are early grammatical and phonological working memory abilities affected by preterm birth?

18
Are early grammatical and phonological working memory abilities affected by preterm birth? Alessandra Sansavini a, * , Annalisa Guarini a , Rosina Alessandroni b , Giacomo Faldella b , Giuliana Giovanelli a , Gianpaolo Salvioli b a Department of Psychology, University of Bologna, Italy b Institute of Neonatology and Pediatrics, University of Bologna, Italy Received 23 January 2006; received in revised form 6 June 2006; accepted 16 June 2006 Abstract There have been few investigations of the effects of very immature preterm birth on specific linguistic competencies and phonological working memory at preschool age. Study 1 aimed to investigate early grammatical abilities in very immature healthy preterms, taking into account their cognitive development and biological and social factors. The linguistic and cognitive differences found between preterms and fullterms led to investigate in Study 2 the role of phonological working memory on preterms’ grammatical development. Very immature preterm birth resulted to affect grammatical, cognitive and phonological working memory abilities until 3.5 years leading to persisting difficulties in comparison with fullterms, albeit not severe deficits. Tight relations between phonological working memory and grammar were found both in preterms and fullterms, that highlights the reciprocal support of these abilities in development. A partial compensatory effect by the maternal level of education on preterms’ grammatical and cognitive abilities was also found. Learning outcomes: The reader will become familiar with the relations between grammatical and phonological working memory abilities in typical and preterm 3.5-year-old children. # 2006 Elsevier Inc. All rights reserved. Journal of Communication Disorders 40 (2007) 239–256 * Corresponding author at: Dipartimento di Psicologia, Universita ` degli Studi di Bologna, Viale Berti Pichat, 5, 40127 Bologna, Italy. Tel.: +39 051 2091879; fax: +39 051 243086. E-mail address: [email protected] (A. Sansavini). 0021-9924/$ – see front matter # 2006 Elsevier Inc. All rights reserved. doi:10.1016/j.jcomdis.2006.06.009

Upload: independent

Post on 23-Nov-2023

0 views

Category:

Documents


0 download

TRANSCRIPT

Are early grammatical and phonological working

memory abilities affected by preterm birth?

Alessandra Sansavini a,*, Annalisa Guarini a, Rosina Alessandroni b,Giacomo Faldella b, Giuliana Giovanelli a, Gianpaolo Salvioli b

a Department of Psychology, University of Bologna, Italyb Institute of Neonatology and Pediatrics, University of Bologna, Italy

Received 23 January 2006; received in revised form 6 June 2006; accepted 16 June 2006

Abstract

There have been few investigations of the effects of very immature preterm birth on specific

linguistic competencies and phonological working memory at preschool age. Study 1 aimed to

investigate early grammatical abilities in very immature healthy preterms, taking into account

their cognitive development and biological and social factors. The linguistic and cognitive

differences found between preterms and fullterms led to investigate in Study 2 the role of

phonological working memory on preterms’ grammatical development. Very immature preterm

birth resulted to affect grammatical, cognitive and phonological working memory abilities until

3.5 years leading to persisting difficulties in comparison with fullterms, albeit not severe deficits.

Tight relations between phonological working memory and grammar were found both in preterms

and fullterms, that highlights the reciprocal support of these abilities in development. A partial

compensatory effect by the maternal level of education on preterms’ grammatical and cognitive

abilities was also found.

Learning outcomes: The reader will become familiar with the relations between grammatical and

phonological working memory abilities in typical and preterm 3.5-year-old children.

# 2006 Elsevier Inc. All rights reserved.

Journal of Communication Disorders 40 (2007) 239–256

* Corresponding author at: Dipartimento di Psicologia, Universita degli Studi di Bologna, Viale Berti Pichat, 5,

40127 Bologna, Italy. Tel.: +39 051 2091879; fax: +39 051 243086.

E-mail address: [email protected] (A. Sansavini).

0021-9924/$ – see front matter # 2006 Elsevier Inc. All rights reserved.

doi:10.1016/j.jcomdis.2006.06.009

1. Introduction

Over the last 15 years medical progress has made possible the increasing survival of

very immature preterm newborns (gestational age < 33 weeks, birthweight < 1500 g). At

the same time, perinatal complications and cerebral damage are increased, because of both

the higher immaturity (frequently associated with respiratory, infective or cardiovascular

perinatal complications) and the consequent long hospitalization (Volpe, 2001). Our paper

aims to investigate the effects of very immature preterm birth, without cerebral damage, on

early grammatical and phonological working memory abilities, taking into account

cognitive development. Indeed, there have been few investigations of the effects of very

immature preterm birth on specific linguistic competencies at preschool age. In addition,

the role of phonological working memory on grammatical development has rarely been

investigated in at risk populations such preterm children.

Very immature preterm birth constitutes a biological risk factor, even in absence of

cerebral damage, since it occurs in a period sensitive for the development of the central

nervous system (CNS). Indeed, between 23 and 40 weeks of gestation an extensive

maturation of the brain occurs. Brain volume and cortical folding increase with gestational

age at different rates in different regions of the brain (Counsell, Rutherford, Cowan, &

Edwards, 2003), glial cells migrate from the germinal matrix and reach the cortex around

30 weeks of gestation (Maalouf et al., 1999), dendrites are growing and synapses start to

develop (Ferrari, Sturloni, & Cavazzuti, 1982). As the CNS is forming, sensory systems

develop as well. The auditory system starts to function at about 25 weeks of gestation and

its reactivity stabilizes around 32 weeks (Kisilevsky, Hains, & Low, 1999). Discrimination

of two auditory stimuli begins in fetuses and preterms between 30 and 36 weeks and

preferences for familiar linguistic stimuli appear between 34 and 39 weeks (Cheour-

Luhtanene et al., 1996; DeCasper, Lecanuet, Busnel, Granier-Deferre, & Maugeais, 1994;

Giovanelli, Sansavini, & Farneti, 1999).

