DO-CLIT3SgMasc has washedSgMasc

The child has washed him

6

(4) Il bambino la ha lavata

The child DO-CLIT3SgFem has washedSgFem

The child has washed her

At the pragmatic level, DO clitics denote a discourse topic (like all pronouns),

a referent that has been already introduced in the discourse. Therefore, the

production of a DO clitic requires knowledge that it should denote a discourse

topic and what the topic of discourse is. Italian has two series of overt pronouns,

clitics and tonic pronouns. Both must refer to discourse topics, but they fill

different discourse functions. Clitics are not felicitous in contrastive contexts, as

shown by the examples below, where the use of the tonic pronoun instead of the

DO clitic appears felicitous.1

(5) Il bambino lo lava, *non il gatto

The child DO-CLIT3SgMasc washes, not the cat

The child is washing him, not the cat

(6) Il bambino lava lui, non il gatto

The child washes him, not the cat

The child is washing him, not the cat

The production of reflexive clitic pronouns (RE clitics) also requires a

number of linguistic pieces of competence. At a syntactic level, like DO clitics,

RE clitics occur in a pre-verbal position when the verb of the sentence is finite

and in a post verbal position when the verb is not finite. They are not marked for

gender but for person and number, with the exception of third person “si” (itself)

which is used for both singular and plural. Interestingly, when the tense of the 1 Notice that tonic pronouns, unlike clitics, must appear in the postverbal position, like full DPs. Therefore, they

are syntactically less complex.

7

sentence is Present Perfect, RE clitics select for the auxiliary “essere” and the

past participle on the verb must agree in number and gender with the referent of

the clitic, i.e., the subject of the sentence, as shown in the examples below2

(7) Il bambino si lava

The child RE-CLIT3 washes

The child is washing himself

(8) a. Il bambino si è lavato

The childSgMasc RE-CLIT3 is washedSgMasc

The child has washed himself

b. La bambina si è lavata

The childSgFem RE-CLIT3 is washedSgFem

The girl has washed herself

c. I bambini si sono lavati

The childPlMasc RE-CLIT3 are washedPlMasc

The children have washed themselves

d. Le bambine si sono lavate

The childPlFem RE-CLIT3 are washedPlFem

The girls have washed themselves

Analogously to DO clitics, the use of RE clitics is subject to pragmatic

constraints, being it banned from contrastive contexts, as shown by the examples

below, where the use of the reflexive tonic pronoun appears felicitous. 2 The 3rd person reflexive clitic is neither marked for gender nor for number; these features are determined by

the subject of the sentence.

8

(9) Il bambino si è lavato, *non il gatto

The child RE-CLIT3Sg is washed, not the cat

The child has washed himself, not the cat

(10) Il bambino ha lavato se stesso, non il gatto

The child has washed RE-selfSgMasc, not the cat

The child has washed himself, not the cat

One important difference to be noticed between (9) and (10) is the occurrence

of the auxiliary “essere” in (9) and the occurrence of the auxiliary “avere” in

(10), where we find the pronominal tonic counterpart “se stesso”. This fact

shows that the contrast between (9) and (10) is not only dependent on the

different pragmatic constraints for the use of the clitic pronoun and its tonic

counterpart, but also on the different underlying syntactic structures and the

related fact that the two pronouns are two different linguistic entities. Since we

have seen that Present Perfect sentences with DO clitics select for the auxiliary

“avere” and Present Perfect sentences with RE clitics select for auxiliary

“essere”, reflexive clitics are also different from DO clitics and sentences in

which these two different pronouns occur are likely to have distinct underlying

syntactic structures, as argued in Burzio (1986), among others. Another well

known difference between DO clitics and RE clitics concerns the fact that

reflexive clitics must be interpreted as coreferential with the sentence subject

while DO clitics must not.

2.2. The acquisition of clitics in Italian

Italian typically developing (TD) children start to produce DO clitics at

around two years, but optionally omit them up to age 4 years. They rarely make

errors in their placement or with their morphology and never replace them with 9

tonic pronouns (Guasti, 1993/94; Leonini, 2006; Schaeffer, 2000; Caprin &

Guasti 2009; Moscati & Tedeschi, 2009). Concerning reflexive clitics, Italian

TD children use them from 2 years; no study has investigated whether they are

used consistently from that age, but likely they are much more used than DO

clitics, as no study on the development of morphosyntax reports omission.

Moreover, out conjecture finds some indirect support from a study on the

acquisition of French pronouns, a language similar to Italian with respect to

clitics. In this language an asymmetry in the use of DO and RE clitics has been

found with an advantage for reflexive clitics (Zesiger, Zesiger, Arabatzi,

Baranzini, Cronel-Ohayon, Franck, Frauenfelder, Hamann, Rizzi, L., 2010).

Italian children have no problem with placement of reflexive clitics and with the

selection of the auxiliary “essere” already around two years (Snyder et. al, 1995;

Caprin & Guasti, 2009).

Italian preschool children with SLI frequently fail to produce DO clitics and

the most persistent error is omission (Leonard et al., 1992; Leonard & Bortolini,

1998), as shown in the ungrammatical sentence below, where the internal

argument of the verb is not overtly realized.

(11) * Il bambino Ø prende Ø

The child Ø catches Ø

This failure has been shown to be a good clinical marker of SLI in Italian

preschool children since Bortolini and colleagues found that DO clitic omission

identified 90.91% of SLI preschool children as language impaired (sensitivity)

and 100% of TD age controls as not language impaired (specificity) in an

elicitation task (Bortolini, Caselli, Deevy, & Leonard, 2002; Bortolini et al.,

2006).

Data concerning the production of RE clitics in Italian children with SLI are

scarce and include some confounding factors. There is a single study by Pozzan 10

(2006), with few subjects, reporting that the production of 3rd singular DO clitics

is not significantly different from the production of 3rd singular RE clitics. In her

study Pozzan argues that this might depend on the fact that four of the ten

reflexives elicited in the task were not proper reflexives since they were

occurring in inherent reflexive predicates; indeed, when these were removed

from her analysis, Pozzan reported a marginal tendency towards a better

preservation of 3rd singular RE clitics in comparison to 3rd singular DO clitics, as

attested for French children affected by SLI in a number of studies (Jakubowicz,

Nash, Rigaut, & Gerard, 1998; Audollent & Tuller, 2003). In summary, DO

clitics are used from 2 years, but optionally omitted up at most 4 years by TD

children and they have been found to be a very good clinical marker of SLI in

pre-school years. Less is known about RE clitics, which are also used from 2

years. They may be employed more consistently by TD children, but their use

by children with SLI is still controversial.

2.3. Challenges: the prosodic account and the rationale for our study

In this section, we are going to discuss how failure to produce clitics in Italian

children with SLI has been interpreted. In their study, Bortolini, and colleagues

(2006) (see also Leonard, Sabbadini, Leonard, & Volterra, 1987; Leonard,

Sabbadini, Volterra, & Leonard, 1988; Bortolini et al., 2002) explained Italian

SLI children’s failure to produce DO clitics in prosodic terms. They argued that

the factor responsible for their omission was a difficulty in the production of

non-final weak syllables. Since preverbal DO clitics are the first weak syllable

of a prosodic unit formed by the clitic and the verb (clitic group or prosodic

word), children with SLI failed to produce them and only produced the verb,

which often is a bisyllabic element with a trochaic structure (strong-weak), i.e.,

they failed to produce an initial weak syllable of a prosodic unit. Bortolini and

colleagues claimed that their hypothesis was corroborated by the fact that these

children also omitted the first weak syllable of trisyllabic words with the 11

prosodic structure weak-strong-weak (the same prosodic structure of a clitic plus

bisyllabic verb) (e.g. the first syllable of “farfalla” (FAR’FALLA, ‘butterfly’).

At the same time, they noticed that prosody could not be the sole factor

underlying the difficulty in producing clitics, since children with SLI omitted the

weak non-final syllable staying for a clitic with greater frequency than they

omitted the non-final weak syllable of trisyllabic words.

In this work, we provide another piece of evidence against the view that

prosody is the main factor responsible for failure to use clitics. In particular, we

will compare different types of clitics with the same prosodic properties: DO

clitics, as in Bortolini and colleagues, and RE clitics. As we mentioned, Pozzan,

the single study including reflexive clitics, did not find strong evidence that the

production of 3rd singular RE clitics was easier than the production of 3rd

singular DO clitics, but her investigation included a confounding factor that

limits the interpretation of her data.

