Sie sind auf Seite 1von 16

INT J LANG COMMUN DISORD, XXXX 2017,

VOL. 00, NO. 0, 1–16

Research Report
Grammaticality differences between Spanish-speaking children with specific
language impairment and their typically developing peers
Donna Jackson-Maldonado† and Ricardo Maldonado‡
†Universidad Autónoma de Querétaro, Santiago de Querétaro, Mexico
‡Universidad Nacional Autónoma de México, Mexico City, Mexico
(Received February 2016; accepted December 2016)

Abstract
Background: A limited number of studies have analyzed grammaticality in monolingual Spanish-speaking children
with specific language impairment (SLI). Most of the available data are based on bilingual speakers.
Aims: To extend previous studies by doing a more detailed analysis of grammatical types in monolingual Spanish-
speakers with and without SLI.
Methods & Procedures: Forty-nine Spanish-speaking children (18 with SLI, 17 age-matched typically developing
controls, 14 language-matched controls) were recruited from schools in Mexico and observed in a spontaneous
narrative task.
Outcomes & Results: The findings were inconsistent with those of previous studies. Significant differences were
found for article, connector and preposition omissions, and the per cent of ungrammatical utterances. There were
no significant differences found for clitics and verb phrases, though clitic substitutions were frequent. Language-
matched controls did not produce different frequencies of ungrammatical utterances.
Conclusions & Implications: Significant differences were found for three main items: the per cent of ungrammatical
utterances, the omission of articles and the omission of prepositions. Therefore, we propose these components be
taken into consideration when distinguishing typically developing children from children with SLI.

Keywords: grammaticality, specific language impairment (SLI), Spanish.

What this paper add


What is already known on the subject
Spanish-speaking children with SLI do not produce the same types of errors as do English speakers, and bilinguals are
different from monolingual children. In a few studies, several grammatical forms have been identified that distinguish
children with and without SLI in Spanish use of articles, clitics and function words. Most studies are based on
experimental designs.
What this paper adds to existing knowledge
This paper presents a study that uses a larger sample of monolingual Spanish speakers than have other studies. It also
uses a spontaneous narrative elicitation method and includes more diverse grammatical categories and processes than
have others. The findings show that the use of different elicitation methods may have distinct outcomes and, thus,
intervention should also use diverse techniques.

What are the potential or actual clinical implications of this work?


Oftentimes grammatical forms are considered as a group rather than identifying the specific process they undergo.
Language evaluations should determine whether the affected concepts are omitted, substituted or added. It is
important to determine how grammatical forms are used erroneously in order to guide the course of intervention
based on acquisition processes.

Address correspondence to: Donna Jackson-Maldonado, Universidad Autónoma de Querétaro, Santiago de Querétaro, Mexico; e-mail:
djacksonqro@gmail.com
International Journal of Language & Communication Disorders
ISSN 1368-2822 print/ISSN 1460-6984 online  C 2017 Royal College of Speech and Language Therapists