Furthermore, preterm birth exposes an immature CNS to external stimuli (those of the

neonatal intensive care unit—NICU) which are frequently invasive, non-contingent and

disorganizing and cause instability to autonomic and motor systems, state organization,

attention and self-regulation (Als, 1992). Indeed, many changes have been made in the NICU

to enhance the physiological stability of these infants, to limit invasive stimulations and to

promote an early relationship between them and their parents. However, high immature

preterms require prolonged medical interventions (e.g., to assure alimentary, respiratory and

cardiac functions) which result in an artificial and often painful environment.

It may thus be assumed that these adverse biological and environmental conditions

affect the development of brain functions and structures and, consequently, the

development of sensory, motor and cognitive systems which depend on the interaction

between neurobiological maturation and environmental stimuli (Elman et al., 1996;

Giovanelli et al., 1999). A recent study showed that, although brain volume in preterm

infants at 40 weeks of post-conceptional age was similar to that of fullterms, the surface

area of the cortex and cortical folding was reduced in the preterms (Ajayi-Obe, Saeed,

Cowan, Rutherford, & Edwards, 2000). If these assumptions are correct, delay might be

expected in infants with higher immaturity, even in the absence of cerebral damage. At the

same time, brain plasticity should be taken into account. For instance, in language

A. Sansavini et al. / Journal of Communication Disorders 40 (2007) 239–256240

development, fullterms, who experienced early post-natal focal lesions, performed at 5

years at the lower level of a normal range, showing that partial compensation of the lesion,

presumably supported by cortical reorganizations, had occurred. In at risk populations,

such as preterms, compensation may partly depend on social factors, such as maternal and

paternal level of education and middle-high socio-economic status (Barsky & Siegel, 1992;

Molfese, Holcomb, & Helwig, 1994).

2. Study 1. Grammatical and cognitive development in very immature preterms

Little agreement exists on the effects of very immature preterm birth, without cerebral

damage, on language and cognitive development. Several studies have found that,

compared with fullterms, preterms show less reactivity to social stimuli and fewer

vocalizations in the first year of life (Beek van, Hopkins, & Hoeksma, 1994; Oller, Eilers,

Steffens, & Lynch, 1994), together with linguistic delays at one (Byers-Brown, Bendersky,

& Chapman, 1986) and 2 years (Vohr, Garcia Coll, & Oh, 1988). However, studies of the

preschool years have given rise to conflicting results, that may partially depend on

methodological choices, such as the criteria of sample selection, the presence of a control

group, the chosen tasks (Gallagher & Watkin, 1998). Some authors did not find significant

differences between preterms and fullterms (Menyuk, Liebergott, & Schultz, 1995),

whereas others found that, until 5 years (Largo, Molinari, Comenale Pinto, Weber, & Duc,

1986; Sansavini, Rizzardi, Alessandroni, & Giovanelli, 1996) and at school age (Anderson

& Doyle, 2003; Wolke & Meyer, 1999), the linguistic and cognitive development of very

low birthweight and gestational age healthy preterms, even if within the normal range, was

significantly lower than that of a matched fullterm group. These studies support the idea

that biological risk factors may have long-term effects, even in presence of protective

social factors.

Regarding the development of specific linguistic competencies, slight delays were

found in 2.5-year-old very immature preterms: lexical development was affected by

birthweight � 1000 g and male gender, while grammatical development by gestational

age < 31 weeks and male gender (Sansavini et al., 2006). Other authors showed that at 3.5

and at 5 years preterms produced fewer types and tokens of verbs compared with fullterms,

and this was true even for those preterms with a birthweight higher than 1500 g (Le-

Normand & Cohen, 1999). Another study reported that, within a sample of 3–4-year-old

preterms with very low gestational age, only some had difficulties in language

comprehension, production and repetition of non-words (Briscoe, Gathercole, & Marlow,

1998). However, causal risk factors were not clearly identified, even if a role for neonatal

respiratory complications was hypothesized.

Our study thus aimed to investigate the effects of very immature preterm birth, without

cerebral damage, on preschool language development. In particular, it focused on

grammatical abilities, the development of which is critical at 3–4 years. Indeed, in the

Italian language, by 3.5 years typically developing children produce complete and

articulated sentences, mostly correct both in morphosyntax and phonology (Bortolini & De

Gasperi, 2002; Chilosi, Cipriani, & Fapore, 2002). At this age, grammatical development is

a good index of the rate of overall linguistic development and, in the case where a

A. Sansavini et al. / Journal of Communication Disorders 40 (2007) 239–256 241

telegraphic style (without function words) is still used or the canonical order of words in

the sentence has not yet been acquired, this suffices to detect grammatical delay.

Furthermore, between 3.5 and 4 years SLI can be diagnosed. If, as hypothesized above,

very immature preterm birth affects language development, differences in grammatical

development should be expected at preschool age between preterms and fullterms.

Cognitive abilities should also be evaluated in order to determinate whether only linguistic

or also cognitive difficulties characterize preterm children. With regard to school-age

children, Aram, Hack, Hawkins, Weissman, and Borawski-Clark (1991) found that

preterms have a general linguistic and cognitive disadvantage with respect to fullterms. A

recent study on very immature preterms’ linguistic and cognitive development found that at

2.5 years language and cognition were strictly related and those preterms who were at risk

for lexical production also had lower grammatical and cognitive scores with respect to

preterms not at risk (Sansavini et al., 2004, 2006).