RE clitics share a number of features with DO clitics, for instance the fact that

they occur pre-verbally with finite verbs. They are also different from DO

clitics, for instance they select for a different auxiliary in present perfect

constructions than DO clitics do. Moreover, while 3rd person DO clitics display

gender and number features, 3rd person RE clitics do not. Establishing whether

clitics are all problematic for children with SLI can shed some light on what

causes the underlying disorder consist of.

If prosodic limitations alone cannot be held responsible for failure to use

clitics, as already hinted by Bortolini and colleagues, what else can be

advocated? In order to identify other possible sources of difficulties, beyond

investigating other clitics, we will evaluate the vulnerability of a number of

other specific linguistic phenomena in the semantic and pragmatic domains. The

use of clitics is also constrained by pragmatic restrictions as clitics are pronouns

used to refer to a discourse topic. In order to investigate whether the failure to

produce DO clitics is dependent on deficits of these pragmatic skills or if these 12

somehow contribute to the difficulty, we examine children’s semantic-pragmatic

competence in two ways. First, we investigate children’s use of the null subject

in sentences, i.e., a null pronoun required to denote a discourse topic. Second,

we examine a different aspect of the semantic-pragmatic competence, the one

involved in the comprehension of quantified sentences. This will help us in

understanding whether semantic-pragmatics is an area of strength or weakness

in these children and whether pragmatic limitation may contribute to the deficit

with clitics. In so doing, we will also delineate the profile of a population of

children with SLI along a number of linguistic dimensions. Next, we turn to the

description of the semantic-pragmatic aspects examined in our study.

3. The comprehension of quantified sentences

In order to gain insights into semantic competence of Italian school age SLI

children and investigate their ability to compute end evaluate additional unsaid

content associated to linguistic utterances, in particular scalar implicatures, we

examined the comprehension of sentences containing quantifiers in subject

position, as in (12) and (13) below.

(12) Qualche mela è nelle scatole

Some apple is in-the boxes

Some apples are in the boxes

(13) Tutte le mele sono nelle scatole

All the apples are in-the boxes

All apples are in the boxes

3.1. The problem

To understand the meaning of sentences like (12) or (13), we need to know

the quantifier meaning, namely the set relation denoted by quantifier that is 13

required to hold for the sentence to be true. For instance, in (12), the existential

quantifier “qualche” (some) requires the intersection between the set of things

that are apples and the set of things that are in the boxes to be a non empty set.

However, it is well known that this might not be enough in order to correctly

understand what speakers consistently intend to convey by the use of (12)

(Horn, 1972). In fact, (12) is consistently rejected as false by Italian adult

speakers when it is uttered in a context where all the apples are in the boxes,

though it is true according to the set relation denoted by the quantifier “qualche“

(some). In fact, by the use of (12), Italian adult speakers intend the meaning

conveyed by (14), which is false in the context, where all apples are in the

boxes.

(14) Qualche mela è nelle scatole, ma non tutte

Some apple is in-the boxes, but not all

Some apples are in the boxes, but not all

In the linguistic literature we find a long lasting debate on how the additional

statement not all apples are in the boxes, conveyed by the use of (12) and not

explicitly said, is derived (see Sauerland (2011) and citations found there for a

clear cut overview). According to the Pragmatic Theory, based on the

pioneering work of Horn (1972) and Grice (1975, 1989), the unsaid content is

not directly part of the sentence truth conditional meaning, but it is rather

derived from conversational principles requiring the speaker to be

conversationally cooperative and to use the most informative statement in a

given context: if the speaker chooses to use (12) she has no evidence for saying

that all apples are in the boxes or she knows that it is not the case that all apples

are in the boxes. Notice that the utterance of the sentence “all apples are in the

boxes” is stronger than the utterance of (12), since (12) is true in a larger set of

situations than “all apples are in the boxes”, and therefore is less informative. In 14

the literature, the additional statement not all the apples are in the boxes,

conveyed by the use of (12), is called scalar implicature since some and all are

expressions that are members of the same entailment scale (Horn, 1972): all is

higher in the scale than some since (13) entails (12). In other words, when “all

apples are in the boxes” is true, it is also true that “some apples are in the

boxes”. According to this theory a scalar implicature is thus derived by means of

a pragmatic reasoning on the speaker intentions.

3.2. The comprehension of quantified sentences: Italian Children and

Children with SLI

Italian five years old TD children are adult-like in the comprehension of

quantifiers which do not require the generation of a scalar implicature like

“tutti” (all) and “nessuno” (no) but they have problems in the comprehension of

sentences with quantifiers that trigger a scalar implicature like “qualche”

(some). They fail to reject under-informative some statements and their

performance falls into a bimodal distribution: children tend either to consistently

accept under informative SOME-statements or to consistently reject them

(Guasti, Chierchia, Crain, Foppolo, Gualmini, Meroni, 2005; Foppolo, Guasti,

Chierchia, in press). Italian children reach adultlike performance in the

comprehension of under-informative some statements at 7 years. However, it has

been shown that the experimental setting modulates this difficulty.

Concerning SLI children, Katsos and colleagues (Katsos, Roqueta, Estevan,

& Cummins, 2011) reported that Spanish-speaking children with SLI between 4

and 9 years of age performed worse than a group of age-matched TD children

and as well as a group of language-matched TD children. However, in this study

the children with SLI were likely to be not homogeneous, displaying different

linguistic deficits, and their age range was so wide that the SLI group included

children who certainly had problems with scalar implicatures, as even typically

developing children have problems up to 7 years. Another study by Skordi and 15

colleagues (Skordi, Katsos, Marshall, van der Lely, submitted) investigated the

comprehension of quantifiers in older English children with SLI. This study

reported that English children with G(rammatical)-SLI between 11 and 13 years

of age, who were severely impaired in their morphosyntactic and grammatical

abilities, performed lower than all control groups in the comprehension of

quantifiers which do not require the generation of a scalar implicature.

Moreover, Skordi and colleagues found that SLI children performed lower than

age-matched and vocabulary-matched TDs but better than younger

grammatically matched TDs in the comprehension of under-informative

utterances. All in all, the comprehension of quantified sentences seem to be only

mildly affected in Spanish and English children with SLI, but some confounding

factors may be present in previous studies.

3.3. The rationale of our study

Since the production of clitics is also constrained by pragmatic restrictions for

the use of nouns and pronouns in discourse, it is important to evaluate whether a

more general pragmatic failure contribute to clitic production. In order to

establish whether and how semantic-pragmatic contribute to the linguistic

deficit, we tested the same children with the clitic production task and a task

tapping semantic-pragmatic competence. We did so with a homogeneous group

of children with SLI, all school aged children, and all selected on the basis of

specific criteria.

We examined whether children understand and evaluate the correct set

relations associated with the quantifiers “qualche” (some) and “tutti” (all);

moreover, we investigated children’s abilities to derive and evaluate scalar

implicatures by examining their comprehension of under-informative SOME

statements containing the quantifier “qualche” (some).

16

4. Our Study

Our study aim at investigating whether the failure to produce DO clitics

persists during school years in Italian SLI children. Since DO clitics and RE

clitics share the same prosodic features, we compared their production to

investigate whether the prosodic characteristics of DO clitics is per se a problem

for Italian children with SLI. Additionally we examined whether this failure

might also depend on a pragmatic weakness in understanding what the discourse

topic is and in understanding sentences including quantified sentences.

Our study includes two experiments. In experiment 1 we tested the

production of 3rd singular DO clitics (3sDO-clitics) and 3rd singular RE clitics

(3sRE-clitics). In experiment 2 we tested the comprehension of the universal

quantifier “tutti” (ALL) and the existential quantifier “qualche” (SOME).