DOI: 10.1111/1460-6984.12312
2 Donna Jackson-Maldonado and Ricardo Maldonado
Introduction has determiners with similar functions, they are not
marked for gender or number as they are in Spanish. As
Most studies that analyze the grammaticality of struc-
a result, determiner substitutions or omissions might
tures of children with specific language impairment
be common in bilingual language-learner acquisition
(SLI) have identified clinical markers that differentiate
of Spanish. Thus, as processes may be different, more
these children from those who are typically developing
monolingual data are needed to describe fully language
(TD). The goal of the grammaticality studies has not
deviance in SLI. Few studies have been performed on
been to provide an in-depth description of grammati-
grammaticality in Spanish speakers, mostly with small
cal processes, but instead to identify some of the initial
middle-class samples of monolingual Spanish speakers.
markers of SLI by discriminant analysis. Most studies
It has been shown that grammatical errors occur
have centered on English. In recent years, studies based
across a select group of linguistic structures and yet
on Spanish and other Romance languages have begun to
seldom has an in-depth analysis been carried out to
be published (Andreu et al. 2012, Bortolini et al. 2006,
determine which forms are omitted or substituted.
Jacobson and Walden 2013, Jakubowicz et al. 1998,
In contrast, this study is descriptive in nature as it
Jakubowicz 2011, Morgan et al. 2009, 2012, Serrat and
identifies the types of ungrammatical structures used by
Aparici 2001, Simon-Cereijido and Gutiérrez-Clellen
children with SLI. Nevertheless, it follows the tenets of
2007). Bortolini et al. (2006) have indicated that clini-
cognitive approaches in which languages are analyzed
cal markers for Italian are not similar to English. These
from a usage-based perspective (Tomasello 2009,
authors posit that data from English are not readily ap-
Langacker 1987). Language use reflects conceptualiza-
plicable to languages with structures that differ from
tion (Langacker 1987, Bates and MacWhinney 1989)
English that have more flexibility in word order, richer
and, thus, the structures already acquired may be more
morphology and in case marking. In order to account
informative, having more language-specific weight and,
for form and function differences, studies should be lan-
as such, could be less cognitively demanding. At the
guage specific. For instance, whereas in English auxil-
same time, grammatical errors are not precisely pre-
iaries are highly vulnerable, in Romance languages it has
sented as clinical markers but rather as manifestations
been shown that articles, clitics, prepositions and con-
of higher cognitive demands that the child may take
nectors are most affected (Aguilar-Mediavilla et al. 2007,
longer to process. Therefore, errors are considered as
Andreu and Sanz-Torrent 2013, Bedore and Leonard
being within the process of acquiring different elements
2001, 2005, Bortolini et al. 2006, Dispaldro
rather than as being simple mistakes. Nevertheless, to be
et al. 2013, Eisenberg & Guo, 2013, Gutiérrez-
consistent with many studies we will use the term ‘error’
Clellen et al. 2006, Jacobson 2012, Jacobson
to describe these processes. By analyzing the errors
and Walden 2013, Leonard and Dispaldro 2013,
themselves, rather than proposing a discriminant anal-
Morgan et al. 2009, 2012, Restrepo and Gutiérrez-
ysis, the relation of the error to its function could aid in
Clellen 2012, Serrat and Aparici 2001, Simon-Cereijido
understanding language-specific processes that are weak
and Gutiérrez-Clellen 2007). The non-generalizability
areas in the development of Spanish. Although this is the
of language-specific indicators of SLI across all languages
basic framework of the study, it is not our goal to address
has been amply discussed by Weismer and Evans (2002)
the implications of different theoretical approaches.
and Weismer and Kaushanskaya (2010).
The goals of this paper focus on issues of gram-
Most grammaticality studies of Spanish-speaking
matical forms and processes of children with SLI: (1)
children are based on bilingual populations with
to describe grammaticality measures that differentiate
two structurally different languages (i.e., English and
Spanish-speaking monolingual children with and with-
Spanish versus Spanish and Catalan). Although these
out SLI; (2) to analyze grammatical forms and errors that
data are relevant, it is well known that bilinguals may
are produced by children with SLI; and (3) to extend
mix both languages or transfer from one language
the information provided by previous studies by using
to the other, and that these processes of mixing and
a larger number of children and eliciting spontaneous
transfer may affect language development and in a way
narrative productions. We ask the following research
not observed in monolinguals (e.g., Bialystok 2005,
questions:
Conboy 2012, Gathercole and Thomas 2009, Genesee
2006, Genesee et al. 2011, Kohnert 2010, Lanza 1997,
Odlin 1989, among others). For example, English has a
reduced number of clitics, whereas Spanish not only has r Which specific grammatical types differentiate
more clitic forms but also they cover a wider range of children with SLI from TD children?
grammatical functions. This difference between English r Are the sensitive grammatical categories identified
and Spanish could provoke clitic omissions in bilingual in other Spanish language studies consistent with
learners of Spanish. In addition, although English the findings of this population?
Grammaticality Spanish SLI 3
r Do children with SLI use the same grammati- (PGU) to investigate grammatical development in 3-
cal processes and grammatical components as TD year-olds (Eisenberg et al. 2012, Guo and Schneider
children? 2016), while Guo and Schneider (2016) counted both
correct and incorrect uses of grammatical forms. Other
To this end, we study monolingual Mexican Spanish studies that analyzed several components of the narra-
speakers. First, we describe how grammaticality is an- tive have also used grammaticality indexes to calculate
alyzed and then summarize studies carried out with the number of correct of utterances (Duinmeijer et al.
Spanish-speaking children. 2012).
A method for counting finite verbs, the finite
verb morphology composite (FVMC), was proposed by
Grammaticality
Bedore and Leonard (1998) for English-speaking chil-
There are different methods for studying grammatical- dren. A similar method identifies the number of correct
ity. First, the notion of grammaticality should not be tense markers (Botting 2002). Data using the FVMC
confused with grammaticality judgment. In some gram- on other samples have not shown consistent findings
maticality studies, the number of correct grammatical (Moyle et al. 2011).
forms is analyzed, while others count the number of un- Spanish grammaticality indexes must be established
grammatical items. Both the terms ‘grammaticality’ and to be language specific and cannot be based on the
‘agrammaticality’ have been used to refer to the study English systems. In this sense, Gutiérrez-Clellen et al.
of grammatical and ungrammatical forms (Eisenberg (2006), Restrepo (1998), and Restrepo and Gutiérrez-
et al. 2012, Gutiérrez-Clellen et al. 2006). Most of these Clellen (2012) have proposed grammaticality indexes
studies have used T- or C-units to analyze grammati- consisting of the average number of grammatical errors
cality rather than clauses. Different methods may yield per T-unit. Unintelligible utterances or cohesion errors
different results. are not included in the count, nor are lexical, phono-
Results may also vary according to the observational logical or articulation errors. Each ungrammatical item
technique used to obtain data of the grammaticality of is given individual credit rather than counting the to-
children with SLI and, thus, should be analyzed accord- tal number of ungrammatical utterances. The authors
ingly: sentence-completion probes, story retell, spon- propose that this index can differentiate children with
taneous language analysis (SLA) and spontaneous nar- SLI and TD. Miller and Iglesias (2012) have proposed
ratives may yield discrepant information. Testing for a grammaticality index (utterance level errors or EU) by
grammaticality in a controlled structure, such as sen- counting the number of errors per utterance. A count of
tence completion, implies obtaining a restricted type of 3 or more errors per utterance is considered an EU.
answer; likewise, a story retell, in which the story and In this study we use a unit to count ungrammatical
its elements have been heard by the child prior to the utterances, similar to the Gutiérrez-Clellen et al. (2006)
child’s re-producing it, may be quite different from spon- model, but with a more stringent criterion for what
taneous samples in which the child will select what to is considered ungrammatical (see below). In order to
say and determine which events or objects/participants understand the types of errors to be analyzed, we will
of the story should be introduced. Given these differ- offer a brief summary of findings for Spanish-speaking
ences in methodologies, the identification of grammati- children that have been published to date.
cal processes may also differ. For example, the omission
of components previously mentioned would not have Review of studies based on monolingual
the same implication as omissions in which they were Spanish-speakers with SLI
not introduced in the current context. Levels of gram-
Articles
maticality have most frequently been computed based
on narrative samples. This method, even in narratives, Substitutions and omissions of articles have been re-
is not homogeneous. Story generation versus story retell ported in elicitation tasks completed by Spanish-
have been found to produce different types of errors as speaking children with SLI between the ages of 3 and
well (Botting 2002, Duinmeijer et al. 2012). 7 (with SLI) (Morgan et al. 2009), in SLA (Anderson
Some studies include semantic irregularities in their and Souto 2005, Restrepo and Gutiérrez-Clellen 2001)
counts, while others do not. For instance, Lee (1974) as well as in a story-retell task (Morgan et al. 2009). Us-
took into account the grammatical and semantic irreg- ing SLA, Anderson and Souto (2005) found that 4- and
ularities in utterances for the developmental sentence 5-year-olds made more omissions than substitutions of
scoring (DSS) procedure. Dunn et al. (1996) computed articles. Andreu and Sanz-Torrent (2013) proposed that
structural errors over the number of utterances. A group only 5% of the utterances produced by the 5- and 6-year-
of researchers have used a measure that instead, com- olds with SLI had determiner omissions. In contrast,
puted the per cent of grammatical (correct) utterances Restrepo and Gutiérrez-Clellen (2001) give evidence
4 Donna Jackson-Maldonado and Ricardo Maldonado
for both article omissions and gender substitutions. In reported more clitic omissions than substitutions in
TD children, article productions may change with age 7-year-olds. In contrast, De la Mora (2004) using a cloze
whereas 3-year-olds show both substitutions and omis- task and SLA found more substitutions for the full-noun
sions, 4-year-olds use more substitutions, and both types phrase than for the clitic, whereas Morgan et al. (2009)
of errors decrease at age 5 (Castilla-Earls and Eriks- reported omissions in a cloze task. In a language closely
Brophy 2012). This indicates that articles are vulnerable related to Spanish, Italian-speaking children with SLI
since they undergo distinct grammatical processes result- between the ages of 3;7 and 5;6 had significantly more
ing in errors (omissions and/or substitutions) and their direct object omissions than TD children in sentence-
correct use changes with age. It is yet to be determined completion probes (Bortolini et al. 2006).
whether omissions or substitutions are most frequent. Particularly with clitics, the data-collection tech-
nique used is critical. Whereas in elicitation tasks and
story retell the referent is active, in SLA and spontaneous
Clitics
narratives the participant selects which referents to talk
Clitics in children with SLI have been of interest for about and, thus, there is less control over whether or
Romance language researchers (Bortolini et al. 2006, not the referent has been mentioned beforehand. The
Grueter and Crago 2012, Köppe 2001, Leonard and difference in task demands could lead children to make
Dispaldro 2013, Paradis et al. 2003) because they may clitic omissions or substitutions because the preceding
refer back to objects or participants previously men- referent may not be clear.
tioned and need to agree with them in person, gender
and number. Furthermore, they are unstressed forms
Prepositions and complex verb forms
that need the phonological support of full words. The
most frequently analyzed types of clitics are the ac- Few publications with Spanish-speaking children have
cusative (lo, la, los, las—masculine, feminine and plural) addressed the use of prepositions and complex verb
and dative (le, les, lo/la—masculine, neutral and femi- forms that include prepositions. In this section we sum-
nine), which are object clitics. But clitics also have differ- marize findings of both uses of prepositions, although
ent functions, though they are not usually differentiated we are not implying that they are related. Most stud-
in most publications. Whereas object clitics must refer ies do not specify the function of the prepositions.
back to the person or object they represent, middle cli- Bedore and Leonard (2005) administered an experi-
tics emphasize or focus on a specific part of the event or mental task to TD-emergent bilinguals and children
the type of action that is being carried out. Object clitics with SLI and found mostly preposition substitutions.
are marked for gender, number and case (accusative, da- In contrast, Castilla-Earls and Eriks-Brophy (2012) and
tive). Thus, in the sentence el niño lo puso en la mesa (‘the Morgan et al. (2009) reported that children mostly omit
boy put it on the table’), the accusative pronoun lo is in- prepositions up until the age of 5, this being determined
flected for gender (masculine) and number (singular) as by a cloze task, story retell and SLA.
it agrees with its masculine, singular antecedent object, Andreu and Sanz-Torrent (2013) studied Catalan
el libro (‘the book’). The middle clitic instead denotes bilinguals and also found that between 8% and 10% of
different types of actions (such as se cayó—‘it fell down the utterances in SLA included preposition omissions.
unexpectedly’ versus cayó—‘it fell down’) (Maldonado Prepositions have also been studied as part of the verb
1988, 1999). argument structure. Andreu et al. (2012) as well as Sanz-
In most child language studies, it has been shown Torrent et al. (2011) found omissions and/or substitu-
that in TD children clitics are acquired by 3 or tions in an experimental task in which children between
4 years of age, though they are not completely mas- 3 and 6 years of age were required to introduce a sen-
tered (Anderson and Souto 2005, De la Mora 2004, tence object. In another study using story retell, Auza
Hernández Pina 1984, Jackson-Maldonado et al. 1998, and Morgan (2013) mostly found omissions in story
Montrul 2004, Silva-Corvalán 1994). Catalan–Spanish retell for three types of multifunctional and polysemic
bilinguals (Bosch and Serra 1997) have been reported to prepositions: a, con and en.
omit these forms. Yet, the type of clitic function is not Contrary to findings in other languages (Rice and
always specified. Wexler 1996, Weismer and Evans 2002), verb errors
Children with SLI have been shown to omit and have been found to be infrequent at early ages for
substitute these forms in SLA between the ages of 4 and Spanish speakers with SLI, and early verb productions do
7 (Andreu and Sanz-Torrent 2013, Castilla-Earls and not reflect optional infinitives. Errors have been found
Eriks-Brophy 2012, Morgan et al. 2009). Andreu and at later stages in specific structures and mostly related to
Sanz-Torrent (2013) found that only 7% of the utter- complex forms. Several studies carried out with Spanish-
ances produced by children with SLI had pronoun omis- speaking children using cloze tasks and story retell have
sions. In another SLA study, Bosch and Serra (1997) shown errors in the use of subjunctive markings and
Grammaticality Spanish SLI 5
other complex verb forms in narratives, cloze tasks and Each group of children was consistent in terms
story retell (Holst 2014, Andreu et al. 2012, Jackson- of socio-economic status, measured by maternal
Maldonado 2012, Morgan et al. 2009, Sanz-Torrent educational level: 51% of the sample had mothers with
et al. 2011). Andreu and Sanz-Torrent (2013) found a high-school education and 49% had mothers with a
that 14% of utterances of 5- and 6-year-olds were com- lower level of education. For obvious reasons, language
plex sentences whereas 30% were simple sentences. controls were of different ages than their peers with SLI.
In summary, the above review of pertinent studies No significant differences were found for children’s age
has shown that Spanish-speaking children with and (t(33) = –.818, p = .42) in the age control groups for
without SLI have produced ungrammatical utterances gender (t(33) = 1.2, p = .25) nor maternal education
in which articles, clitics and prepositions are affected. (t(33) = .151, p = .88) across all groups. Also, the chil-
Verb structures have been studied to a lesser degree. dren’s IQs were similar (t(33) = –.814, p = .421) across
Results may vary according to the type of elicitation groups. Most children with SLI were male (78.5%).
task and the number of participants. There are Language controls were contrasted by MLU-w scores
inconsistencies as to whether grammatical forms are on the narrative task.
more affected by omissions or substitutions. Both entail The SLI criteria for monolingual Spanish-speaking
different processes and could affect the way errors are children have been problematic and inconsistent across
interpreted and/or treated. Additional studies with dif- studies and, especially difficult to determine in Spanish,
ferent elicitation methods and more diverse samples are a language for which language assessment instruments
needed. are limited. Because of these concerns, we were strict in
labelling a child as SLI. We based our criteria on the most
frequently used assessment instruments in recent studies
Methods based on Spanish-speakers (Bedore and Peña 2008, Peña
2007) (see below). Criteria for group inclusion were
Participants
based on Tomblin et al. (1997). The language assess-
A total of 49 children participated in the study. Some ment measures selected were based on Spanish-language
of these children were included in a previous study publications and we established a clear protocol for SLI
that addressed different goals (Jackson-Maldonado identification, which was also based on published stud-
2011). There were 18 participants identified as SLI ies and on extensive discussions with Hispanic experts
(12 male and six female), 17 age-matched TD children in the field.
(14 male and three female) and another 14 TD In order to be included in the SLI group, the child
children, matched for gender, who were added as had to meet one of the first two criteria AND one or
language-matched for the younger SLI group (seven both of the third and fourth criteria, as follows: (1) to
male and seven female). Language matches were be on a speech and language therapist’s caseload be-
included as a contrast taking into consideration that cause they had previously been identified with language
differences in grammaticality among children with a delay/disorder (i.e., clinical judgment); and/or (2) to
large age differences could represent differences of have parental or teacher concern for expressive language
cognitive and neurophysiological levels (Coady and development (Restrepo 1998, Restrepo and Gutiérrez-
Evans 2008, French and O’Brien 2008, Gathercole Clellen 2001), plus, (3) to fall below the 16th percentile
et al. 1999, Royle and Courteau 2013). The children or 1.25 SD below the mean on at least two distinct
ranged in age from 5;1 to 9;3 (mean = 7;3, SD = 1;3) language assessment measures, using an adjusted score
in the SLI group, and from 5;0 to 9;3 (mean = 7;3, based on a Mexican sample (developed for this study
SD = 1;3) in the TD age-matched group. For the with 35 TD children), and (4) to produce less than
language controls of the younger children with SLI, ages 80% of the phonemes on a non-word repetition task
ranged from 3;0 to 4;11 (mean = 3;6, SD = 1;0). Mean correctly (see below). Furthermore, exclusionary criteria
length of utterance-word (MLU-w) was used as the included an IQ below 85 and any known auditory, cog-
language measure to contrast the groups. Although nar- nitive, emotional, neurological or developmental disor-
ratives are typically shorter than a full-language sample, ders. All children had normal hearing based on a hearing
they are considered an appropriate measure for calculat- screening test.
ing MLU-w since only using the longest three utterances
has been shown to be adequate to determine language
Instruments
levels and MLU has been used frequently in narratives
(Hoffman 2009, Jackson-Maldonado and Conboy Language assessment measures were selected from those
2007). Language controls were only contrasted for mostly used in other published studies of Spanish-
grammatical index and not for all grammatical process speaking children. Norms were adjusted based on a
measures. small-scale study with a monolingual Spanish-speaking
6 Donna Jackson-Maldonado and Ricardo Maldonado
population in Mexico, similar to that proposed by Procedure
Morgan et al. (2009).
Children were recruited by multiple means. Since many
An IQ score was obtained based on a non-verbal
children have not been previously identified with a lan-
intelligence measure the Test de Inteligencia No-Verbal
guage disorder in Mexico, research assistants visited pub-
(TONI-II; Brown et al. 2000).
lic and private schools in the metropolitan area of a
The most frequently used test to identify both
central state in Mexico and asked the teachers to iden-
monolingual and bilingual Spanish-speaking children
tify children who had trouble understanding instruc-
with SLI has been the Spanish Clinical Evaluation
tions, answering questions, explaining concepts as well
of Language Fundamentals 4th Edition (SCELF-4)
as those who frequently produced incomplete, unintel-
(Semel et al. 2006). In particular, two subscales are
ligible and/or ungrammatical sentences. We also asked
used: Estructura de Palabras (Word Structure) and
if there were children who were ‘dysphasic’, the fre-
Repetición de Oraciones (Sentence Repetition) (e.g.,
quently used term in Mexico for children with limited
Andreu and Sanz-Torrent 2013, Jacobson and Walden
communicative abilities. These children were tested for
2013, Morgan et al. 2009, 2012, Smyk et al. 2013,
the inclusionary criteria stated above. TD children were
Windsor et al. 2010, among others). The Technical
also recruited from the same schools. Other participants
Manual of the SCELF-4 (Semel et al. 2006) confirms
were recruited through clinics that serve children with
that it is appropriate for monolingual Spanish-speaking
language disorders. All parents signed an informed con-
children. Both scales were used in this study.
sent form and completed a clinical history to determine
To assess receptive and expressive vocabulary we used
group inclusion. Children were tested for hearing and
the Expressive and Receptive One-Word Picture Vo-
all had levels within normal ranges, except for one child
cabulary Tests, Bilingual-Spanish Editions (EOWPVT
who had a hearing loss and, thus, did not participate in
and ROWPVT; Brownell 2000a, 2000b). These tests
this study.
use a forced-choice (for receptive) or picture-naming
Evaluations were carried out in the school, clinic
(for expressive) format and include both nouns and
or the university laboratory and lasted over one or two
verbs.
sessions, depending on the child’s attention span. Pre-
Formal tests were complemented with informal
viously trained research assistants who were psychology
measures in the SLI identification process, as suggested
and/or linguistics students applied the standardized tests
by Simon-Cereijido and Gutiérrez-Clellen (2007) and
according to the test protocols. All test sessions were
Restrepo (1998). We used the per cent of correct
video or audio recorded to permit offline scoring by re-
phonemes criteria on a non-word repetition (NWR)
search assistants who were blind to group membership.
measure because its discriminant strength has been
Transcriptions were done in SALT (Miller and Iglesias
amply proven for Spanish-speaking bilingual children
2012). All transcripts followed strict reliability criteria
(e.g., Aguado et al. 2006, Calderón 2010, Ebert et al.
and were independently transcribed and coded by two or
2008, Girbau and Schwartz 2007, Gutiérrez-Clellen
more trained research assistants both for transcriptions
and Simón-Cereijido 2010, Summers et al. 2010,
as well as grammatical errors. All discrepancies were re-
among others). We developed a NWR task specifically
solved by another specialist. Inter-rater reliabilities were
for a larger research project. Although several studies
90% or above based on this method.
had developed similar tests, none was published at the
time that the data were collected for this study and non-
word lists were not available. Our task was developed
Analysis units and rules for determining errors
based on the Dollaghan and Campbell (1998) model of
minimum wordlikeness, but it also varied in articulatory We completed two levels of analysis: a general grammat-
complexity as suggested by Gathercole and Baddeley icality count index and an error-by-error description.
(1990). It consisted of 24 non-words: four instances Our first level, the grammaticality index, consisted of
for each of the six syllable lengths (see appendix A). In a count based on the number of utterances with one
this paper, the NWR task was used as one of several or more ungrammatical items. The index was the total
criteria for SLI identification. Comparisons of this number of ungrammatical utterances divided by the to-
study’s NWR test with the Aguado et al. (2006) list and tal number of utterances produced by the child. This
with Calderón’s (2010) list applied to a group of adults procedure follows the general proposal by Gutiérrez-
and a group of TD children showed the efficacy of the Clellen et al. (2006), Restrepo and Gutiérrez-Clellen
measure (Jackson-Maldonado et al. 2015, Rodrı́guez (2012) and Restrepo (1998) as well as Miller and Igle-
Hernández and Jackson-Maldonado 2009). sias (2012). One difference was that we counted total
Finally, a spontaneous narrative was elicited using number of utterances with errors rather than computing
the wordless picture story, Frog, Where Are You? (Mayer individual errors. Any utterance, in this case a T-unit,
1969) for the grammaticality count. with a grammatical error was given a count of 1. Thus,
Grammaticality Spanish SLI 7
Table 1. Grammaticality categories by error type