A second aim was to examine the effects of biological (gestational age, gender) and

social (maternal and paternal level of education) factors on preschool grammatical and

cognitive abilities and investigate whether the considered social factors have a

compensatory role for neonatal biological risk factors. If these social factors play a

role, both preterms and fullterms should have better performance in the presence of

protective social factors (i.e., higher levels of maternal and paternal education), as found by

Molfese et al. (1994) in 3-year-old children. However, Largo et al. (1986), comparing

language development in preterms and controls from 1 to 5 years, have found that the

socio-economic status affected fullterms’ linguistic abilities at all considered ages, while it

affected those of preterms only at 5 years. By contrast, some authors claim that social

factors may not have great compensatory effects at preschool and school age, in the case of

cerebral lesions or of very low gestational age and birthweight (Wolke & Meyer, 1999). In

the study of Largo et al. (1986) male gender also had a negative effect on language

development until preschool age, independently of social factors, both in preterms and

controls. A recent study did not find any effect, at 2.5 years, of parental level of education

on both preterms’ and fullterms’ lexical and grammatical development, which were mainly

affected by biological risk factors (Sansavini et al., 2006).

2.1. Method

2.1.1. Participants

Ninety monolingual Italian preterms took part in a follow-up study which was

conducted at the Institute of Neonatology and Pediatrics of Bologna University. Birth

dates ranged from October 1995 to November 1999. At birth as well as at the presumed

date of birth and at 3 months (corrected age), cerebral echography was carried out

routinely. Contact between mothers and their preterm neonates in the incubators was

encouraged.

For the present study, preterm children were recruited if, at birth, they had a

gestational age � 33 weeks. We allowed for some degree of medical complication

related to their premature birth (respiratory distress with or without mechanical

ventilation, bronchodysplasia, apnea, intra-ventricular hemorrhage of I or II grade,

intrauterine growth retardation, ROP at birth of I or II grade, visual problems at 3.5

A. Sansavini et al. / Journal of Communication Disorders 40 (2007) 239–256242

years, persistent hyperechogenicity, hyperbilirubinemia with phototherapy). By

contrast, those with cerebral palsy, leukomalacia, intra-ventricular hemorrhage >II

grade, hydrocephalus, motor handicaps or significant sensory impairments, at birth or at

the subsequent medical assessments, were excluded. Preterms, whose parents were not

of Italian mother-tongue, were also excluded, assuming that this might affect the infant’s

language development.

The mean gestational age at birth of the preterm group was 30.1 weeks (S.D. = 2.3),

with a range from 25 to 33 weeks. Thirty-nine had a gestational age < 31 weeks and 51 had

�31 weeks. Their mean birthweight was 1213.1 g (S.D. = 260.6), with a range from 550 to

1600 g. Twenty-five children weighed at birth �1000 g and 65 weighed >1000 g. Fifty-

two were males and 38 females. Parental educational level was distributed as follows.

Twenty-two mothers had a low educational level (basic), 43 a medium level (high school)

and 25 a high level (University). Thirty-six fathers had a low educational level (basic), 36 a

medium level (high school) and 18 a high level (University). Preterm children were tested

at 3.5 years (M = 42.3 months, S.D. = .7, range = 41–45). The ages of the preterm children

were based on their expected date of delivery, thereby corrected for prematurity. The

choice of corrected age seemed to us proper to evaluate preterms’ language development

considering the high neonatal immaturity of our preterm sample, as done by other studies

evaluating very immature preterms in the first years of life (Grunau, Kearney, & Whitfield,

1990; Hindmarsh, O’Callaghan, Mohay, & Rogers, 2000). Indeed, in studying early

language development the interaction between CNS level of maturation and environmental

stimuli and the consequent rapid and continuous changes in linguistic abilities should be

taken into account (Bates & Goodman, 1997).

A comparison group of 40 Italian fullterm children, who had experienced normal birth

(gestational age > 37 weeks and birthweight > 2800 g) and absence of neonatal

complications and whose parents were of Italian mother-tongue, was recruited. Seventeen

were males and 23 females. Parental educational level was distributed as follows. Seven

mothers had a low educational level (basic), 20 a medium level (high school) and 13 a high

level (University). Eight fathers had a low educational level (basic), 19 a medium level

(high school) and 13 a high level (University). Fullterm children were tested at 3.5 years

(M = 42.1 months, S.D. = .7, range = 42–45).

A Chi-square analysis indicated that the preterm and the fullterm sample did not differ

by gender [x2 (1, N = 130) = 2.60, p = .107], nor by maternal level of education [x2 (2,

N = 130) = .84, p = .658] nor by paternal level of education [x2 (2, N = 130) = 5.46,

p = .065]. An independent samples t-test indicated that the preterm and the fullterm group

were equivalent in terms of children’s age [t (128) = 1.49, p = .14].

2.1.2. Materials

2.1.2.1. Grammatical ability. An Italian test of repetition of phrases and sentences – test

di ripetizione di frasi (TRF, Devescovi & Caselli, 2001) – was administered to each child.

The TRF, designed to investigate Italian children’s grammatical ability from 2 to 4 years,

consists of 24 noun phrases (constituted by articles and nouns) and 27 sentences of

different length and grammatical complexity (constituted by three, four, five or six words

belonging to the following categories: nouns, verbs, adjectives/adverbs and function

words, i.e., articles and prepositions). For the present study the 27 sentences of the TRF

A. Sansavini et al. / Journal of Communication Disorders 40 (2007) 239–256 243

were administered and analyzed. The sentences were presented in a randomized sequence

for all children. Some examples of sentences were: la macchina e rossa, ‘the car is red’ and

Anna porta la torta in cucina, ‘Anna carries the cake into the kitchen’. The test was

presented as a game, introduced as follows: ‘‘Now let’s play a game together: I will say

something and you will say the same thing’’. If the child did not repeat, the sentence was

repeated a second time. Each sentence was presented together with a colored card depicting

its meaning. The mean length of utterance (MLU) score and total morphological errors

were scored. The latter included omissions, substitutions, and errors in bounded

morphemes (which define number and grammatical gender of nouns, adjectives and

articles, and number and person of verbs).

2.1.2.2. Cognitive ability. The Italian version of the L–M form of the Stanford–Binet

Intelligence Scale (Bozzo & Mansueto Zecca, 1968) was also administered to determine an

intelligence quotient (IQ). Since the 4th revised edition of this test (i.e., that with

subscales), was neither translated nor standardized in Italy, the L–M version was used.