4.1. Experiment 1: Clitic Production

4.1.1. Subjects

One group of 16 school-aged Italian monolingual children with SLI (mean

age 7,3 years, range 72-118 months, 3 females) were tested in the production of

clitics pronouns; children were also tested in the standardised grammar test

TCGB (Chilosi, Cipriani, Giorgi, Fazzi, Pfanner, 1995), in the standardised

receptive vocabulary test PPVT (Dunn, L.M. & Dunn L.M. (Italian version by

Stella, Pizzioli & Tressoldi, 2000)) and in the standardised Raven’s Coloured

Progressive Matrices test (Raven, Court, & Raven, 1998; Belacchi, Scalisi,

Cannoni & Cornoldi, 2008) in order to have an homogeneous measure of their

verbal and non verbal cognitive abilities. SLI children were recruited from

speech therapy centres in the area of Milano and Pavia (Italy), they had been

diagnosed as being specifically language impaired based on standard inclusion

and exclusion criteria by expert clinicians (ICD-10; World Health Organization,

1992, 2000) and were receiving clinical services. In order to participate to the 17

study, SLI children had to score less than 2 SD below the mean score for their

age in the TCGB test or in the PPVT test; SLI children who scored 1.5 SD

below the mean score for their age in both tests were also admitted to the study.

All children had non verbal IQ within normal limits (IQ>85 at Raven).

Children with SLI were matched to forty-eight typically developing children

on the basis of chronological age (± 3 months), of grammatical age as measured

in terms of score at the TCGB test (± 3 points) and of vocabulary age as

measured in terms of score at the PVVT test (± 3 points). Control children were

recruited from schools and kindergartens in the same residential areas as the

children in the SLI group. All children scored within 1 SD from the mean score

for their age on the TCGB and the PVVT tests. No child was receiving clinical

services and all children have non verbal IQ within normal limits (IQ>85 at

Raven). All children with SLI were individually matched to typically developing

peers; gender matching was also observed. Group data are represented in Table

1 below.

Subjects

Age Mean

in months (SD)

PPVT raw score

Mean (SD)

PPVT Z Score Mean

TCGB raw score

Mean (SD)

TCGB Z Score Mean

SLI (n16) 87.44 (13.23)

74.13 (25.03)

-1.34 13.13 (6.14)

3.64

CA (n16) 87.56 (11.99)

108.30 (22.3)

0.21 4.41 (4.15)

0.10

VA (n16) 64.13 (13.89)

75.56 (24.37)

-0.20 12.05 (7.95)

-0.20

GA (n16) 63.12 (7.96)

74.69 (20.48)

-0.43 13.06 (6.74)

-0.08

Table 1: Children group data for experiment 1

18

No differences between the SLI group and the TD groups were found according

to a number of one-way ANOVAs: no age differences between SLI children and

CA children (p=.9281), no difference for raw scores at the PVVT test between

SLI children and VA children (p=.8636), no differences in raw scores at the

TCGB test between SLI children and GA children (p=1).

4.1.2. Materials and Procedure

Children were tested in two different sessions on different days within two

weeks. TD children were tested in a quite room at their schools, SLI children in

a quite room at their speech therapy centres. Informed consent prior to testing

was collected from children’s parents. Parents, educators and speech therapists

have been informed of the results of the study during dedicated meetings. The

study was approved by the Ethics committee of the University of Milano-

Bicocca according to the standards of the Helsinki Declaration (1964).

In experiment 1 we tested SLI children’s production of 3rd singular DO clitics

(3sDO-clitics) and 3rd person RE clitics (3sRE-clitics). We elicited 8 sentences

in the simple present containing a 3sDO-clitic as in (1), 8 sentences in the

present perfect containing a 3sDO-clitic as in (3), where the past participle

agrees in number and gender with the clitic, 5 simple present sentences

containing a proper 3sRE-clitic as in (7). Half of the 3sDO-clitics were

masculine half were feminine. All verbs used were obligatory transitive. During

the elicitation task, children were presented pairs of drawings on the monitor of

a PC administrating the experiment. When the first drawing appeared on the

screen, children heard the description of the event represented in the drawing.

Descriptions were digitally recorded by a female native speaker of Italian and

played through loudspeakers connected to the PC. When the second drawing

appeared on the screen, children heard a question eliciting a sentence containing

a clitic pronoun. Eliciting questions were also digitally recorded and played

19

through loudspeakers. A trial example for the elicitation of a sentence containing

a 3sDO-clitic is given in Figure 1 below.

Drawing 1: Drawing 2: The child has caught

the butterfly A child is trying to

catch a butterfly

In questa storia un bambino Cosa ha fatto il bambino vuole prendere una farfalla alla farfalla? In this story a child wants What did the child to catch a butterfly do to the butterfly?

EXPECTED ANSWER:

L’ ha presa (he) ItCL’has caught

Figure 1: Example of an experimental trial.

The 21 experimental trials were randomly ordered. Three familiarization trials

eliciting the production of clitics preceded the experimental session; if

necessary, feedback was given during the familiarization session.

Crucially, in the experimental context, a felicitous answer to the question

eliciting the production of clitics is a sentence with a null subject and not a

sentence in which the full lexical subject is infelicitously repeated. This allowed

us to investigate children’s ability to produce a sentence with a null subject as

required by pragmatic restrictions for the use of pronouns in discourse.

4.1.3. Results of Experiment 1

4.1.3.1. Response Coding

Children’s responses were first classified into six different categories.

Responses were considered as Target when they matched the target responses 20

(i.e. the correct clitic was used); they were classified as Arg_Clitic, when a

correct or incorrect clitic was used (wrong gender or number); they were

classified as Ind_Clitic when a grammatical sentence containing an indirect

object clitic instead of a DO clitic or a RE clitic was produced; they were

classified as Omission when the clitic was omitted, which is ungrammatical in

Italian; they were classifieds as Full DP when a sentence with a full DP object

(determiner plus noun) rather than a clitic was produced; finally, they were

classified as Other when a sentence not fitting in any of the previous category

was produced (generally sentences that were irrelevant). Additionally, according

to a second layer categorization, children’s responses were classified as

containing or not an overt subject (pro). All productions were digitally recorded,

scored by one experimenter and checked by two additional ones; discrepancies

were resolved by an agreeing discussion. A list of children’s responses

categories for 3sDO-clitics together with response examples is given in Table 2.

For 3sRE-clitics, children produced Target sentences (correct clitic in the

correct position) and irrelevant sentences only.

21

Table 2: children’s responses categories and example productions

CATEGORY CLITIC pro

USE EXAMPLE PRODUCTION

Target 3sDO-clitic pro

La prende

(he) ItFemSg catches

Arg_Clitic 3sDO-clitic pro

Lo prende

(he) ItMascSg catches

Le prende

(he) ItFemPl catches

Ind_Clitic 3sDO-clitic pro

Gli prende la coda

(he) ItMascPl_dat catches the tail

Omission 3sDO-clitic pro

Prende

(he) catches

Full DP 3sDO-clitic pro

Prende la farfalla

(he) catches the butterfly

Other 3sDO-clitic -

La farfalla vola

the butterfly flies

4.1.3.2. Analyses of the Use of Clitics

Raw percentages for categories of responses of SLI children’s and control

groups are represented in Figure 2 and in the Table 3 below.

22

FIGURE 2: PRECENTAGES OF PRODUCED STRUCTURES  IN THE CLITIC ELITITATION TASK PRODUCTION TASK 

3sDO-clitics 3sRE-clitics

TARGET CL_arg CL_ind Full DP Omission Others TARGET Others

Mean (SD)

Mean (SD)

Mean (SD)

Mean (SD)

Mean (SD)

Mean (SD)

Mean (SD)

Mean (SD)

SLI 40.07 (29.33)

45.22 (32.57)

1.47 (4.70)

33.82 (27.5)

8.45 (15.77)

4.77 (6.82)

84.70 (30.43)

15.3 (3.24)

CA 88.24 (24.19)

88.24 (24.19)

0.00 (0.00)

2.34 (4.49)

1.47 (4.70)

1.47 (2.73)

92.94 (24.43)

7.06 (1.34)

VA 78.67 (24.71)

82.72 (25.34)

1.83 (2.93)

5.14 (7.40)

1.83 (5.30)

1.83 (3.67)

91.76 (24.55)

8.24 (1.63)

GA 77.94 (22.87)

81.98 (23.78)

2.94 (4.99)

5.14 (7.06)

1.47 (4.70)

1.47 (2.73)

91.76 (24.55)

8.24 (1.63)

Table 3: raw percentages of produced structures

23

As we can observe, contrary to TD children with SLI children do not use

3sDO-clitics consistently and tend to produce a full DP in its canonical post

verbal position. They do so more than all three groups of control children. Rate

of omission of the verb internal argument is very low. By contrast, 3sRE-clitics

are correctly used. In our analysis we first investigated group differences in the

production of 3sDO-clitics and in the production of 3sRE-clitics. Group

differences were analysed by means of three different pairwise comparisons: (i)

SLI children vs CA children, (ii) SLI children vs VA children, (iii) SLI children

vs GA children. For each response category (Target, Arg_Clitic, Ind_Clitic,

Omission, Full DP, Other) we carried out repeated measure logistic regression

analyses in a mixed model, with Group (SLI vs TD) and Sentence Type (present

vs present perfect) as fixed factors, subject and items as random factors, and

response accuracy as dependent variable. We used a backwards elimination

procedure to compare the goodness of fit of the models with the Group factor

(Baayen, 2008). All statistical analyses were conducted using R (version 2.13.1);

for mixed modeling LanguageR package (version 1.4) was used (Baayen, 2011).