Type Example
Omissions
Article Se quitó EL pantalón y los zapatos (‘He/she took THE off pants and shoes’)
Noun Phrase Y el perro estaba persiguiendo A LA RANA (‘and the dog was chasing THE FROG’)
Clitic LO estaban buscando (‘IT was looking for’)
Prepositions Se asomó POR un árbol (‘peaked THROUGH a tree’)
Verb El perro ESTABA tratando de agarrar un panal de abejas (‘the dog WAS trying to grab a bee hive’)
Substitutions
Article Llevó al niño hasta UN montaña (‘(He) took the boy to a mountain’; masculine for feminine)
Clitic Y no lo encontraba (la rana) (‘and he didn’t find it’; masculine for feminine)
Verb El perro y el niño se DURMIÓ (‘the dog and the boy fell asleep’ (plural missing))
Preposition Gritaba por todo el bosque POR su ranita (‘he shouted throughout the woods for his frog’; por for a- for instead of to)
Additions
Article Lo tiró al UN charco de LA agua (‘he threw him down into the a puddle of the water’)
Preposition Se cayó DE el niño (‘of the boy fell’)
Word Order De noche estaba (‘nightime it was’; also verb substitution)
Unintelligible Y el niños gritaba una y otra donde vivı́a ahı́ (‘and the boys screamed one and again where (he) lived there’)
Note: Omissions are in CAPITALS; substitutions and additions are underlined.

in the utterance, ‘el niño se fue A buscar el A LA ranita’ repetitions because clitics would be expected in that
(‘the boy went looking FOR THE frog’, i.e., where context. Referent-clitic agreement (gender and number)
capitals represent omissions) where there are multiple was always obligatory. To determine clitic appropriate-
errors, we would give a count of 1 for the grammatical- ness, we also incorporated suprasegmental cues such
ity index, and also each ungrammatical type would be as intonation contours and pauses as these clues could
counted separately in the next level of analysis of gram- imply potential barriers to topic maintenance that
matical categories: preposition omissions and article would affect clitic use (Du Bois 2003, Givón 1983).
substitutions. Another important issue is that not all omissions
Our second level of analysis, grammaticality types were considered errors. For example, auxiliary deletion,
and errors, was subdivided into omissions, substitutions el niño# comiendo versus está comiendo ‘eating versus
and additions. This level of errors was then further is eating’ with an intonation shift or a pause after
classified according to grammatical criteria (articles, niño ‘boy’, would be an appropriate predicative form
prepositions, etc.). Thus, there was an error count that does not require an auxiliary and, thus, it was
for general type (all omissions) and for grammatical not counted as ungrammatical. However, cases in
category (article omission, article substitution, etc.), which the child omitted the auxiliary in a sequence
and also a composite by grammatical type that included of events without an intonation shift were counted
all types of errors within each grammatical category as an error (Alarcos-Lorach 1978, Gutiérrez-Ordóñez
(omissions and substitutions of articles together). 1986, Lope-Blanch 1984, Rodriguez-Espineira
Table 1 presents the categories and examples of each 1992).
grammatical error type found in the data. The absence of the auxiliary in the direct command,
We established a criterion for scoring ungram- a comer! (‘Let’s eat’) was not accounted as an error either.
matical forms based on obligatory contexts for both Most Spanish studies treat this as an error allegedly be-
linguistic form and discourse (in which context and cause the auxiliary of the periphrastic future vamos (for
language use played an important part). Further vamos a comer, ‘go to eat’) is missing. A comer is a fixed
we took into consideration information from the expression in Mexican Spanish, not an error. Also, the
images and narrative situation to control for context absence of the plural form -s in the dative le (le di un
interpretation. Most importantly, for clitics, referential regalo a los niños, ‘I gave a gift to the children’, instead
distance, i.e., the number of words between the clitic of les di un regalo a los niños with number agreement) is
and the noun phrase it referred to, were fundamental for well established not only in Mexican Spanish but also
grammaticality decisions. In the same vein, we also took in several dialects of Latin American Spanish and, thus,
into account whether the clitic could refer back to an it was not counted as an omission error. It should be
explicit image in the story. This was crucial because the noted that this accepted omission is not the case for all
visual image could operate as a contextual antecedent dialects of Spanish.
for the clitic even if the child did not actually name To summarize, the forms that were considered un-
it in the narrative. Full phrases, when the referent was grammatical in this study were: (1) agreement er-
previously mentioned, were considered as inappropriate rors in gender number in determiner-noun and/or
8 Donna Jackson-Maldonado and Ricardo Maldonado
Table 2. Descriptive statistics for SLI and language controls