Indeed, other recent Italian studies on at risk populations have used this test (Vicari,

Caravale, Carlesimo, Casadei, & Allemand, 2004).

2.1.3. Procedure

Preterm children were tested individually with the TRF and the Stanford–Binet

Intelligence Scale in a silent room of the Institute of Pediatrics and Neonatology of

Bologna University. Fullterm children were tested individually with the same tests in a

silent room of their kindergarten. During the administration of the tests the children were

videotaped. Two independent observers, previously trained, made the transcriptions of

children’s repetitions of the sentences of the TRF and coded them.

2.2. Results

2.2.1. Differences between preterms and controls in grammatical and cognitive

abilities

Descriptive statistics for the preterm group and the control group on key measures are

provided in Table 1. Scores for the total preterm group were lower than those for the control

group.

To investigate the differences between the preterm and fullterm sample, one-way

analyses of variance (ANOVAs) were performed using group membership (preterm and

fullterm) as the between-subjects factor (see Table 1). Measurements of the TRF (MLU,

total morphological errors including omissions, substitutions and errors in bounded

morphemes, and their distribution in each word category – nouns, function words, verbs,

adjectives/adverbs) and of the cognitive test (IQ) were used as the dependent variables.

Significant differences were found in MLU [F(1, 128) = 10.43, p = .002], in total

morphological errors [F(1, 128) = 10.67, p = .001] and in omissions [F(1, 128) = 10.43,

p = .002]. Significant differences were found in most word categories. Preterms showed

more errors than controls in nouns [F(1, 128) = 8.41, p = .004], function words [F(1,

128) = 11.25, p = .001], and verbs [F(1, 128) = 8.37, p = .004]. Finally, a significant

difference was found between preterms and fullterms in IQ [F(1, 128) = 27.36, p < .001].

A. Sansavini et al. / Journal of Communication Disorders 40 (2007) 239–256244

2.2.2. Effects of biological and social factors on grammatical and cognitive abilities

To understand the role of biological (gestational age, gender) and social (maternal and

paternal level of education) factors on preterms’ and fullterms’ grammatical competence

(MLU) and IQ, several ANOVAs were run. As in other follow-ups (e.g., Taylor, Klein,

Minich, & Hack, 2000), ANOVA was used to compare groups with a different level of risk.

With regard to the preterm sample, 2 four-way ANOVAs were run using gestational age

(<31 weeks versus �31 weeks), gender (male versus female), maternal level of education

(low versus medium versus high) and paternal level of education (low versus medium versus

high) as the independent variables. In the first ANOVA (dependent variable: MLU) a

difference was found with respect to maternal level of education, corresponding to the

expected direction, even if it did not attain significance [F(2, 60) = 2.78, p = .07].

Comparisons among the means using the Bonferroni procedure showed that children of

mothers with high education had a higher MLU (M = 4.19) than that of mothers with low

education (M = 3.48, p = .045). Gestational age, gender and paternal level of education

showed no main effects. Moreover, interaction effects were not found. In the second ANOVA

(dependent variable: IQ) a significant main effect of maternal level of education was found

[F(2, 60) = 4.53, p = .015]. Comparisons among the means using the Bonferroni procedure

showed that children of mothers with high education had a higher IQ (M = 108.34) than that

of mothers with low education (M = 99.98, p = .028). Gestational age, gender and paternal

level of education showed no main effects. Moreover, no interaction effects were found.

With regard to the control sample, 2 three-way ANOVAs were run using gender (male

versus female), maternal level of education (low versus medium versus high) and paternal

level of education (low versus medium versus high) as the independent variables. Both in

the first ANOVA (dependent variable: MLU) and in the second ANOVA (dependent

A. Sansavini et al. / Journal of Communication Disorders 40 (2007) 239–256 245

Table 1

Scores in grammatical abilities (TRF) and IQ in preterm and control (fullterm) children (Study 1)

Preterms Fullterms

Mean S.D. Range Mean S.D. Range

TRF

MLU 3.8 (N = 90) .9 1.3–4.7 4.2** (N = 40) .6 1.9–4.7

Types of morphological errors

Omissions (O) 24.5 (N = 90) 23.3 0–90 11.4** (N = 40) 15.5 0–75

Substitutions (S) 1.1 (N = 90) 1.8 0–11 1.2 (N = 40) 2.1 0–10

Errors in bounded

morphemes (E)

1.2 (N = 90) 1.8 0–9 .7 (N = 40) 1.1 0–4

Total morphological

errors (O + S + E)

26.7 (N = 90) 23.4 0–91 13.3** (N = 40) 16.6 0–78

Distribution of morphological errors in each category

Nouns 8.1 (N = 90) 8.1 0–31 4** (N = 40) 6.1 0–28

Function words 12.9 (N = 90) 10.6 0–39 6.8** (N = 40) 7 0–30

Verbs 4.9 (N = 90) 5.1 0–21 2.3** (N = 40) 3.8 0–17

Adjectives/adverbs .9 (N = 90) 1.3 0–5 .5 (N = 40) 1.4 0–6

IQ 104.2 (N = 90) 11.1 81.8–135.7 114.7*** (N = 40) 9.2 90.5–130.9

One-way ANOVA test. *p < .05; **p < .01; ***p < .001.

variable: IQ) gender, maternal and paternal level of education showed nor main effects nor

interaction effects.