We found a Group effect between SLI children and all three control groups

on the following response categories: Target, CL_arg, Full DP. Estimated

coefficients, their standard errors, Z-values and associated p-values for the

Group factor are summarized in the Table 4 below.

24

CA comparison Estimate SE Z P

Target Analysis

(Intercept) -0.6575 0.5255 -1.251 0.211 Group 3.9091 0.7795 5.015 < 0.001

CL_arg Analysis

(Intercept) -0.3830 0.5665 -0.676 0.499 Group 3.7437 0.8403 4.455 < 0.001

FULL DP Analysis

(Intercept) -0.8209 0.4771 -1.720 0.085 Group -2.9357 0.7551 -3.888 0.001

VA comparison Estimate SE Z P

Target Analysis

(Intercept) -0.5401 0.3966 -1.362 0.173 Group 2.6464 0.5476 4.833 < 0.001

CL_arg Analysis

(Intercept) -0.2936 0.4546 -0.646 0.518 Group 3.0052 0.6486 4.633 < 0.001

FULL DP Analysis

(Intercept) -0.7789 0.3737 -2.084 0.037 Group -2.7612 0.5846 -4.723 < 0.001

GA comparison Estimate SE Z P

Target Analysis

(Intercept) -0.5021 0.3594 -1.397 0.162 Group 2.4020 0.4834 4.969 < 0.001

CL_arg Analysis

(Intercept) -0.2639 0.4162 -0.634 0.526 Group 2.6905 0.5871 4.583 < 0.001

FULL DP Analysis

(Intercept) -0.7769 0.3696 -2.102 0.035 Group -2.7317 0.5756 -4.746 < 0.001

Table 4: 3sDO-clitics production, estimated coefficients for Group effect in

the CA, VA and GA comparisons.

As indicated by the coefficients reported in Table 5, CA children, younger VA

and GA children produced significantly more Target structures, more CL_arg

structures and less structures where the clitic pronoun was substituted by a full

DP than SLI children did. We did not find any effect of Sentence Type (Present

25

vs Present Perfect) nor any interaction between Sentence Type and Group. In

order to investigate whether age is an influential factor for the use of DO clitics

in children with SLI, for each response category we performed a number of

repeated measure logistic regression analyses in a mixed model, with subjects’

Age and Group (SLI vs TD) as independent fixed factors, subject and items as

random factors, and response accuracy as dependent variable. These analyses

did not revealed any effect of age nor any interaction between age and group.

Additional analyses with age clusters also did not revealed any Age effect nor

any interaction between Age and Group. Thus, we can conclude that the problem

with clitics affects children in our group regardless of their age, that is, it is a

persistent problem for a while at least.

In the analyses of 3sRE-clitic production, we did not find any effect of Group

on any produced structure in any comparison. No age differences or interactions

between age and group differences were found according to a number of

repeated measure logistic regression analyses.

4.1.3.3 Direct object clitic production failure as a persistent clinical marker

of SLI in Italian.

An important outcome for research on SLI is the identification of linguistic

phenomena that are clinical markers of SLI, namely, that identify SLI children

with respect to their same-age TD peers (Rice & Wexler, 1996; Fletcher &

Peters, 1984; Gavin, Klee, & Membrino, 1993). What distinguishes the

goodness of a clinical marker is its sensitivity and its specificity, namely the

probability of identifying SLI subjects as true cases of SLI and the probability of

identifying TD subjects as true cases of typical development as opposed to SLI.

Bortolini and colleagues (2002, 2006) found that the failure to produce DO

clitics is a good clinical marker of SLI in Italian children, aged from 3;7 years to

5;6 years, since it shows a good sensibility and the specificity rate. Our study

aimed at extending this result to older children, since a number of studies have 26

shown that SLI is persistent (Weiner, 1974; Kerschensteiner & Huber, 1975;

Clegg, Hollis, Mawhood, & Rutter, 2005; Records, Tomblin, & Freese, 1992;

Tomblin, Zhang, Buckwalter, & O’Brien, 2003; Stothard, Snowling, Bishop,

Chipchase, & Kaplan, 1998). In order to evaluate how well the failure to

produce 3sDO-clitics could distinguish older SLI children from same age peers,

we performed a logistic regression analysis (Bortolini et al., 2006) on Group

(SLI vs CA) with subject’s percentages of accuracy of production of Target

sentences containing 3sDO-clitics as covariate variable. According to the

analysis, our test obtained a good sensitivity value and a good specificity value

(Plante & Vance, 1994) since it correctly identified 15 SLI children out of 16

(93.8%) and it accurately identified 15 TD children out of 16 (93.8%) as not

being language impaired. This piece of data confirms that the failure to produce

DO clitic pronouns is a good clinical marker of SLI in Italian even at school age.

4.1.3.4. Analysis of the Use of Null Subjects.

In order to evaluate whether the use of pronouns is generally problematic for

children with SLI and whether there is a problem in resuming discourse topics,

we analysed the use of subjects. Raw percentages of sentences including a null

subject (pro) produced by children with SLI and TD control groups are

represented in Figure 3 and in Table 5 below.

27

Table 5 & Figure 3: raw percentages of produced null subjects

As we can observe, TD children and children with SLI both produced a

high rate of sentences with a null subject; in our statistical analyses, we did not

find any effect of Group in any comparison.

4.1.4. Summary of Experiment 1.

Our data indicates that children with SLI differ from age matched TD

children and younger language matched TD children in the production of DO

clitic pronouns. Contrary to TD children, SLI children do not use the clitic and

tend to produce a full DP in its canonical post verbal position. This extends

Bortolini et al.’s finding to older Italian SLI children, confirming that clitics are

challenging even after 5 years. However, in our SLI group, the failure takes on a

different manifestation than it did in Bortolini et al., where SLI children were

younger than in our group. In fact, our children with SLI tend to produce the

corresponding full DP instead of omitting the internal verbal argument as

Bortolini and colleagues found. Interestingly, this fact has been also observed in

pro

Mean (SD)

SLI 87.10 (33.01)

CA 82.10 (38.53)

VA 96.81 (17.62)

GA 94.04 (23.80)

100

80

60

40

20

0

28

young (age 3 to 5 years) Spanish children with SLI (Bedore & Leonard, 2001).

SLI children do not differ from TD children in the production of reflexive clitic

pronouns nor in the use of null subjects. Finally, the failure to produce clitics

remain a good clinical marker of SLI in the school years with a high sensitivity

and specificity, not much different from those observed in the pre-school aged

children.

4.2. Experiment 2: The Comprehension of Quantifiers.

.2.1. Subjects

children with SLI who participated in Experiment 1 also

4

The same 16

participated in Experiment 2. They were matched to three groups of TD

children, different from those of Experiment 1, following the same criteria used

for Experiment 1. Group data are represented in Table 6 below.