SLI 5–6 N = 8 TDlang1 N = 14 Effect SLI 7–8 N = 11 TDlang2 N = 9 Effect


∗ ∗
%UNGRA 31.9 (4–62) 16.1 (6.7–32) .480 23.7 (38.7–64) 11 (2.2–22.7) .535
%OMISS 18.73 (0–43.2) 8.46 (0–18.18) .379 13.18 (0–50)∗ 3.58 (0–16) .465
%SUBSTIT 7.14 (0–19.05) 4.08 (0–22.22) .232 8.47 (3.3–20) 7.1 (2.86–14) .12
%ADDIT 1.79 (0–14.29) 2.08 (0–10) .247 1.21 (0–13.3) 0 .178
#T UNIT 34.63 (15–69) 50.93 (27–94) n.a. 36.73 (14–58) 34 (19–56) n.a.
Notes: Means (ranges) effect size and significance of per cent error types. ∗p < .04.
UNGRA, ungrammatical; OMISS, omissions; SUBSTIT, substitutions; ADDIT, additions; SLI 5–6 or 7–8, specific language impairment 5–6- or 7–8-year-olds; TDlang1, language
controls for SLI group (5–6-year-olds); TDlang2: language controls for SLI group (7–8-year-olds).

adjective-noun phrase, (2) obligatory contexts where r = .480) and the older group of children had strong
subject–verb agreement was missing (in Spanish verb effects (U = 18, Z = 2.39, p < .016, r = .535)
agreement is obligatory but overt subject mention is op- (table 2).
tional), (3) grammatical verb inflection errors, (4) verb Contrasts for per cent of general error types were
auxiliary omission with no pause or intonation pred- carried out with a Mann–Whitney U-test (table 2). For
icative contour (see above), and (5) clitics that did not the younger language control group, there were no sig-
agree in gender and number with the preceding refer- nificant differences and effect sizes were medium for
ent or with the image depicted in the story. Utterances omissions (U = 30, Z = 1.77, p < .082, r = .379), yet
were considered unintelligible or incomplete if part of negligible for other error types: substitutions (U = 40,
the utterance could not be deciphered or if part of the Z = 1.09, p < .297, r = .232) and additions (U = 42,
utterance was not completed, but did not correspond to Z = 1.15, p < .365, r = .247). Thus, compared with
a specific omission or substitution. language controls at this young age, there are no signifi-
cant differences in omissions, though there is a medium
effect size, and there were no differences for substitu-
Results tions and additions. Language control comparisons for
the older age group were significant for omissions with
We first contrast children with SLI and language- a medium effect size (U = 16, Z = 2.08, p < .04, r =
matched controls based on the grammaticality index .465), and non-significant with small effects for substi-
and general error types (omissions, substitutions and tutions (U = 32.5, Z = .544, p < .587, r = .12) as well
additions) and then present a more in-depth analysis as for additions (U = 35, Z = .79, p < .791, r = .178).
comparing the children with SLI only with age con- Thus, for both age groups, only omissions had medium
trols. The contrasts of children with SLI and language effects and they were significant for the older group.
controls were divided in two language groups based on The next level of analysis contrasted children with
age. The first group consisted of the younger children SLI (N = 18) with TD (N = 17) age-matched con-
with SLI who were between the ages of 5 and 6 and trols. Children with SLI were compared with TD chil-
their language-matched controls (SLI = 8, TD = 14). dren on criterion measures: scores on vocabulary and
The second group included older children with SLI be- grammar tests as well as their scores on the NWR task
tween the ages of 7 and 8 and their language-matched (see Instruments above). Homogeneity of variance across
peers (SLI = 10, TD = 9). Language controls were these tests was established. A t-test for individual sam-
based on MLU-w and were usually 2 years younger than ples yielded significant differences for all expected mea-
the SLI group. Homogeneity of variance was not estab- sures. As a group, children with SLI had significantly
lished and, thus, non-parametric analyses were carried lower scores than the TD group on receptive vocabulary
out. Based on the American Psychological Association’s (ROW) (t(33) = 2.79, p = .009), expressive vocabulary
(APA) 5th Manual (2001) and proposals by Grissom (EOW) (t(33) = 3.42, p = .002), sentence repetition
and Kim (2012) and Leech and Onwuegbuzie (2002), (CELFEP) (t(32) = 4.55, p < .000), word structure
in this study we report both effect sizes and p-values (CELFRO) (t(33) = 3.38, p = .002) and per cent of
given the sample size. Effect sizes were calculated for r correct phonemes produced during the non-word repe-
as: small ࣙ .10, medium ࣙ .30 and large ࣙ .50 based tition task (PPC) (t(32) = 5.81, p < .000) (table 3). In
on Cohen (1988). this case, Cohen’s d effects were calculated (small ࣙ .20,
In the first analysis we compared language controls medium ࣙ .50, large ࣙ .80). All had high effect sizes at
with children with SLI on grammaticality index and er- or above (d = .98).
ror type. Non-parametric analyses were significant for A test for homogeneity of variance showed that dis-
the grammaticality index, while the younger group had tributions of grammaticality measures were not normal
medium effect sizes (U = 23, Z = 2.25, p < .024, and, thus, only non-parametric analyses were carried out
Grammaticality Spanish SLI 9
Table 3. Results of a t-test for language, non-word (per cent p < .612, r = .264). Thus, as with the language controls,
phoneme correct) and IQ tests only general omissions were different between groups.
SLI (SD) TD (SD) Effect Cohen’s d As it can also be seen in table 4, a composite of
ROW 93 (15.8) ∗∗∗
105 (10) 0.908
grammatical types that summed errors (total articles,
EOW 98 (19.9)∗∗ 117 (16.7) 1.03 total clitics, etc.) had significant differences and strong
SCELFEP 80 (25.9)∗∗ 113 (13.9) 1.59 effect sizes for total article errors (U = 70, Z = 2.82,
SCELFRO 76 (23.1)∗∗∗ 105 (27.9) 1.13 p < .006, r = .564) while preposition and verb phrase
MLU-w 4.75 (0.98) 5.21 (0.79) 0.517 errors had medium effects and only prepositions were
PPC 71.4 (11.1)∗∗∗ 88.7 (4.5) 2.04
TONI-II 106.9 (16.3) 111.8 (19.1) 0.276
significant (prepositions U = 91.5, Z = 2.07 p < .04,
r = .418 and verb phrases U = 100.5, Z = 1.73, p <
Note: Mean score and (standard deviations). ∗∗ p < .002, ∗∗∗ p < .0001.
Percentile scores: ROW, Receptive One Word Test; EOW, Expressive One Word
.08, r = .346). Clitic errors only had small effect sizes
Test; SCELFEP, Spanish CELF-4 Estructura de Palabras; SCELFRO, Spanish CELF-4 (U = 112, Z = 1.37, p < .193, r = .267). Unintelligible
Repetición de Oraciones; MLU-w, mean length of utterance in words; PPC, per cent utterances and word order problems were negligible and,
phonemes correct on non-word task; TONI-II, Test of Non-verbal Intelligence, Spanish
Version. IQ Score. thus, not discussed.
A more in-depth analysis is presented in table 5.
Comparisons were not significant for grammatical cate-
gory combined with error type and medium effect sizes
for all grammaticality categories. As above, p-values and were found for several omission categories: noun phrases
effect sizes are provided to show significance. Table 4 ex- (U = 115, Z = 1.59, p < .230, r = .368), verb phrases
hibits means and ranges for the general variables studied (U = 111, Z = 1.6, p < .182, r = .315), articles (U =
and for each group of children. Differences in gram- 110, Z = 1.8, p < .172, r = .362), prepositions (U = 99,
maticality index scores between children with SLI and Z = 1.84, p < .082, r = .368), clitics (U = 144, Z = .36,
age-matched controls were calculated. We then com- p < .806, r = .320), and frequency of unintelligible or
pared groups by error types and grammatical categories. incomplete utterances (U = 114.5, Z = 1.77, p < .217,
Contrasts between children with SLI and age- r = .354.). Some substitution errors also had medium
matched controls based on a Mann–Whitney U-test effect sizes, but, again, were not significant: articles
yielded significant differences and a large effect size (U = 99.5, Z = 1.93, p < .082, r = .386), clitics (U =
for per cent of ungrammatical utterances (U = 49, 109.5, Z = 1.56, p < .61, r = .312), and verb phrases
Z = 3.412, p <. 001, r = .682). Also, both groups (U = 142, Z = .367, p < .756, r = .312). Curiously,
produced a similar number of utterances (M = 35.84 clitic substitutions were more frequent in TD children
and 33.25) (table 4). compared with children with SLI. No additions were
General omissions, substitutions and additions be- significantly different and effect sizes were negligible.
tween the SLI and TD age-matched groups were further These results of error types illustrate that language
analyzed (table 4). Effects for status (SLI or TD by age) and age-matched controls are distinct and, thus, it would
were considered to determine differences. Significant seem to be that comparing children with SLI with
differences were found and effect sizes were large for to- age-matched controls provides more information about
tal omissions (U = 70.5, Z = 2.74, p < .006, r = .548). grammaticality than contrasting children with SLI to
No significant differences were found and effect sizes language-matched controls. Most errors only had low
were low for total substitutions (U = 128.5, Z = .779, effect sizes when comparing children with SLI with
p < .44, r = .156) and additions (U = 136, Z = 1.32,