2.3. Discussion

The first aim of this study was to investigate the effects of very immature preterm birth,

without cerebral damage, on grammatical abilities at a time, 3.5 years, when typically

developing children produce complete and articulated sentences. Cognitive abilities were

also evaluated in order to assess whether only linguistic or also cognitive difficulties

characterize preterm children. The results show that preterms have significantly lower scores

compared to fullterms both in grammatical (MLU, omissions, total morphological errors)

and cognitive (IQ) measures. These results are consistent with the few studies which have

investigated specific aspects of language development in preterms at preschool age (Briscoe

et al., 1998; Le-Normand & Cohen, 1999). The specific and new contribution of the present

study is to show that MLU, which is a sensitive measure for assessing grammatical

development at this age, is an aspect of language affected by preterm birth. Furthermore, the

results show that all main word categories (nouns, verbs and function words) necessary for

producing complete sentences gave rise to significantly more morphological errors by

preterms than fullterms. In both preterms and fullterms morphological errors are constituted

mostly by omissions, whereas substitutions and errors in bounded morphemes are rare, and

function words are the most omitted word category followed by nouns and verbs. It can be

concluded that preterm birth affects the development of several grammatical categories and

not only of a specific one and that the types and categories of preterms’ mistaken words are

similar to those of fullterms, even if in a higher measure.

The second aim of our study was to examine the effects of biological (gestational age,

gender) and social factors (maternal and paternal level of education) on preschool

grammatical and cognitive abilities and investigate whether social factors may have a

compensatory role over neonatal biological risk factors. The results showed that the only

factor having an effect on preterms’ grammatical and cognitive development was maternal

level of education, while no effects were found for the other social factor (paternal level of

education) nor for biological factors (gestational age, gender). Maternal level of education

appeared thus to be a relevant factor for grammatical and cognitive development at 3.5 years

showing to partially compensate the effects of a preterm birth, as other authors found

examining cognitive and linguistic development in at low risk 3-year-old preterms (Molfese

et al., 1994). The effect of the maternal level of education emerged in the preterm sample but

not in the fullterm one. A possible explanation for this result may lie in the fact that in

fullterms this social factor has an effect on grammatical and cognitive development in the

case of more complex tasks. It cannot be excluded that social factors may affect other aspects

of preterms’ and fullterms’ language development not examined in the present study, such as

lexical competence or spontaneous language samples (Dollaghan et al., 1999).

The differences found between preterms and controls in both grammatical and cognitive

abilities raise the question of whether other abilities connected to grammatical

development and supporting it, are impaired by very immature preterm birth. To further

investigate this question a second study was planned. In particular, some authors have

hypothesized that language development is supported by phonological working memory

A. Sansavini et al. / Journal of Communication Disorders 40 (2007) 239–256246

development, whose function is to temporary retain and rehearse verbal material, and that

phonological working memory skills and vocabulary knowledge are reciprocally

facilitative (Gathercole & Baddeley, 1993). According to these authors, for as long as

speech production and comprehension have not become automatic processes, phonological

working memory may help short- and long-term storage of new words and syntactic

structures. We thus hypothesize that phonological working memory plays a role on the

development of preterms’ grammatical abilities.

3. Study 2. Phonological working memory, grammar and their relations in very

immature preterms

Relations between phonological working memory and some aspects of language

development have been studied at preschool and school age in typical, at risk – preterms

and twins –, and atypical populations – Williams Syndrome (WS) and specific language

impairment (SLI) children (Adams & Gathercole, 1995; Bishop, North, & Donlan, 1996;

Briscoe et al., 1998; Gathercole & Baddeley, 1990; Grant et al., 1997). A study examining

3-year-old typical (fullterm) children through both phonological working memory tests

(non-word repetition and auditory digit span) and spontaneous speech, showed that

children with good phonological working memory abilities produced a richer array of

words and longer and more complex utterances than children with poorer phonological

working memory abilities (Adams & Gathercole, 1995). Another study examining 3–4-

year-old preterms found that a subgroup was at risk for both language production and

comprehension and phonological working memory (non-word repetition and auditory digit

span) (Briscoe et al., 1998). Other studies showed that SLI children, at school age, have a

deficit in phonological working memory (non-word repetition) and confirmed the

hypothesis that phonological working memory supports the development of language

(Bishop et al., 1996; Gathercole & Baddeley, 1990; Speidel, 1993). Research on WS

children, with two control groups (one matched for non-verbal test age and the other for

verbal test age), showed that phonological working memory is relatively unimpaired in WS

and strongly related to vocabulary knowledge (Grant et al., 1997).

The results of the above studies suggest that children with linguistic difficulties have

deficits in phonological working memory abilities. However, while relations between

phonological working memory and vocabulary acquisition, language comprehension and

speech production at preschool age have been investigated in several studies, relations

between phonological working memory and grammatical development have rarely been

examined at preschool age. Only the study by Adams and Gathercole (1995) on typically

developing children showed a specific role of phonological working memory in learning new

grammatical structures, besides supporting the storage of vocabulary and the production of

utterances. The relation between phonological working memory and grammar development

should thus be further investigated in at risk and atypical populations.

The first aim of this second study was thus to investigate the effects of a very immature

preterm birth on phonological working memory abilities, besides grammatical and

cognitive ones. Differences on these abilities were expected in preterm children compared

to fullterms. The second aim was to investigate the relations between phonological

A. Sansavini et al. / Journal of Communication Disorders 40 (2007) 239–256 247

working memory and grammatical abilities. If, as hypothesized by several authors quoted

above, these abilities are interdependent in their development, significant relations among

them should hold both in preterm and fullterm children. In particular, it was hypothesized

that phonological working memory helps the mastery of function words which, in several

languages such as Italian, begin to be learned after nouns and verbs and make sentences

complete. Consequently, it was predicted that children with less developed phonological

working memory abilities will have lower grammatical abilities (MLU).

3.1. Method

3.1.1. Participants

Sixty-two monolingual Italian preterms took part in a follow-up study which was

conducted at the Institute of Neonatology and Pediatrics of Bologna University. Birth dates

ranged from October 1997 to November 1999. The preterm children employed in this

second study also participated in Study 1.

The mean gestational age at birth of the preterm group was 30.3 weeks (S.D. = 2.2),

with a range from 25 to 33 weeks. Twenty-five had a gestational age < 31 weeks and 37

had �31 weeks. Their mean birthweight was 1202.7 g (S.D. = 276.6), with a range from

550 to 1630 g. Seventeen children weighed at birth �1000 g and 45 weighed >1000 g.