Subjects

Age Mean

PPVT raw score

PPVT TCGB raw score

TCGB

(SD) Mean (SD)

Z score Mean

Mean (SD)

Z score Mean

SLI (n16) 87.44 (13.23)

74.13 (25.03)

-1.34 13.13 3.64 (6.14)

CA (n16) (11.99)

0.21 4.40 (4.15)

0.10 87.56 108.30 (22.30)

VA (n16) (14.73) (25.57)

-0.29 14.25 -0.02 63.69 75.5 (9.80)

GA (n16) (   (  

-0.49 13.16 0.16 64.87 11.39)

75.19 21.57) (7.23)

able 6: Children group data for experiment 2. T

29

No differences between the SLI group and the TD groups were found according

.2.2 Materials and Procedures

stigated SLI children’s comprehension of

intra

cars are in the boxes) .

to a number of one-way ANOVAs: no age difference between SLI children and

CA children (p=1); no difference in raw scores at the PVVT test between SLI

children and VA children (p=.8884); no difference in raw scores at the TCGB

test between SLI children and GA children (p=1).

4

In experiment 2 we inve

nsitive sentences containing a quantified subject. We made use of a

shortened version of Katsos’ et al. (2011) Cavegirl and Boxes design. In this

task children were told that the Cavegirl, a fictional character, was learning

Italian and they were asked to help her when she made errors. While children

saw the picture of the Cavegirl and the picture of the boxes containing toys on

the screen of a PC administrating the experiment, they heard the Cavegirl saying

how many toys were in the boxes; after that, they were invited to say whether

the Cavegirl said right of wrong. Sentences uttered by the Cavegirl were

digitally recorded by a female native speaker of Italian and played through

loudspeakers connected to the PC. An example of a trial is given below in

Figure 4. We tested the comprehension of the universal quantifier “tutti” (all)

and the existential quantifier “qualche” (some). Sentences containing the

quantifier “tutti” were logically true or false with respect to a given situation as

represented in Table 7; sentences containing the quantifier “qualche” were

logically true or false with respect to a given situation and also logically true but

pragmatically under-informative with respect to a given situation (i.e. the

sentence “some cars are in the boxes” was uttered in a situation where all cars

were in the boxes). We tested 3 items for each condition as represented in the

table below for a total of fifteen items. Items were randomly ordered and

preceded by five familiarization items including sentences with numerals (three

30

Situation1 Situation2 Situation3

ALL

True False

/

SOME

true False

under-informative

Table 7: experimental conditions for the quantifier comprehension task.

Qualche mela è nelle scatole Some apple

e 4. Example trial. Under informative Design from Katsos et al. (2011)

if I said RIGHT or WRONG

is in-the boxes

accuracy for quantifier

w.

Hi, I am Susan and I am learning Italian. Would you help me? I will say how many things are in the boxes and you will judge

Figur use of SOME. .

.2.3. Results of Experiment 2 4

Raw percentages of SLI and TD children’s response

and Table 8 belotypes are represented in Figure 5

31

OVERALL SOME TRUE

SOME FALSE

SOME UNDER-INF

ALL TRUE

ALL FALSE

Mean Mean Mean Mean Mean Mean

(SD)

(SD) (SD) (SD) (SD) (SD)

SLI 87.91

9) (19) (37) (9) 13)

98.00

(14) 91.66 (1

92.03 64.58 98.00 94.32 (

CA 93.33

(12) 91.24 (34)

98.00 (8)

85.42 (34)

100 (0)

98.66 (8)

VA 91.25

(13) 91.08 (30)

100 (0)

77.08 (39)

100 (0)

98.00 (6)

GA 90.41

(14) 88.00 (32) (8)

77.08 (42)

100 (0)

98.57 (8)

Table 8: Accuracy raw percentages ntifier hensio

FIGURE 5: ACCURACY IN QUANTIFIER COMPREHENSION 

in qua compre n 32

As we can observe from Figure 5, children with SLI and TD children were at

lse sentences,

OMEtrue sentences and SOMEfalse sentences. Contrary to CA, children with

SL

ups differed in the comprehension of SOMEunder-inf only.

Th

ceiling in the comprehension of ALLtrue sentences, ALLfa

S

I had problems in rejecting SOMEunder-inf sentences. In our analyses,

analogously to the analyses of clitic production data, group differences in the

comprehension of quantified sentences were analysed by means of different

pairwise comparisons between SLI children and the three control groups. Also

in this case, for each comparison we carried out repeated measure logistic

regression analyses in a mixed model, with Group (SLI vs TD) and Condition

(SOMEtrue vs SOMEfalse vs SOMEunder-inf vs ALLtrue vs ALLfalse) as

independent fixed factors and response accuracy as dependent variable; in these

analyses, we also used a backwards elimination procedure to compare the

goodness of fit of the models. All analyses were conducted using R (version

2.13.1); for mixed modeling we used the LanguageR package (version 1.4;

Baayen, 2011).

In the CA comparison, we found a Group effect, an effect of Condition and

an interaction between Group and Condition. The interaction was due to the fact

that the two gro

us, the difference between groups is due to just one condition:

underinformative SOME. Estimated coefficients, their standard errors, Z-values

and associated p-values for the Group and Condition fixed factors are

summarized in the Table 9 below.

33

CA Estimate SE Z P

(Intercept) 1.4241 0.8041 1.771 0.076 Group 2.5079 1.3131 1.910 0.056 SOMEunder-inf vs SOMEtrue 0.7 3.559 < 0.001 SOMEunder-inf vs SOMEfalse 0.7 3.559 < 0.001

r-infor vs ALLtrue <r-inform vs ALLfalse <

-2.6344 1.0729 -

orm orm

2.6344 2.6343

401401

SOMEundende

m 4.2618 1.2021 3.545 0.001 SOMEu 3.1656 0.9176 3.450 0.001 Group*Condition

2.455

0.014

s Table 9: estimated coefficient for an itio th

Since it has been shown that Italian children reach adult like performance in

the comprehension of under-informative some statements at around 7 years

(G

with

Group d Cond n in e CA

comparison

No effect of Group and Condition was found in the VA and GA comparisons.

uasti et al., 2005), we divided children with SLI in two groups, one group

age minor than 7;6 years and one with age greater than 7;6 years and we

analyzed accuracy in the comprehension of under-informative some statements

in these two groups. Percentages of accuracy by age groups is represented in

Table 10 and in Figure 6 below.

34

Table 10

Accuracy raw percentages of SOMEunder-inf in SLI and CA

A Age > 7,6

SD

:

ge < 7,6

Mean SD Mean

SLI 54.55 850.56 6.67 35.19

CA 77.77 42.37 95.23 21.82

Figure 6: Accuracy raw percentages of SOMEunder-inf in SLI and CA per AGE_group

35

In our analyses, we found an effect of Age_group on the comprehension of

SOMEunder-inf sentences in children with SLI, but not in the CA group. As we

can observe from Table 10, while younger SLI children have problems in the

comprehension of SOMEunder-inf sentences, older SLI children are on a peer

with both older and younger TD children. Estimated coefficients, their standard

errors, Z-values and associated p-values for the Age-group effect in SLI are

summarized in the Table 11 below.

Age_group in SLI Estimate SE Z P

(Intercept) -0.3054 0.3522 -0.867 0.385 Age_group 2.1771 0.8372 2.600 0.009

Table 11: estimated coefficients for Age_group in SLI children.

ary of Experiment 2

e problems with

e semantic component associated to the meaning of the quantifiers ALL and

weak in computing and evaluating scalar

impl

4.2.4. Summ

Our data show that school age children with SLI do not hav

th

SOME and that they are mildly

icatures triggered by the quantifier “qualche” (some). However, our data

show that this weakness is only observable when we compare school age SLI

children, as a group, with CA matched children. Children with SLI obtain score

comparable to those obtained by 5 years old children. Moreover, when we split

the children with SLI in two groups based on age, the difference with the CA

control group remains evident only in the younger children. These data suggest

that in our SLI group the semantic-pragmatic competence is only mildly

affected. Interestingly, the ability to compute scalar implicatures is only delayed

and children with SLI catch up their peer by about 8 years. These conclusion

36

sharply contrasts with the findings concerning clitic production. Even with age,

we did not observed any noticeable improvement in our group.