Table 4. Descriptive statistics of total per cent error types and total grammatical categories for SLI and age controls

SLI N = 18 SD TD N = 17 SD r
#T UNIT 35.84 (14–69) 15.21 33.25 (19–56) 9.32 .126
%UNGRAM 27.19 (9–64.3) 16.8∗∗ 11.41 (2.2–24) 6.63 .682
%OMISSION 15.52(0–50) 13.72∗∗ 4.9 (0–17.39) 6.7 .548
%SUBSTITUT 7.91 (0–20) 5.87 5.98 (0–13.5) 3.89 .156
%ADDITION 1.45 (0–14.29) 4.36 0 0 .264
%ARTICLES 8.07 (21.62) 7.83∗∗ 1.48 (0–6.9) 2.27 .564
%CLITIC 2.94 (0–14.29) 4.4 4.48 (0–10) 3.81 .267
%PREPOSITION 5.96 (0–28.57) 7.33∗ 2.04 (0–13) 3.67 .418
%VERB 4.58 (0–14.49) 4.08 2.24 (0–6.06) 2.33 .346
%UNINTEL 0.199 (0–2.33) 0.612 0.189 (0–3.03) 0.758 .354
%ORDER 0.198 (0–2.33) 0.613 0.168 (0–2.7) 0.675 .075
Notes: Means (ranges), standard deviation, effect size and significance. ∗ p < .04, ∗∗ p < .006.
SLI, primary language impairment; TD, typical development age control; SD, standard deviation.
10 Donna Jackson-Maldonado and Ricardo Maldonado
Table 5. Descriptive statistics of per cent error types by grammatical categories for SLI and age controls

SLI N = 18 SD TD N = 17 SD Effect
Omissions
%ARTICLE 3.71 (0–16.22) 5.77 0.369 (0–4) 1.07 .362
%CLITIC 1.36 (0–14.29) 2.95 1.24 (0–8.7) 2.83 .07
%PREPOSITION 5.68 (0–28.57) 7.37 1.78 (0–8.7) 2.87 .367
%NOUNPHRASE 1.9 (0–8.11) 2.95 0.62 (0–5.36) 1.70 .31
%VERB 2.86 (0–13.64) 4.04 0.9 (0–6.06) 1.82 .32
Substitutions
%ARTICLE 4.19 (0–20) 6.04 1.28 (0–6.9) 2.21 .386
%CLITIC 1.58 (0–9.09) 2.94 3.25 (0–10) 3.55 .312
%PREPOSITION 0.27 (0–2.63) 0.81 0.27 (0–4.35) 1.09 .076
%VERB 1.72 (0–8) 2.41 1.34 (0–5.41) 1.94 .312
Additions
%ARTICLE 0.175 (0–3.3) 0.76 0 .184
%CLITIC 0.175 (0–3.3) 0.76 0.17 (0–2.7) 0.68 .016
%PREPOSITION 1.28 (0–14.29) 3.89 0 .264
Note: Means (ranges), standard deviation and effect size.