Thirty-three were males and 29 females. Parental educational level was distributed as

follows. Fourteen mothers had a low educational level (basic), 30 a medium level (high

school) and 18 a high level (University). Twenty-three fathers had a low educational level

(basic), 24 a medium level (high school) and 15 a high level (University). Preterm children

were tested at 3.5 years corrected age (M = 42.4 months, S.D. = .8, range = 41–45).

A comparison group of 28 monolingual Italian fullterms was recruited who also

participated in Study 1. Eleven were males and 17 females. Three mothers had a low

educational level (basic), 14 a medium level (high school) and 11 a high level (University).

Seven fathers had a low educational level (basic), 10 a medium level (high school) and 11 a

high level (University). The fullterm children were tested at 3.5 years (M = 42.2 months,

S.D. = .8, range = 42–45).

A Chi-square analysis indicated that the preterm and the control sample did not differ by

gender [x2 (1, N = 90) = 1.5, p = .221], nor by maternal level of education [x2 (2,

N = 90) = 2.08, p = .354] nor by paternal level of education [x2 (2, N = 90) = 2.41,

p = .299]. An independent samples t-test indicated that the preterm and the fullterm group

were equivalent in terms of children’s age [t (88) = .9, p = .369].

3.1.2. Materials

3.1.2.1. Grammatical and cognitive abilities. The Italian test of repetition of phrases and

sentences – TRF (Devescovi & Caselli, 2001) and the Stanford–Binet Intelligence Scale

(Bozzo & Mansueto Zecca, 1968) were used as in Study 1.

3.1.2.2. Phonological working memory abilities. Two Italian phonological working

memory tests were administered to each child: a test of non-word repetition (test di

ripetizione di non parole, Ciccarelli, 1998) and an auditory word span (test di span,

Ciccarelli, 1998).

A. Sansavini et al. / Journal of Communication Disorders 40 (2007) 239–256248

The test of non-word repetition consists of 40 non-words, 10 each containing 2 (e.g., mipo,

dabi), 3 (e.g., lumapa, bidana), 4 (e.g., peparoni, cotiboda) and 5 (e.g., napolebana,

ralebonuba) syllables. All non-words were stressed on the penultimate syllable, which is the

more frequent stress pattern in Italian words. The test was presented as a game and the child

was told that he/she would hear some funny words which he/she should try to repeat. Non-

words were spoken aloud by the experimenter at a constant rate, as done in other studies on

young children (Briscoe et al., 1998). The stimuli were presented in a randomized sequence

for all children. Responses were scored as incorrect if the child produced phonemic

differences from the target non-word. When a child consistently misarticulated a phoneme in

spontaneous speech (e.g., /s/ or /r/), the misarticulation of the target phoneme in a non-word

was scored as correct, since it did not depend on phonological working memory abilities.

Total correct repetitions, phonemic errors and omissions were scored. The number of non-

words correctly repeated at each of the four syllable lengths was also scored.

The auditory word span consists of lists of words (names of animals with a high

frequency in the Italian language) of increasing length (two, three, four words). The

experimenter spoke aloud the lists at a constant rate and asked to repeat each list

immediately after it was presented. Some examples of lists of words were: at length 2,

corvo, delfino, ‘crow, dolphin’; at length 3, pecora, rana, toro, ‘sheep, frog, bull’; at length

4, pecora, tigre, topo, balena, ‘sheep, tiger, mouse, whale’. Three lists of each length were

given, starting at length two. If the child repeated correctly two lists out of three, the length

of the next list was increased by one. If the child failed to repeat correctly two lists out of

three, testing stopped. Span was scored as the maximum length at which the child correctly

repeated at least two lists.

3.1.3. Procedure

The procedure was the same used in Study 1. Two independent observers, previously

trained, made the transcriptions of children’s repetitions of the sentences of the TRF, of the

non-words of the test of repetition and of the list of words of the auditory word span and

coded them.

3.2. Results

3.2.1. Differences between preterms and controls in phonological working memory,

grammatical and cognitive abilities

Descriptive statistics for preterm and control samples are shown in Table 2. All children

completed the TRF and the Stanford–Binet Intelligence Scale. Instead, 19 out of 62

preterms and 2 out of 28 fullterms did not complete non-word repetition (i.e., they did not

repeat at least 70% of the total number of non-words). Ten out of 62 preterms and 3 out of

28 fullterms did not complete the auditory word span (i.e., they failed to repeat correctly

two lists out of three of the length two list). To compare the preterms and fullterms not

completing the phonological working memory tests, a Chi-square analysis was run. This

revealed that more preterms (30% of the sample) did not complete non-word repetition

than fullterms (7% of the sample) [x2 (1, N = 90) = 5.96, p = .015]. No significant

difference was found in auditory word span, although 16% of the preterms and 11% of the

fullterms did not complete the test ( p = .373, Fisher’s exact test).

A. Sansavini et al. / Journal of Communication Disorders 40 (2007) 239–256 249

To investigate the differences between the preterm and fullterm sample, one-way

ANOVAs were performed using group membership (preterm and fullterm) as the between-

subjects factor. Measures of phonological working memory – non-word repetition –

(omissions, phonemic errors, total correct repetitions, correct repetitions of 2-, 3-, 4- and 5-

syllable non-words), – auditory word span- (word span), grammar – TRF (MLU), and

cognitive abilities (IQ) were used as the dependent variables (see Table 2). Differences

were found in phonological working memory abilities. In particular, differences were

found in omissions [F(1, 67) = 7.41, p = .008] and in correct repetitions of 4-syllable non-

words [F(1, 67) = 5.24, p = .025]. A tendency in the total correct repetitions [F(1,

67) = 3.57, p = .063] and in correct repetitions of 5-syllable non-words [F(1, 67) = 3.7,

p = .059] was also found. In phonemic errors, in correct repetitions of 2- and 3-syllable

non-words and in the auditory word span, no differences were found. Significant

differences between preterms and fullterms were also found in the MLU [F(1, 88) = 7.52,

p = .007] and in the IQ [F(1, 88) = 17.69, p < .001].