5. Discussion

Our study was motivated by two goals (1) extend our understanding of

s to school age Italian speaking children with SLI; (2) investigate

diffe

ith respect to younger ones. Contrary to

youn

clinical marker

rent language abilities in these children useful for defining the origin of the

deficit and for differential diagnosis. With respect to the first goal, our data

confirm that the failure to produce DO clitics is still a good clinical marker of

SLI after age 5 years in Italian children. This means that clitic production is a

persistent challenge for children with SLI during school years. By contrast,

comprehension of quantified sentences and generation of scalar implicatures is

only mildly affected and children with SLI catch up their peers at 8 years,

shortly after the same ability is achieved by CA matched children. Since

children with dyslexia are referred to clinical services, often for the first time,

during the school years and often they have also language problems, our results

are good news for them. It is possible to establish whether these children are

also affected by SLI by examining their production of clitics. Of course, this

conjecture needs further investigation.

We observed that the failure to produce clitics takes on a different

manifestation than in older children w

ger children with SLI, who tend not to produce the internal argument of the

verb (Bortolini, 2006), older children with SLI tend to produce a corresponding

full DP instead of the DO clitic. One could object that the infelicitous production

of the object full DP is an artifact of our experimental design in which the

repetition of the full object DP in the lead-in could trigger the production of the

same DP instead of the DO clitic (see Figure 1). That this is not the case is

indicated by the fact that children correctly produced null subject in their

answers, as required by restrictions on the use of pronouns in discourse, though 37

a full subject DP is repeated in the lead-in. It would be therefore difficult to

argue that the production of the full DP object in the lead-in triggers the

production of the full DP in the response but the production of the subject full

DP does not. The specific structure of children response might be rather due to

the fact that older school age children with SLI know that the internal argument

of a transitive verb should be overtly realized in Italian, but they cannot do so

consistently with a DO clitic.

With respect to our second goal, namely that of investigating different

language abilities in children with SLI with the purpose of finding the locus of

the

ed on their prosodic properties, as we discussed

earli

deficit and for differential diagnosis, we can conclude that our group of

children with SLI has a deficit circumscribed to the area of syntax and

morphosyntax, with the exclusion of the semantic and pragmatic area. This

conclusion is based on a series of analysis that we performed and on the task

administered in Experiment 2.

Bortolini and colleagues argued that the failure to overtly realize DO clitics

in Italian SLI children depend

er, although the admitted that this could not be the sole cause. Our findings

further prove and sharpen this point. The production of reflexive clitics reveals

that no differences exists between children with SLI and TD children (either in

the age and language matched comparisons): SLI children produced a high rate

of target responses containing a reflexive clitic (84.70%). This data suggest that

the prosodic characteristic of DO clitics cannot be the cause, let alone the only

one, of SLI children’s failure to produce them, since DO clitics and reflexive

clitics share the same prosodic features. At most, the prosodic component can

add up to other challenging aspects of DO clitic production, an hypothesis that

our data cannot exclude. As we argued in our initial discussion, pragmatic

competence is also relevant for the production of clitics, since they denote a

discourse topic. Given that SLI children produced a corresponding full DP

instead of a DO clitic, we could conjecture that these children lack the pragmatic 38

competence for the use of pronouns. Two facts suggest that this hypothesis is

not correct. First, in our experiment, pragmatic competence for the use of

pronouns requires the production of a sentence with a null subject as a felicitous

answer to the question eliciting the production of clitics. Eighty-seven percent of

SLI children’s responses were sentences with a null subject, suggesting they do

not lack pragmatic competence for the use of pronouns. Second, if we assume

that scalar implicatures are derived by means of a pragmatic reasoning on the

speaker intentions, our data concerning the comprehension of under-informative

SOME statements further suggest that the SLI children’s failure to produced DO

clitics is not dependent on a general pragmatic weakness. In fact, we found that

children with SLI, as a group, were different from chronological age matched

children but not from language matched TD children in the comprehension of

under-informative SOME statements (while such a difference was found in the

case of clitics). Moreover, the ability to reject under informative SOME

statements was in place in older children with SLI, as these were better in the

comprehension of under informative SOME statements than younger ones and

are like their peers. This was not so for clitics, as no age difference was found

and clitics were equally problematic for all of our children with SLI. This

suggest that the failure to reject under-informative SOME statements is mild,

likely derivative on a general language weakness. and also not a good candidate

for distinguishing SLI children from TD children.

On the basis of our findings, we hypothesize that a major source of

difficulty for the production of DO clitics lies in their syntactic and

morp

alian

cano

hosyntactic complexity, likely coupled with the prosodic complexity.

Unlike full DPs, DO clitics are realized in an uncanonical preverbal

position that is associated to a non local thematic assignment (in It

nical word order is SVO but clitics in finite sentences occur in SOV

structures). According to Belletti’s (1999) analysis of clitic placement in Italian,

third person accusative clitics are D-heads of an impoverished DP containing 39

solely the clitic and are inserted in the derivation of the sentence in the internal

verb argument position with a strong accusative features that needs to be

checked overtly, in the syntax, in a dedicated projection called AgrOP before LF

(Chomsky, 1993). These features force movement of the DP including the clitic

and of the clitic itself, as is sketched in Figure 7. The clitic first moves as a DP

to the specifier of the AgrPartP, as it is shown by the fact that the participle of a

present perfect sentence should agree with the clitic in Italian. Next, it moves as

a head to the head of AgrOP, were it cliticizes with the verb. Finally, it moves

together with the verb to the head of AgrSP, where the verb checks agreement

with the subject.

(15) (pro) la prende

(he) it_CL catches

He is catching it

40

Figure 7. Clitic movement in Italian (Belletti, 1999).

According to Chomsky (1993), a lexical DP is instead inserted in the verb

argument position and checks its accusative feature covertly at LF and not

overtly in the syntax. This means that a lexical DP must not move overtly in the

AgrO projection. If we follow Belletti’s (1999) analysis, the morphosyntactic

operations required for overt case checking in clitics’ production are therefore

more complex than case checking of a full lexical DP since the first has to move

overtly in the syntax, as we illustrated above, while the second must not. As a

result, SLI children do not seem to be able to accomplish the complex overt

movement of the DO clitic into AgrO and prefer to produce a full lexical DP.

proj

+NOM AgrS

VP

+ACC AgrO

(TP)

AgrSP

AgrS‘ spec

AgrOP

V | tv

AgrO‘ spec

(AuxP)

(AgrPartP)

V‘

t3

DP2

D‘ t2

k

tk

la3 - prendev her_CL - catches

tj

41

We argue that SLI children have no problem in the production of RE clitics

since reflexive clitics, contrary to DO clitics, are not realized in a uncanonical

syntactic position associated to non local thematic assignments. It has been

suggested that reflexive constructions are derived by a lexical operation

(reflexivization) on the verb that affects the internal theta-role, linking it to the

external theta-role and thereby blocking its mapping onto object position

(Chierchia, 2004). In other words, RE clitics by virtue of a lexical process are

coindexed with the subject of the sentence and thus sentences containing them

do not display any uncanonical order. It is also likely that reflexive clitics are

directly inserted in the pre-verbal position and thus do not feature movement,

unlike DO clitics. Thus, challenging aspects in the production of DO clitics lie at

the syntactic level (i.e., movement and (un)canonical order), but also at

morphosyntactic level. In fact, 3rd person singular DO clitics in Italian are

marked for person, number and gender, while 3rd person RE clitics are only

marked for person. This additional factor of complexity might add up and make

the production of DO clitics difficult for Italian children with SLI. Our data are

compatible with this explanation, but further comparative investigation of the

production of the other morphologically different DO clitics and reflexive clitics

in Italian children with SLI is needed to substantiate it.

Thus, our findings suggest that the avoidance of DO clitics cannot be

grounded on their lack of prosodic saliency, since children with SLI produce RE

clitics which share the same prosodic features of DO clitics; however, it is

possible that prosodic weakness is challenging when coupled with other

difficulties, like the syntactic and morphosyntactic difficulties exhibited by DO

clitics. Under this view, vulnerable aspects at different linguistic levels add up to

make the production of specific linguistic objects particularly challenging for

children with SLI. This view is compatible with theories that see SLI as a deficit

in the processing of complex aspects of linguistic relations (van der Lely, 1998,

2005). 42

Our results have further implications for educational purposes, since pronouns

serve to maintain discourse coherence. Failure to produce clitics means that

children with SLI will have problems with aspects of literacy development and

likely will not develop a written language similar to that of typically developing

children, something that needs further investigation.