language-matched controls (both groups). Although age 2005, Jacobson and Walden 2013, Morgan et al. 2009,
differences were large between children with SLI and 2012, Restrepo and Gutiérrez-Clellen 2001, 2012,
language controls, differences were not significant, al- Simon-Cereijido and Gutiérrez-Clellen 2007).
though there were medium effect sizes for some cate- A finding derived from the results of this study,
gories. Errors in children with SLI are similar to those although not a research question per se, was that there
produced by children who are about two years younger. are differences between language and age-matched
controls. Our study, in line with most research, first
compared children with SLI with language-matched
Discussion controls who were matched based on the MLU-w
measure (as in Botting 2002, and Duinmeijer et al.
The aims of this study were to describe grammatical cat- 2012). Language controls were divided into two groups
egories by omissions, substitutions and additions that (one compared with 5- and 6-year-olds with SLI and
differentiate children with SLI from children who are a second group compared with 7- and 8-year-olds
TD and relate these findings to other studies. First we with SLI). Language controls were, on average, 2 years
analyzed ungrammaticality. Then we carried out a more younger. Because of this large age difference, it could be
detailed analysis by grammatical type. As mentioned expected that there would be differences despite similar
above, observational techniques may affect grammati- language levels. There was a significant difference
cality results and, thus, we will centre our discussion on between the 5–6-year-old group and the language
results that have been obtained by narrative samples and controls for the ungrammaticality index, and there were
SLA in other studies. medium effects for this variable, as well as for per cent
Previous research has mostly been clinically based, of omissions. Similarly, the results for the 7–8-year-old
using discriminant analysis, to propose markers that group were significant for grammaticality, and also for
differentiate children with SLI from those without per cent omissions. In these cases effect sizes were strong
SLI while not discussing the grammatical errors and (if rounded). Thus, data from this study would agree
types themselves (Jacobson 2012, Morgan et al. 2009, with researchers who have proposed that grammatical
Simon-Cereijido and Gutiérrez-Clellen 2007). In con- errors change with age (Castilla-Earls and Eriks-Brophy
trast, this study is descriptive in nature and purposely 2012). Furthermore, it may be more methodologically
designed to illustrate several ungrammatical forms that pertinent to include age-matched controls if the goal is
are produced in a spontaneous narrative task. Most to illustrate grammatical errors. This methodological is-
published studies have not shown consistent evidence sue can be considered since Bedore and Leonard (2001)
for one error type in particular, much less related to did not find differences when comparing children with
grammatical categories. Although most studies concur SLI with MLU-matched controls and may explain why
that the forms that are most affected in SLI are articles, many studies mostly use age- rather than language-
clitics and other function words, such as prepositions, matched controls. The need to compare age rather than
it is not clear whether they all undergo omissions, language controls is further evidenced by the fact that
substitutions or additions (Anderson and Souto 2005, children who are much younger than the experimental
Andreu and Sanz-Torrent 2013, Bedore and Leonard group may be quite different because of maturation.
Grammaticality Spanish SLI 11
Thus, results could be confounding maturation and are mostly omitted and, less frequently, substituted
language ability (Royle and Courteau 2013). (Auza and Morgan 2013). In this sample, children also
The detailed errors and grammatical category anal- added prepositions where they were not required. In
ysis compared children with SLI with age-matched con- a study with a different sample, using a story-retell
trols. Based on our research questions, the grammatical- method, additions of prepositions have also been found
ity measure that most clearly differentiates children with (Stansbury and Jackson-Maldonado 2016a, 2016b) and
SLI from those without SLI is the general grammaticality specific preposition types are analyzed. Thus, it appears
index, independently of whether TD language or age- that prepositions are also susceptible to errors of all
matched controls are used as contrasts with children with types.
SLI. Although many studies use different calculation Verb productions have also been identified as clinical
techniques (as explained above), researchers have consis- markers of SLI. As mentioned above, verb errors are fre-
tently shown that grammaticality is affected in children quent in English-speaking children with SLI and are the
with SLI. In this study we used a more conservative mea- basis of the Rice and Wexler (1986) model. Although
sure than previous studies in that we counted grammat- verb errors were found in this study, differences between
icality by utterance rather than by each ungrammatical the SLI and TD groups were not significant and there
form. Using this more conservative count we still found were only medium effects. Results also showed medium
that grammaticality persisted as the strongest measure. effects for omissions and substitutions, though differ-
The general grammatical errors (omissions, substi- ences were not significant. In a limited number of studies
tutions and additions) across all grammatical categories of Spanish-speaking children, verbs have been shown to
in children with SLI when compared with language- be a problematic structure, but more so when included
matched controls yielded no significant differences, as part of the argument structure or in complex forms
though there were medium effect sizes for omissions. (particularly the subjunctive). Our findings are similar to
Omissions were also the strongest error type that dif- other studies (Andreu and Sanz-Torrent 2013, Jackson-
ferentiated SLI from the TD age control group. They Maldonado 2012, Sanz-Torrent et al. 2008). However,
were significant differences and the effect size was strong. in order to address verb-specific structures that are af-
Thus, an overall finding is that taking into considera- fected by SLI, it is necessary to describe verb types, as
tion all grammaticality, children with SLI omit forms well as tense and mood inflexions, arguments, and to
more frequently than they substitute them when com- contrast simple and complex forms. These analyses ex-
pared with TD children. It would seem, then, that these ceed the goals of this study, but findings suggest that
grammatical forms (articles, clitics, prepositions) are not verbs do not appear to be as affected as other grammat-
yet part of the grammatical repertoire of children with ical categories.
SLI. Several researchers have addressed clitic productions
When grammatical types were analyzed separately to proposing that their use may differentiate children with
determine which form is most affected by SLI, we found SLI from TD when they are observed in story-retell and
consistent evidence to support most published research direct elicitation cloze tasks (Bedore and Leonard 2001,
related to articles. Total article errors showed significant Bosch and Serra 1997, De la Mora 2004, Jacobson
differences with strong effect sizes and, though not sig- 2012, Jacobson and Walden 2013, Morgan et al. 2009,
nificant, medium effect sizes were found for both omis- 2012, Simon-Cereijido and Gutiérrez-Clellen 2007).
sions and substitutions. Studies that have used narratives As mentioned above, results from cloze tasks may not
or SLA (Anderson and Souto 2005, Morgan et al. 2009) be relevant to the discussion of our data because the
have found evidence for both error types, though Andreu referent is still active and processes would be different.
and Sanz-Torrent (2013) only found article omissions Taking this into account, clitic productions were not
in SLA. Thus, articles are problematic for children with significantly different when comparing both groups of
SLI; however, it remains unclear which error type is the children and effects sizes fell below medium levels. This
most salient in these children. finding differs from results presented from story-retell
Children with SLI in this study also had more and SLA studies. In our study, more clitic substitutions
preposition errors. Differences were significant when were found for TD age-matched controls than for
children with SLI were compared with age-matched children with SLI; differences were not significant
controls and there were medium effect sizes. Further- and medium effect sizes were found and no effects
more, prepositions that were omitted, though not with were found for omissions. Thus, using spontaneous
significant differences, did show medium effect sizes. narrative procedures, clitic errors do not seem to
We found no effects for substitutions. Moreover, prepo- be as strong as indicated by prior investigations in
sition additions, which have seldom been discussed which other methodologies are used. Some possible
in the existing literature, approximated medium effect explanations for this discrepancy could be that clitics
levels. Previous studies have suggested that prepositions would more readily be expressed in a narrative retell
12 Donna Jackson-Maldonado and Ricardo Maldonado
where the characters or objects have been previously Clinical implications
established or in an elicitation task in which the referent
The findings of this study are particularly relevant for
is mentioned immediately before the required clitic
intervention strategies. Oftentimes grammatical forms
form. In this study there are multiple clitic omissions
are considered as a group rather than identifying the
and substitutions but statistical differences were not
specific error type. Thus, language evaluations should
established. What occurred is that clitic omissions and
determine not only which grammatical concepts are
substitutions were frequent in both groups of children.
affected, but also whether the affected concepts are
Yet, the TD group had more complex utterances
omitted, substituted or added. In other words, it is im-
(Jackson-Maldonado and Maldonado 2015) and the
portant to determine how grammatical forms are used
referential distance from antecedent (Givón 1983) was
erroneously in order to guide the course of intervention.
greater than for the children with SLI. A reanalysis of
Whereas substitutions entail basic grammatical knowl-
the data, based two experts’ reviews of the transcripts,
edge, such as identifying and inserting some tentative
showed differences in clitic recovery. In the TD group
form in the correct position but not necessarily knowing
clitics were more likely to appear than three utterances
all of its characteristics (gender and number agreement,
after the first mention and, thus, clitic installation for
for instance), omissions may imply that the child has
the TD group could be considered to be more complex
not yet identified that a form has a function and an
than for the SLI group. The children with SLI followed
obligatory place within an utterance. Therefore, in that
three basic strategies: (1) they would never state the ref-
case, children must learn the category in itself. The
erent bur rather merely look at the image; (2) they would
intervention for each type of error would be different,
state the referent at the beginning of the story, and not
but first a correct identification of the process involved
refer back to it; or (3) they would reinstall the referent in
is required. Working with prepositions that are omitted
the production with a complete noun phrase and avoid
is quite distinct than inserting a missing determiner or
using a clitic altogether. For example, they would refer
correcting its gender. Also, intervention should be based
to the frog with a masculine clitic lo rather than the fem-
on the individual characteristics of the child and not
inine clitic la, but they had never used the noun phrase