3.2.2. Relations between phonological working memory and grammatical abilities

To investigate the relations between phonological working memory (non-word

repetition) and grammatical abilities both in preterms and in fullterms a Pearson product

moment correlation test was used. With regard to the preterm sample, significant

A. Sansavini et al. / Journal of Communication Disorders 40 (2007) 239–256250

Table 2

Scores in grammatical abilities (TRF), phonological working memory abilities and IQ in preterm and control

(fullterm) children (Study 2)

Preterms Fullterms

Mean S.D. Range Mean S.D. Range

TRF

MLU 3.7 (N = 62) .9 1.3–4.7 4.2** (N = 28) .7 1.9–4.7

Phonological working memory non-word repetition

Total correct repetitions

(2-syllable + 3-syllable

+ 4-syllable + 5-syllable

non-words)

28.5 (N = 43) 6.6 9–39 31.3^ (N = 26) 4.9 20–39

Phonemic errors 6.6 (N = 43) 4.2 0–17 7.6 (N = 26) 4.4 1–16

Omissions 4.9 (N = 43) 6 0–20 1.4** (N = 26) 3.2 0–14

Correct 2-syllable

non-words

9.1 (N = 43) 1.3 5–10 9.3 (N = 26) 1 7–10

Correct 3-syllable

non-words

8.5 (N = 43) 1.6 3–10 8.4 (N = 26) 1.3 5–10

Correct 4-syllable

non-words

6.6 (N = 43) 2.2 1–10 7.7* (N = 26) 1.4 5–10

Correct 5-syllable

non-words

4.3 (N = 43) 3.4 0–10 5.8^ (N = 26) 2.6 0–10

Auditory word span

Word span 2.6 (N = 52) .6 2–4 2.7 (N = 25) .6 2–4

IQ 104.4 (N = 62) 10.5 81.8–120.9 114.1*** (N = 40) 9.1 90.5–130.9

One-way ANOVA test. ^p < .07; *p < .05; **p < .01; ***p < .001.

correlations were found between MLU and omissions [r(41) = �.57, p < .001], total

correct repetitions [r(41) = .67, p < .001], correct repetitions of 2-syllable [r(41) = .46,

p = .002], 3-syllable [r(41) = .50, p = .001], 4-syllable [r(41) = .58, p < .001] and 5-

syllable [r(41) = .51, p = .001] non-words. With regard to the fullterm sample, significant

correlations were found between MLU and omissions [r(24) = �.56, p = .003] and correct

repetitions of 5-syllable non-words [r(24) = .58, p = .002].

To investigate whether both preterms and fullterms with more difficulties in the auditory

word span also had poorer grammatical abilities (MLU), 2 one-way ANOVAs were run

(one for the preterm sample and one for the fullterm sample). Auditory word span at four

levels (no span versus 2-word span versus 3-word span versus 4-word span) was the

independent variable, while MLU was used as the dependent variable. A significant main

effect of phonological working memory ability was found both in preterms and fullterms

(see Fig. 1).

In particular, children who did not complete the task (no span) had a significantly lower

MLU than those who completed the task [preterms: F(3, 58) = 12.07, p < .001; fullterms

F(3, 24) = 8.79, p < .001] (see Fig. 1). Comparisons among the means using the

Bonferroni procedure were computed to establish which groups differed significantly from

one another. Preterms who did not complete the auditory word span had a lower MLU

(M = 2.6) than preterms with a 2-word span (M = 3.6, p = .001), 3-word span (M = 4.1,

p < .001) and 4-word span (M = 4.2, p = .002). In the same way, fullterms who did not

complete the auditory word span had a lower MLU (M = 2.9) than fullterms with a 2-word

span (M = 4.3, p = .001), 3-word span (M = 4.4, p < .001) and 4-word span (M = 4.5,

p = .005).

3.3. Discussion

The first aim of this study was to investigate whether very immature preterm birth

affects phonological working memory abilities, beyond grammatical and cognitive ones.

The results show that preterms had lower scores in comparison with fullterms in the non-

word repetition task and in grammatical (MLU) and cognitive (IQ) tests. In particular, they

A. Sansavini et al. / Journal of Communication Disorders 40 (2007) 239–256 251

Fig. 1. Preterms’ and fullterms’ MLU score (mean length of utterance) in function of their auditory word span.

encountered difficulties in coping with non-word repetition. Indeed, 30% of the preterms in

comparison with 7% of the fullterms did not complete this test. Furthermore, among

children who completed this test, preterms omitted more non-words and made fewer

correct repetitions in comparison with fullterms, specifically with regard to the longer (4-

and 5-syllable) non-words. Increased word length resulted in a performance decrease in

non-word repetition in both groups, as found in other studies comparing normal and SLI

children (Marton & Schwartz, 2003), but preterms encountered more difficulties in the

longer non-words than fullterms. By contrast, the other phonological working memory test,

the auditory word span, yielded no differences between preterms and fullterms and the

range of the auditory word span showed less variability in both samples (most of the

children had a two-word or three-word length span). These results thus show that preterm

birth affects phonological working memory abilities, except auditory span. This happens

probably because the latter is composed by names of animals, which normally are already

part of the lexicon of a child of this age, while the test of repetition includes non-words

which cannot be supported by lexical competence. It thus seems that preterms have

difficulties when a higher competence of phonological working memory is required and no

lexical support can be called upon as it was found in studies on SLI children (Gathercole &

Baddeley, 1990; Marton & Schwartz, 2003).