6. Conclusions

Our study confirms that the failure to produce DO clitics is a good clinical

marker of SLI after age 5 years in Italian children but the failure takes on a

different manifestation than in younger children. While younger children with

SLI omit the internal argument of the verb producing an ungrammatical

sentence, older SLI children infelicitously realize it as a full DP. RE clitics are

unproblematic for school age children with SLI. This discrepancy between two

types of clitics suggests that DO clitic omission cannot only be grounded in the

prosodic component but might depend on the complexity of morphosyntactic

and syntactic operations that are required for their production. Further

investigation of the production of differently marked clitics to especially

concerning person contrasts is needed to tease apart the contribution of these

factors. In contrast, to clitics, pragmatic understanding is not particularly

affected in our children with SLI. In fact no problems was observed in the

comprehension of sentences containing the quantifiers ALL and SOME when

felicitously uttered in our experimental context. The fact that children with SLI

are different from age matched children but not from language matched children

in the comprehension of under informative SOME statements suggests that they

have also some mild problems in the pragmatic component of language, but

these could be dependent on some more general problems with language,

especially in the light of the fact that we observed a later development of this

ability in Italian school aged children with SLI.

43

REFERENCES

Audollent, C. and Tuller, L. (2003) “ La dysphasie: quelles séquelles en français? ” ANAE 74-75: 264-270.

Aguilar-Mediavilla, E., Sanz-Torrent, M. & Serra-Raventos, M. (2002). A comparative study of the phonology of pre-school children with specific language impairment (SLI), language delay (LD) and normal acquisition. Clinical Linguistics & Phonetics, 16 , 573-596.

Bedore, L.M. & Leonard, L.B. (2001). Grammatical morphology deficits in Spanish-speaking children with specific language impairment. Journal of Speech, Language, and Hearing Research, 44, 905-924.

Baayen, R.H. (2008). Analyzing Linguistic Data. A Practical Introduction to Statistics Using R. Cambridge University Press

Baayen, R.H. (2011). languageR: Data sets and functions with "Analyzing Linguistic Data: A practical introduction to statistics". http://cran.r-project.org/web/packages/languageR/.

Belacchi, C., Scalisi, T.G., Cannoni, E. & Cornoldi, C. (2008). Manuale CPM. Coloured Progressive Matrices. Standardizzazione italiana. Firenze: Giunti O.S. Organizzazioni Speciali.

Belletti, A. (1999). Italian/Romance clitics: Structure and derivation. in H. van Riemsdijk (eds) Clitics in the Language of Europe. Mounton de Gruyter, Berlin. 543-579.

Bishop, D.V.M. (2000). Pragmatic language impairment: A correlate of SLI, a distinct subgroup, or part of the autistic contiuum? In Bishop, D.V.M & Leonard, L.B. (Eds). Speech and language impairments in children: causes, characteristics, intervention and outcome (pp 99-113). Hove: Psychology Press.

Bishop, D.V.M. (1998). Development of the children’s communication checklist (CCC): a method for assessing qualitative aspects of communicative impairment in children. Journal of Child Psychology and Psychiatry, 39, 879-892.

Bishop, D.V.M.., Chan, J., Hartley, J., Adams, C. & Weir, F. (2000). Conversational responsiveness in specific language impairment: Evidence of disproportionate pragmatic difficulties in a subset of children. Development and Psychopathology 12, 177-199.

44

Bishop, D.V.M., & Snowling, M.J. (2004). Developmental dyslexia and specific language impairment: Same or different? Psychological Bulletin, 130, 858-886.

Bortolini, U., Arfé, B., Caselli, M. C., Degasperi, L., Deevy, P. & Leonard, L.B. (2006). Clinical marker for specific language impairment in Italian: the contribution of clitics and non-word repetition. International Journal of Language and Communication Disorders, 41(6), 695-712

Bortolini, U., Caselli, M.C., Deevy, P. & Leonard, L.B, (2002). Specific language impairment in Italian: first steps in the search of a clinical marker. International Journal of Language and Communication Disorders, 37, 77–93.

Bortolini, U. & Leonard, L.B. (1996). Phonology and grammatical morphology in specific language impairment: accounting for individual variation in English and Italian. Applied Psycholinguistics, 17, 85–104.

Bortolini, U., Leonard, L.B., Caselli, M. C. (1998). Specific language impairment in Italian and English: Evaluating alternative accounts of grammatical deficits. Language and Cognitive Processes, 13 (1), 1-20.

Bottari, P., Cipriani, P. & Chilosi, A. (1994). Dissociations in the acquisition of clitic pronouns by dysphasic children: A case study from Italian. Working paper, Pisa: Scientific Institute Stella Maris.

Bottari, P., Cipriani P., Chilosi, A.M. & Pfanner, L. (2001). The Italian determiner system in normal acquisition, specific language impairment, and childhood aphasia. Brain and language, 77(3), 283-93.

Botting, N. & Adams, C. (2005). Semantic and inferencing abilities in children with communication disorders. International Journal of Language and Communication Disorders, 40(1), 49-66.

Burzio, L. (1986) Italian Syntax: A Government-Binding Approach, Reidel, Dordrecht

Cantiani, C. Lorusso, M.L., Perego, P. Molteni, M., Guasti, M.T. (in press) ERPs reveal anomalous morphosyntactic processing in developmental dyslexia. Applied Psycholinguistics. Caprin, C., & Guasti, M.T. (2009). The acquisition of morphosyntax in Italian: A cross-sectional study. Applied psycholinguistics, 30(1), 23-52.

Catts, H. W. (1993). The relationship between speech– language impairments and reading disabilities. Journal of Speech and Hearing Research, 36, 948–958.

45

Catts, H. W., Adlof, S. M., Hogan, T. P., & Weismer, S. E. (2005). Are specific language impairment and dyslexia distinct disorders? Journal of Speech, Language, and Hearing Research, 48, 1378-1396.

Chilosi, A.M., Cipriani, P., Giorgi, A., Fazzi, B. & Pfanner, L. (1995). Test di Comprensione grammaticale per i bambini. Pisa: Edizioni del Cerro.

Chomsky, N. (1982). Lectures on Government and Binding: The Pisa Lectures. Holland: Foris Publications. Reprint. 7th Edition. Berlin and New York: Mouton de Gruyter, 1993.

Chomsky, N. (1993). A minimalist program for linguistic theory. MIT occasional papers in linguistics no. 1. Cambridge, MA: Distributed by MIT Working Papers in Linguistics

Cipriani, P., Chilosi, A.M., Bottari, P., Pfanner, L, Poli, P. & Sarno, S. (1991) L‘uso della morfologia grammaticale nella disfasia congenita. Giornale Italiano di Psicologia 18, 765–779.

Clegg, J., Hollis, C., Mawhood, L., & Rutter, M. (2005). Developmental language disorders – a follow-up in later adult life. Cognitive, language and psychosocial outcomes. Journal of Child Psychology and Psychiatry, 46 (2), 128-149.

Chierchia, G. (2004). A semantics for unaccusatives and its syntactic consequences. In The unaccusativity puzzle: explorations of the syntax-lexicon interface, Artemis Alexiadou, Elena Anagnostopoulou, and Martin Everaert (eds.), 22-59. Oxford: Oxford University Press.

Conti-Ramsden, G., & Botting, N. (1999). Classification of children with specific language impairment. Journal of Speech, Language and Hearing Research, 42, 1195–1204.

Conti-Ramsden,G., Crutchlet,A., & Botting, N. (1997). The extent to which psychometric tests differentiate subgroups of children with SLI». Journal of Speech, Language and Hearing Research, 40, 765-777.

Cummings, L. (2008). Clinical linguistics. Edinburgh: Edinburgh University Press.

Dockrell, J.E., Messer, D., and Murphy, V. (2005) “Language profiles and naming in children with word finding difficulties,” presentation at the 10th IASCL, Freie Universitat Berlin.

46

Dunn, L.M. & Dunn, L.M. (2000). Peabody - Test di Vocabolario Recettivo - P.P.V.T.-R. Peabody Picture Vocabulary Test – Revised. Adattamento italiano e standardizzazione a cura di: Giacomo Stella - Claudia Pizzioli - Patrizio E. Tressoldi. Omega Edizioni.