on broad programmes that do not address individual
la rana (‘the frog’) at the beginning of the story. Most
needs.
clitic errors in the TD group appeared when the referent
Findings from this study also emphasize the need
was not explicitly installed in the story or was more than
to base intervention methods on language-specific
three utterances after the initial reference. As distance
information and not to derive strategies from research
increases the initial referent can be lost as other parts of
based on speakers of other languages. Ungrammaticality
discourse may interfere. In contrast, children with SLI
in Spanish is manifested in very different forms and
are not particularly sensitive to tracking back the ref-
functions than is true for English. In particular, auxil-
erent in discourse. Children with SLI had trouble with
iaries are less sensitive to SLI in Spanish than in English.
clitics in all contexts and distances, though differences
In contrast, articles, clitics, prepositions and connector
were not significant when compared with their TD
words are more vulnerable in Spanish than in English.
peers. This issue requires a more in depth analysis with
As English is one of the primary languages for which
a larger sample of children in order to truly understand
intervention programmes have been developed, given
the nature of clitics. Furthermore, findings gathered
that there is more extensive research on this language,
from different observational techniques should not be
many therapists simply translate programmes and pub-
contrasted.
lications into Spanish without considering its specific
Our study supports some findings in previously
structures and needs. This strategy could be detrimental
published research on Spanish-speaking children with
to a child’s progress, as language-specific forms that are
SLI. Undoubtedly, articles are one of the most prob-
particularly vulnerable may not be targeted.
lematic grammatical forms in this group. Prepositions,
Finally, it is of utmost importance to understand
although often not analyzed, are also forms that are
that ungrammatical forms may be produced in differ-
difficult. The study of clitics requires more in-depth
ent scenarios. Elicitation techniques, as an intervention
analysis and we would propose that clitic functions
strategy, may instigate the child to produce distinct
and reference be considered. What remains clear, as
forms. It is recommended that therapist use multiple
has already been determined in the existing research
means to elicit problematic forms. For instance, both
base, is that Spanish-speaking children with SLI
story-retell and spontaneous narratives should be used
produce more ungrammatical utterances than their TD
with different goals, while sentence-completion exer-
peers. It is also clear that grammatical types specific
cises may yield still different results. These considera-
to the language under analysis should be considered
tions should also be taken into account when assessing
thoroughly.
children.
Grammaticality Spanish SLI 13
Acknowledgements BEDORE, L. M. and PEÑA, E. D., 2008, Assessment of bilingual
children for identification of language impairment: current
This research was supported by a grant to the first author from findings and implications for practice. International Journal
the Consejo Nacional de Ciencia y Tecnologı́a (grant numbers of Bilingual Education and Bilingualism, 11(1), 1–29.
CONACYT-60765 and CONACYT-NSF 2007–2008), an agency BIALYSTOK, E., 2005, Consequences of bilingualism for cognitive
of the Mexican government. The authors also thank the clinics and development. In J. F. Kroll and A. M. B. de Groot (eds),
schools and all the families who participated. They greatly appreciate Handbook of Bilingualism: Psycholinguistic Approaches (New
the extensive corrections of this manuscript done by Lauren Cycyk York: Oxford University Press), pp. 417–432.
and Daniel Stauffer. They also acknowledge the work performed by BORTOLINI, U., ARFÉ, B., CASELLI, C. M., DEGASPERI, L., DEEVY,
Ma Lourdes Martı́nez Nieto, Alejandra Aguillón, Tania Rodrı́guez, P. and LEONARD, L. B., 2006, Clinical markers for spe-
Jan Holst and Ana Susana Mejı́a Villalobos. Declaration of interest: cific language impairment in Italian: the contribution
The authors report no conflicts of interest. The authors alone are of clitics and non-word repetition. International Journal
responsible for the content and writing of the paper. of Language and Communication Disorders, 41(6), 695–
712.
BOSCH, L. and SERRA, M., 1997, Grammatical morphology deficits
References of Spanish-speaking children with specific language impair-
ment. Amsterdam Series in Child Language Development,
AGUADO, G., CUETOS-VEGA, F., DOMEZÁIN, M. and PASCUAL, B., 6(69), 33–45.
2006, Repetición de pseudopalabras en niños españoles con BOTTING, N., 2002, Narrative as a tool for the assessment of linguis-
Trastorno Especı́fico del Lenguaje: marcador psicolingüı́stico. tic and pragmatic impairments. Child Language Teaching and
Revista Neurologı́a, 43, 201–208. Therapy, 18(1), 1–21.
AGUILAR-MEDIAVILLA, E., SANZ-TORRENT, M. and SERRA- BROWN, L., SHERBENOU, R. J. and JONSHEN, S. K., 2000, TONI-2,
RAVENTÓS, M., 2007, Influence of phonology on morpho- Test de Inteligencia No Verbal (Madrid: TEA).
syntax in Romance languages in children with specific lan- BROWNELL, R., 2000a, Expressive One-Word Picture Vocabulary Test
guage impairment (SLI). International Journal of Language (San Antonio, TX: Pearson).
and Communication Disorders, 42(3), 325–347. BROWNELL, R., 2000b, Receptive One-Word Picture Vocabulary Test
ALARCOS-LORACH, E., 1978, Estudios de Gramática Funcional del (San Antonio, TX: Pearson).
Español, 2nd edn (Madrid: Gredos). CALDERÓN, J., 2010, Nonword repetition in emerging bilingual
ANDERSON, R. T. and SOUTO, S., 2005, The use of articles by children with language impairment. Unpublished doctoral
monolingual Puerto Rican Spanish-speaking children with dissertation, University of California—San Diego.
specific language impairment. Applied Psycholinguistics, 26, CASTILLA-EARLS, A. P. and ERIKS-BROPHY, A., 2012, Medidas de eval-
621–647. uación del lenguaje espontáneo en niños hispanohablantes en
ANDREU, L. and SANZ-TORRENT, M., 2013, Evaluación e iden- edad preescolar. Revista de Logopedia, Foniatrı́a y Audiologı́a,
tificación de los niños con TEL. In L. Andreu, G. 32(3), e7–e19.
Aguado, M. C. Cardona, I. Pera and M. Sanz-Torrent COADY, J. A. and EVANS, J. L., 2008, Uses and interpretations of non-
(eds), El Trastorno Especı́fico del Lenguaje (Barcelona: UOC), word repetition tasks in children with and without specific
pp. 89–136. language impairments (SLI). International Journal of Language
ANDREU, L., SANZ-TORRENT, M., BUIL, L. and MACWHINNEY, B., and Communication Disorders, 43(1), 1–40.
2012, The effect of verb argument structure on picture nam- COHEN, J., 1988, Statistical Power Analysis for the Behavioral Sciences,
ing in children with and without specific language impair- 2nd edn (Hillsdale, NJ: Erlbaum).
ment (SLI). International Journal of Language and Communi- CONBOY, B. T., 2012, Language processing and production in infants
cation Disorders, 47(6), 637–653. and toddlers. In B. A. Goldstein (ed.), Bilingual Language
ANDREU, L., SANZ-TORRENT, M., GUARDIA, J. and MACWHINNEY, Development and Disorders in Spanish–English Speakers, 2nd
B., 2013, The formulation of argument structure in SLI: an edn (Baltimore, MD: Paul H. Brookes), pp. 47–71.
eye-movement study. Clinical Linguistics and Phonetics, 27(2), DE LA MORA, J., 2004, Direct object clitics in Spanish-speaking chil-
111–133. dren with and without specific language impairment. Master’s
AUZA, A. and MORGAN, G., 2013, Uso de preposiciones en el re- thesis, University of Alberta.
cuento de una historia. Comparación de niños hispanohab- DISPALDRO, M., LEONARD, L. B. and DEEVY, P., 2013, Clinical
lantes con y sin trastorno del lenguaje. Infancia y Aprendizaje, markers in Italian-speaking children with and without specific
36(1), 35–49. language impairment: a study of non-word and real word rep-
BATES, E. and MACWHINNEY, B., 1989, Functionalism and the com- etition as predictors of grammatical ability. International Jour-
petition model. In B. MacWhinney and E. Bates (eds), The nal of Language and Communication Disorders, 48(5), 554–
Crosslinguistic Study of Sentence Processing (Cambridge: Cam- 564.
bridge University Press), pp. 73–112. DOLLAGHAN, C. and CAMPBELL, T. F., 1998, Nonword repetition
BEDORE, L. M. and LEONARD, L. B., 1998, Specific language impair- and child language impairment. Journal of Speech, Language,
ment and grammatical morphology: discriminant function and Hearing Research, 41(5), 1136–1146.
analysis. Journal of Speech, Language, and Hearing Research, DU BOIS, J. W., 2003, Discourse and grammar. In M. Tomasello (ed),
41(5), 1185–1192. The New Psychology of Language: Cognitive and Functional
BEDORE, L. and LEONARD, L., 2001, Grammatical morphology Approaches to Language Structure, Vol. 2 (London: Erlbaum),
deficits in Spanish-speaking children with specific language pp. 47–87.
impairment. Journal of Speech, Language, and Hearing Re- DUINMEIJER, I., DE JONG, J. and SCHEPER, A., 2012, Narrative
search, 44, 905–924. abilities, memory and attention in children with a specific
BEDORE, L. and LEONARD, L., 2005, Verb inflections and noun language impairment. International Journal of Communication
phrase morphology in the spontaneous speech of Spanish- and Language Disorders, 47(5), 542–555.
speaking children with specific language impairment. Applied DUNN, M., FLAX, J., SLIWINSKI, M. and ARAM, D., 1996, The use
Psycholinguistics, 26, 195–225. of spontaneous language measures as criteria for identifying
14 Donna Jackson-Maldonado and Ricardo Maldonado
children with specific language impairment: an attempt to GUTIÉRREZ-ORDÓÑEZ, S., 1986, Variaciones sobre la atribución (Col.
reconcile clinical and research incongruence. Journal of Speech, Contextos No. 5) (Leon: University of Leon).
Language and Hearing Research, 39(3), 643–650. HERNÁNDEZ PINA, F., 1984, Teorı́as psicosociolinguı́sticas y su apli-
EBERT, K. D., KALANEK, J., CORDERO, K. N. and KOHNERT, cación a la adquisicion del español como lengua materna
K., 2008, Spanish nonword repetition: stimuli development (Madrid: Siglo XXI).
and preliminary results. Communication Disorders Quarterly, HOFF, E., CORE, C., PLACE, S., RUMICHE, R., SENOR, M. and PARRA,
29(2), 67–74. M., 2012, Dual language exposure and early bilingual devel-
EISENBERG, S. L. and GUO, L., 2013, Differentiating children with opment. Journal of Child Language, 39(1), 1–27.
and without language impairment based on grammatical- HOFFMAN, L. M., 2009, The utility of school-age narrative mi-
ity. Language, Speech, and Hearing Services in Schools, 44, crostructure indices: INMIS and the proportion of restricted
20–31. utterances. Language, Speech, and Hearing Services in Schools,
EISENBERG, S. L., GUO, L. and GERMEZI, M., 2012, How grammati- 40(4), 365–375.
cal are three-year-olds? Language, Speech, and Hearing Services HOLST, J., 2014, Uso del subjuntivo en niños con y sin trastorno
in Schools, 43, 36–52. del lenguaje: el veterinario quiere que se está. Master’s thesis,
FRENCH, L. M. and O’BRIEN, I., 2008, Phonological memory and Universidad Autónoma de Querétaro.
children’s second language grammar learning. Applied Psy- JACKSON-MALDONADO, D., 2011, La identificación del Trastorno
cholinguistics, 29, 463–487. Especı́fico de Lenguaje en niños hispano—hablantes
GATHERCOLE S. E. and BADDELEY, A. D., 1990, The role of phono- por medio de pruebas formales e informales. Revista
logical memory in vocabulary acquisition: a study of young Neuropsicologı́a Neuropsiquiatrı́a y Neurociencias, 11(1),
children learning new names. British Journal of Psychology, 81, 33–50.
439–454. JACKSON-MALDONADO, D., 2012, Verb morphology and vocabu-
GATHERCOLE, S. E., SERVICE, E., HITCH, G. J., ADAMS, A. M. lary in monolinguals, emerging bilinguals, and monolingual
and MARTIN, A. J., 1999, Phonological short term mem- children with primary language impairment. In B. Gold-
ory and vocabulary development: further evidence on the stein, (ed.), Bilingual Language Development and Disorders in
nature of the relationship. Applied Cognitive Psychology, 13, Spanish–English Speakers, 2nd edn (Baltimore, MD: Brookes),