Flax, J. , Realpe-Bonilla, T., Hirsch, L.S., Brzustowicz, L.B., Bartlett, C.W., Tallal, P. (2003). Specific language impairment in families: Evidence for co-occurence with reading impairments. Journal of Speech, Language, and Hearing Research, 46: 530-543.

Fletcher, P. & Peters, J. (1984). Characterizing language impairment in children: an exploratory study. Language Testing 1, 33-49 .

Foppolo, F., Guasti, M.T., Chierchia, G. (in press). Scalar Implicatures in Child Language: Give Children a Chance. Language, Learning and Development.

Friedmann, N., & Novogrodsky, R. (2004). The acquisition of relative clause comprehension in Hebrew: A study of SLI and normal development. Journal of Child Language, 31, 661–681.

Friedmann, N., & Novogrodsky, R. (2008). Subtypes of SLI: SySLI, PhoSLI, LeSLI, and PraSLI. In A. Gavarró, & M. João Freitas (Eds.), Language acquisition and development. Newcastle UK: Cambridge Scholars Press/CSP. pp. 205-217.

Gavin, W. J., Klee, T., & Membrino, I. (1993). Differentiating specific language impairment from normal language development using grammatical analysis. Clinical Linguistics and Phonetics 7: 191-206.

Grice, H.P. (1975). Logic and Conversation. In: Syntax and Semantics, Vol. 3, Speech Acts, ed. by Peter Cole and Jerry L. Morgan.New York: Academic Press 1975, 41–58.

Grice, H.P. (1989) Studies in the Way of Words. Cambridge MA: Harvard University Press.

Guasti, M. T. (1993/1994). Verb syntax in Italian child grammar: Finite and nonfinite verbs. Language Acquisition, 3, 1–40.

Guasti, M.T., Chierchia, G., Crain, S., Foppolo, F., Gualmini, A. & Meroni, L. (2005). Why children and adults sometimes (but not always) compute implicatures. Language and Cognitive Processes, 20(5), 667-696.

Horn, L.R. (1972). On the semantic properties of logical operators in English. Los Angeles, CA: University of California doctoral dissertation.

47

Kerschensteiner, M., Huber, W. (1975). Grammatical impairment in developmental aphasia. Cortex. 11(3): 264-82.

Katsos, N., Roqueta, A.C., Estevan, R. A. C. & Cummins, C. (2011). Are children with Specific Language Impairment competent with the pragmatics and logic of quantification? Cognition, 119: 43-57

Jakubowicz, C., Nash, L., Rigaut, C., & Gerard, C. (1998). Determiners and clitic pronouns in French-speaking children with SLI. Language Acquisition, 1998, 7, 2-4, 7(2-4), 113-160.

Leonard, L. B. (1998). Children with specific language impairment. Cambridge, Mass: The MIT Press

Leonard, L.B. & Bortolini, U. (1998). Grammatical morphology and the role of weak syllables in the speech of Italian-speaking children with specific language impairment. Journal of Speech, language, and Hearing Research, 41, 1363–1374.

Leonard, L.B., Bortolini, U., Caselli, M. C., McGregor, K. & Sabbadini, L. (1992). Morphological deficits in children with specific language impairment: the status of features in the underlying grammar. Language Acquisition, 2, 151–179.

Leonard, L.B., Bortolini, U., Caselli, M. C., & Sabbadini, L. (1993). The use of article by Italian-speaking children with specific language impairment. Clinical Linguistics and Phonetics, 7(1), 19-27.

Leonard, L.B., Sabbadini, L., Leonard, J. & Volterra, V. (1987). Specific language impairment in children: a crosslinguistic study. Brain and Language, 32, 233–252.

Leonard, L.B., Sabbadini, L., Volterra, V. & Leonard, J. (1988). Some influences on the grammar of English- and Italian-speaking children with specific language impairment. Applied Psycholinguistics, 9, 39–57.

Leonini, C. (2006). The acquisition of object clitics and definite articles: Evidence from Italian as L2 and L1. Ph.D Dissertation. Università degli Studi di Firenze.

McArthur, G. M., Hogben, J. H., Edwards, V. T., Heath, S. M., & Mengler, E. D. (2000). On the "specifics" of specific reading disability and specific language impairment. The Journal of Child Psychology and Psychiatry 41, 869-874.

48

McGregor, K.K., Newman, R.M., Reilly, R.M., & Capone, N. (2002). Semantic representation and naming in children with specific language impairment. Journal of Speech, Language, and Hearing Research, 45, 998-1014.

Moscati, V. and Tedeschi, R. (2009). ‘The Delay of Italian Past Participle Agreement.’ In BUCLD 33: Proceedings of the 33rd annual Boston University Conference on Language Development. Cascadilla Press.

Paradis, J., Crago, M., & Genesee, F. (2003). Object clitics as a clinical marker of SLI in french: Evidence from French-English bilingual children. Proceedings of the Annual Boston University Conference on Language Development, 2003, 27, 2, 27(2), 638-649.

Plante E. & Vance R. (1994). Selection of preschool language tests: A data-based approach. Language, Speech, and Hearing Services in Schools. 25:15–24.

Pozzan, L. (2006). Clitics and Determiners in a group of Italian SLI children. Paper presented at the 8th CUNY-SUNY-NYU Mini-conference, SUNY–Stony Brook.

Rapin, I., & Allen, D. A. (1983). Developmental language disorders: Nosological considerations. In U. Kirk (Ed.), Neuropsychology of language, reading and spelling. NewYork: Academic Press. pp 155–184.

Raven, J., Court, J.H. e Raven, J.C. (1998). Raven Manual, Section 1 (General overview) and Section 2 (Coloured Progressive Matrices). Oxford, UK: Oxford Psychologist Press.

Records, N., Tomblin, B., & Freese, P. (1992). The quality of life of young adults with histories of specific language impairment. American Journal of Speech Language Pathology, 2, 44-53

Rice, M.L., & Wexler, K. (1996). Toward tense as a clinical marker of Specific Language Impairment in English-speaking children. Journal of Speech and Hearing Research. 39:1239-1257.

Sabbadini, L., Volterra, V., Leonard, L.B. & Campagnoli, M. G. (1987). Bambini con disturbo specifico del linguaggio: Aspetti morfologici. Giornale di Neuropsichiatria in Età Evolutiva, 7, 213-232.

Schaeffer, J. (2000) The Acquisition of Direct Object Scrambling and Clitic Placement. Amsterdam/Philadelphia: John Benjamins.

Sauerland , U. (2011). The Computation of Scalar Implicatures: Pragmatic, Lexical or Grammatical? Language and Linguistics Compass. 1–14.

49

50

Schaeffer, J. (2003a) Specific Language Impairment as evidence for the division of labor between interface pragmatics and grammar. GLOW 26 Workshop II: Generative approaches to language development. Sweden.

Schaeffer, J. (2003b). Pragmatics and SLI. In: Levy Y. & Schaeffer J. (eds) Language competence across populations: Toward a definition of specific language impairment. Mahwah, NJ: Erlbaum , 135-50.

Snyder, W., Hyams, N. & Crisma, P. (1995). Romance Auxiliary Selection with Reflexive Clitics: Evidence of early knowledge of Unaccusativity. Proceedings of Child Language Research Forum 26, Stanford CSLI.

Stothard, S., Snowling, M., Bishop, D., Chipchase, B., & Kaplan, C. (1998). Language impaired preschoolers: A follow-up into adolescence. Journal of Speech, Language, and Hearing Research, 41, 407-418.

Tomblin, B., Zhang, X, Buckwalter, P, & O’Brien, M. (2003). The stability of primary language disorder: Four years after kindergarten diagnosis. Journal of Speech, Language, and Hearing Research, 46, 1283-1296.

van der Lely, H. K. J. (1996). Language Modularity and Grammatically Specific language impaired children. In M. Aldridge (ed.) Child Language. Multilingual Matters Ltd. Avon: UK. pp 188-201.

van der Lely, H.K.J. (1998). SLI in children: Movement, economy and deficits in the computational syntactic system. Language Acquisition 72, 161-192.

van der Lely, H. K. J. (2005). Domain-specific cognitive systems: Insight from grammatical specific language impairment. Trends in Cognitive Sciences 9, 53-59.

Weiner, P. (1974). A language-delayed child at adolescence., Journal of Speech and Hearing Disorders. 39: 202-212.