65–77. pp. 153–173.
GATHERCOLE, V. C. M. and THOMAS, E. M., 2009, Bilingual first- JACKSON-MALDONADO, D. and CONBOY, B. T., 2007, Utterance
language development: dominant language takeover, threat- length measures for Spanish-speaking toddlers: the mor-
ened minority language take-up. Bilingualism: Language and pheme versus word issue re-visited. In J. G. Centeno, R.
Cognition, 12, 213–237. T. Anderson and L. K. Obler (eds), Communication Disorders
GENESEE, F., 2006, Bilingual first language acquisition in perspective. in Spanish Speakers: Theoretical, Research and Clinical Aspects
In P. McCardle and E. Hoff (eds), Childhood Bilingualism: Re- (Clevedon, UK: Multilingual Matters), pp. 142–155.
search on Infancy through School Age (Clevedon: Multilingual JACKSON-MALDONADO, D. and MALDONADO, R., 2015, La com-
Matters), pp. 45–67. plejidad sintáctica en niños con y sin Trastorno Primario de
GENESEE, F., PARADIS, J. and CRAGO, M. B., 2011, Dual Language Lenguaje. In I. Rodrı́guez Sánchez and E. Vázquez (eds),
Development & Disorders: A Handbook on Bilingualism & Sec- Lingüı́stica Funcional (Querétaro: Universidad Autónoma de
ond Language Learning, 2nd edn (Baltimore, MD: Paul H. Querétaro), pp. 253–301.
Brookes). JACKSON-MALDONADO, D., MALDONADO, R. and THAL, D. J., 1998,
GIRBAU, D. and SCHWARTZ, R., 2007, Nonword repetition in Reflexive and middle markers in early child language acquisi-
Spanish-speaking children with specific language impairment tion: evidence from Mexican Spanish. First Language, 18(54),
(SLI). International Journal of Language and Communication 403–429.
Disorders, 42(1), 59–75. JACKSON-MALDONADO, D., MEJÍA VILLALOBOS, A. S. and HOLST, J.,
GIVÓN, T., 1983, Topic Continuity in Discourse. A Quantitative Cross- 2015, Similitudes y diferencias entre tareas de repetición de
Language Study (Amsterdam: John Benjamins). no-palabras. ¿Qué formato es mas acertado? [Similarities and
GRISSOM, R. J. and KIM, J. J., 2012, Effect Sizes for Research: Univari- differences between non-word repetition tasks. Which format
ate and Multivariate Applications, 2nd edn (New York, NY: works best?]. Paper presented at the 16th Annual Meeting of
Taylor & Francis). Child Language, Mexico.
GRUETER, T. and CRAGO, M., 2012, Object clitics and their omission JACOBSON, P. and WALDEN, P. R., 2013, Lexical diversity and omis-
in child L2 French: the contributions of processing limitations sion errors as predictors of language ability in the narratives of
and L1 transfer. Bilingualism: Language and Cognition, 15(3), sequential Spanish–English bilinguals: a cross-language com-
531–549. doi:10.1017/S1366728911000113 parison. American Journal of Speech Language Pathology, 22,
GUO, L. Y. and SCHNEIDER, P., 2016, Differentiating school-aged 554–565.
children with and without language impairment using tense JACOBSON, P. F., 2012, The effects of language impairment on the
and grammaticality measures from a narrative task. Journal of use of direct object pronouns and verb inflections in her-
Speech, Language, and Hearing Research, 59(2), 1–13. itage Spanish speakers: a look at attrition, incomplete acquisi-
GUTIÉRREZ-CLELLEN, V. F., RESTREPO, M. A. and SIMON-CEREIJIDO, tion and maintenance. Bilingualism: Language and Cognition,
G., 2006, Evaluating the discriminant accuracy of a gram- 15(1), 22–38.
matical measure with Spanish-speaking children. Journal JAKUBOWICZ, C., 2011, Measuring derivational complexity: new
of Speech, Language, and Hearing Research, 49(6), 1209– evidence from typically developing and SLI learners of L1
1223. French. Lingua, 121, 339–351.
GUTIÉRREZ-CLELLEN, V. and SIMÓN-CEREIJIDO, G., 2010, Using JAKUBOWICZ, C., NASH, L., RIGAUT, C. and GERARD, C. L., 1998,
nonword repetition tasks for the identification of language Determiners and clitic pronouns in French-speaking children
impairment in Spanish–English-speaking children: does the with SLI. Language Acquisition, 7, 113–160.
language of assessment matter? Learning Disabilities Research KOHNERT, K., 2010, Bilingual children with primary language
and Practice, 25(1), 48–58. impairment: issues, evidence and implications for clinical
Grammaticality Spanish SLI 15
actions. Journal of Communication Disorders, 43(6), 456– RESTREPO, M. A., 1998, Identifiers of predominantly Spanish-
473. speaking children with language impairment. Journal of
KÖPPE, R., 2001, Object omissions in young bilingual children: Speech, Language, and Hearing Research, 41(6), 1398–1411.
assessing the evidence. Bilingualism: Language and Cognition, RESTREPO, M. A. and GUTIÉRREZ-CLELLEN, V. F., 2001, Article use in
4, 33–35 doi:10.1017/S1366728901250111 Spanish-speaking children with specific language impairment.
LANGACKER, R. W., 1987, Foundations of Cognitive Grammar: Theo- Journal of Child Language, 28(2), 433–452.
retical Prerequisites (Stanford, CA: Stanford University Press). RESTREPO, M. A. and GUTIÉRREZ-CLELLEN, V. F., 2012, Grammat-
LANZA, E., 1997, Language Mixing in Infant Bilingualism: A Soci- ical impairments in Spanish–English speaking children. In
olinguistic Perspective (Oxford: Clarendon). B. Goldstein (ed.), Bilingual Language Development and Dis-
LEE, L. L., 1974, Developmental Sentence Analysis (Evanston, IL: orders in Spanish–English Speakers, 2nd edn (Baltimore, MD:
Northwestern University Press). Brookes), pp. 213–232.
LEECH, N. L. and ONWUEGBUZIE, J., 2002, A call for greater use RICE, M. L. and WEXLER, K., 1996, Toward tense as a clinical
of nonparametric statistics. Paper presented at the Annual marker of specific language impairment in English-speaking
Meeting of the Mid-South Educational Research Association, children. Journal of Speech and Hearing Research, 39(6), 1239–
Chattanooga, TN, USA. 1257.
LEONARD, L. B. and DISPALDRO, M., 2013, The effects of pro- RODRÍGUEZ-ESPINEIRA. M. J., 1992, Sobre la codificación infor-
duction demands on grammatical weaknesses in specific mativa de las cláusulas con predicativo en español. Revue
language impairment: the case of clitic pronouns in Italian. Romane, 27(1), 30–60.
Journal of Speech, Language, and Hearing Research, 56, 1272– RODRÍGUEZ HERNÁNDEZ, T. and JACKSON-MALDONADO, D., 2009,
1286. Repetición de no-palabras en niños con Trastorno Especı́fico
LOPE-BLANCH, J. M., 1984, El Concepto de Oración en la Lingüı́stica del Lenguaje. Paper presented at the XIth Conference of
Española (Mexico City: Universidad Nacional Autónoma). the Latin American Neuropsychological Society, Puebla,
MALDONADO, R., 1988, Energetic reflexives in Spanish. Annual Mexico.
Meeting of the Berkeley Linguistics Society, 14, 153–165. ROYLE, P. and COURTEAU, E., 2013, Language processing in children
MALDONADO, R., 1999, A Media Voz: Problemas Conceptuales del with specific language impairment: a review of event-related
Clı́tico Se (Mexico City: Universidad Nacional Autónoma de potential studies. In Language Processing: New Research (Haup-
México). pauge, NY: Nova), pp. 1–32.
MAYER, M., 1969, Frog, Where Are You? (New York, NY: Dial). SANZ-TORRENT, M., ANDREU, L., BADIA, I. and SIDERA, F., 2011,
MILLER, J. F. and IGLESIAS, A., 2012, Systematic Analysis of Language Argument omissions in preschool Catalan and Spanish
Transcripts (SALT), English & Spanish (Version 12) [computer speaking children with SLI. Infancia y Aprendizaje, 34(1),
software] (Madison, WI: University of Wisconsin, Madison, 49–66.
Waisman Center, Language Analysis Laboratory). SANZ-TORRENT, M., SERRAT, E., ANDREU, L. and SERRA, M., 2008,
MONTRUL, S., 2004, The Acquisition of Spanish: Morphosyntactic Verb morphology in Catalan and Spanish in children with
Development in Monolingual and Bilingual L1 Acquisition and specific language impairment: a developmental study. Clinical
Adult L2 Acquisition (Amsterdam: John Benjamins). Linguistics and Phonetics, 22(6), 459–474.
MORGAN, G., RESTREPO, M. A. and AUZA, A., 2009, Variability in SEMEL, E., WIIG, W. H. and SECORD, W., 2006, Clinical Evaluation of
the grammatical profiles of Spanish-speaking children with Language Fundamentals. 4th Edition (CELF-4) (San Antonio,
specific language impairment. In J. Grinstead (ed.), Hispanic TX: Psychological Corp.).
Child Language: Typical and Impaired Development (Philadel- SERRAT, E. and APARICI, M., 2001, Morphological errors in early
phia, PA: John Benjamins), pp. 283–302. language acquisition: evidence from Catalan and Spanish. In
MORGAN, G., RESTREPO, M. A. and AUZA, A., 2012, Comparison M. Almgren, A. Barreña, M. J. Ezeizabarrena, I. Idiazabal and
of Spanish morphology in monolingual and Spanish–English B. MacWhinney (eds), Research on Child Language Acquisition
bilingual children with and without language impairment. (Somerville, MA: Cascadilla), pp. 1260–1277.
Bilingualism: Language and Cognition, 16, 1–19. SILVA-CORVALÁN, C., 1994, Language Contact and Change: Spanish
MOYLE, M. J., KARASINSKI, C., ELLIS WEISMER, S. and GORMANB, in Los Angeles (Oxford: Oxford University Press).
B. K., 2011, Grammatical morphology in school-age chil- SIMON-CEREIJIDO, G. and GUTIÉRREZ-CLELLEN, V. F., 2007, Spon-
dren with and without language impairment: a discriminant taneous language markers of Spanish language impairment.
function analysis. Language, Speech, and Hearing Services in Applied Psycholinguistics, 28, 317–339.
Schools, 42, 550–560. SMYK, E., RESTREPO, M. A., GORIN, J. S. and GRAY, S., 2013, De-
ODLIN, T., 1989, Language Transfer: Cross-Linguistic Influence in velopment and validation of the Spanish–English Language
Language Learning (Cambridge: Cambridge University Press). Proficiency Scale (SELPS). Language, Speech, and Hearing
PARADIS, J., CRAGO, M. and GENESEE, F., 2003, Object clitics as Services in Schools, 44(3), 252–265.
a clinical marker of SLI in French: evidence from French– STANSBURY, A. and JACKSON-MALDONADO, D., 2016a, Las preposi-
English bilingual children. In B. Beachly, A. Brown and ciones en las narrativas de niños con Trastorno Especı́fico
F. Conlin (eds), Proceedings of the Annual Boston University del Lenguaje. Paper presented at the Congreso Asociación
Conference on Language Development, 27 (Somerville, MA: Española de Logopedia, Foniatrı́a y Audiologı́a. Bilbao, Spain.
Cascadilla), pp. 638–649. STANSBURY, A. and JACKSON-MALDONADO, D., 2016b, Not all
PEÑA, E. D., 2007, Lost in translation: methodological consider- prepositions are created equal: grammaticality in Spanish-
ations in cross-cultural research. Child Development, 78(4), speaking children with SLI. Poster presented at the American
1255–1264. Speech, Language and Hearing Association (ASHA) Confer-
Publication Manual of the American Psychological Association, ence, Philadelphia, PA, USA.
5th Ed., 2001 (Washington, DC: American Psychological SUMMERS, C., BOHMAN, T. M., GILLAM, R. B., PEÑA, E. D. and
Association). BEDORE, L. M., 2010, Bilingual performance on nonword
16 Donna Jackson-Maldonado and Ricardo Maldonado
repetition in Spanish and English. International Journal of WEISMER, S. E. and EVANS, J. L., 2002, The role of processing limita-
Language and Communication Disorders, 45(4), 480–493. tions in early identification of specific language impairment.
TOMASELLO, M., 2009, Constructing a Language: A Usage-Based The- Topics in Language Disorders, 22, 15–29.
ory of Language Acquisition (Cambridge, MA: Harvard Uni- WEISMER, S. E. and KAUSHANSKAYA, M., 2010, The use of descriptive
versity Press). data from bilingual children to inform theories of specific
TOMBLIN, J. B., RECORDS, N. L., BUCKWALTER, P., ZHANG, X., language impairment. Applied Psycholinguistics, 31, 277–282.
SMITH, E. and O’BRIEN, M., 1997, Prevalence of spe- WINDSOR, J., KOHNERT, K., LOBITZ, K. and PHAM, G., 2010, Cross-
cific language impairment in kindergarten children. Jour- language nonword repetition by bilingual and monolingual
nal of Speech, Language and Hearing Research, 40, 1245– children. American Journal of Speech–Language Pathology, 19,
1260. 298–310.

Appendix A: Non-Word Repetition Test

One syllable Two syllables Three syllables Four syllables Five syllables
Tar Piadur Numetia Dicastelo Empuleconi
Fun Muca Tablecó Catı́fago Acubénipo
Sil Dantú Etibo Depolima Milocadunte
Fol Quiesa Jemipa Alefico Kiyatanidul

Note: Words are by syllable length (Jackson-Maldonado 2011).

Das könnte Ihnen auch gefallen