The second aim of this study was to investigate whether phonological working memory

and grammatical abilities are interrelated in preterms. Significant relations were found

between these abilities both in preterm and fullterm children: MLU correlates with

omissions of the non-word repetition test in both groups, with the total and the 2-, 3-, 4-, 5-

syllable correct repetitions in preterms, and with the 5-syllable correct repetitions in

fullterms. Furthermore, preterms and fullterms with less developed phonological working

memory abilities had lower grammatical abilities (MLU). Indeed, in both groups those few

children who did not complete the auditory word span had a lower MLU than those who

completed the task. Therefore, in both preterms and fullterms, there were strict relations

between phonological working memory and grammatical abilities. These results are in

agreement with those found in fullterms’ spontaneous speech by Adams and Gathercole

(1995). The novel contribution of the present study is to show that phonological working

memory skills and grammatical abilities are interrelated and therefore reciprocally

facilitative both in fullterms’ and in preterms’ development.

4. General discussion

Considering the results of our present and past studies, it may be affirmed that very

immature preterm birth may cause slight but persisting difficulties in linguistic, cognitive

and phonological working memory development up to 3.5 years. Differently from 2.5 years

(Sansavini et al., 2004, 2006), not only those preterms with a higher biological risk

(gestational age < 31 weeks) but the total sample has less advanced grammatical and

cognitive development in comparison with fullterms, even if preterms’ mean scores still

fall within the normal range, as already found in other studies (Anderson & Doyle, 2003;

Sansavini et al., 1996, 2006). Considering that preterms were examined at the corrected

age, both at 2.5 and 3.5 years, and therefore their neurobiological development was taken

A. Sansavini et al. / Journal of Communication Disorders 40 (2007) 239–256252

into account, our data suggest that the gap between preterms’ and fullterms’ linguistic and

cognitive development persists and becomes more evident in the fourth year of life. This

happens probably because in typical development a consolidation of grammatical abilities

and a decrease of interindividual variability in the rhythm of linguistic development occurs

with respect to the preceding years. By contrast, this variability remains large (in terms of

MLU, IQ, omissions and errors in grammatical and phonological working memory tasks)

in preterms. Whether this condition of risk in preterms depends on subtle differences in

cortical and synaptic development remains to be examined with studies using both

behavioral and sophisticate physiological measures, such as magnetic resonance imaging.

Even if preterm birth has a persistent effect on linguistic, cognitive and phonological

working memory abilities up to 3.5 years, no singular biological factor yielded a causal

effect by itself, in contrast to earlier ages when gestational age and gender had specific

effects on grammatical and cognitive development (Sansavini et al., 2004, 2006). Instead, a

social factor, maternal level of education, which did not show any influence at 2.5 years,

becomes relevant at 3.5 years partially compensating for the effects of preterm birth on

language and cognitive development in agreement with other studies (Molfese et al., 1994;

Sansavini et al., 1996).

With regard to the relation between phonological working memory and language, the

results of our study show that preterms’ phonological working memory and grammatical

development, although slower than those of fullterms, maintain a strict relation mirroring

what happens in typical development. Our results are in line with the hypothesis by

Gathercole and Baddeley (1993) of a reciprocal influence between phonological working

memory and language acquisition and show that both in fullterms and in preterms, for as

long as speech production and comprehension have not yet become automatic processes,

phonological working memory may help short- and long-term storage not only of new

words (lexical competence) but also of syntactic structures and of the length of utterance

(grammatical competence).

Finally, the methodological contribution of our study is to show that ecologically

structured tasks can be used at preschool age to assess grammatical and phonological

working memory abilities. In particular, the TRF proved to be very effective in assessing

MLU and evaluating preterms’ and fullterms’ grammatical competence as well as the test

of non-word repetition proved to be sensitive in highlighting phonological working

memory skills.

Several questions remain for future research. A longitudinal study could address

whether at risk development continues to characterize preterms at even later ages and

whether the tight relation found between linguistic and phonological working memory

abilities (particularly when using non-words) persists. Furthermore, the effects of

biological and social factors need to be further investigated at later ages, since no

agreement exists regarding the role of compensation of social factors at school age. At the

same time, the results hitherto obtained suggest that an early evaluation (before 4 years) of

preterms’ linguistic, cognitive and phonological working memory abilities should be done

in order to identify those children presenting with more difficulties and to begin an

intervention at preschool age. Such an intervention could be particularly effective in this

specific developmental phase both for recovering and preventing the development of more

severe difficulties subsequently at school age.

A. Sansavini et al. / Journal of Communication Disorders 40 (2007) 239–256 253

Acknowledgements

This research was supported by research grants from the University of Bologna (Basic

Oriented Research ex 60% 2001; University Research Projects 1998–2000 and 2004–

2006). We would like to thank Mara Armaroli, Francesca Ruffilli and Silvia Savini for their

help with the data collection; Silvia Galletti, Laura Malaigia and Vittoria Paoletti, for their

help with the medical examination; the children and parents for their participation in the

research. We would also like to thank Maria Cristina Caselli and Antonella Devescovi for

their precious suggestions concerning the TRF and Laura Ciccarelli for the tests of

phonological working memory. Finally, we are grateful to Annette Karmiloff-Smith for her

careful comments and helpful suggestions.

References

Adams, A. M., & Gathercole, S. E. (1995). Phonological working memory and speech production in preschool

children. Journal of Speech and Hearing Research, 38, 403–414.

Ajayi-Obe, M., Saeed, N., Cowan, F. M., Rutherford, M. A., & Edwards, A. D. (2000). Reduced development of

cerebral cortex in extremely preterm infants. Lancet, 356, 1162–1163.

Als, H. (1992). Individualized, family-focused developmental care for the very-low-birthweight preterm infant in

NICU. In S. Friedman, & M. D. Sigman (Eds.), The psychological development of low birthweight children

(pp. 341–388). Norwood, NJ: Ablex Publishing Corporation.

Anderson, P., & Doyle, L. W. (2003). Neurobehavioral outcomes of school-age children born extremely

low birth weight or very preterm in the 1990s. JAMA, The Journal of American Medical Association, 289,

3264–3272.

Aram, D., Hack, M., Hawkins, S., Weissman, B., & Borawski-Clark, E. (1991). Very low birth weight children and

speech and language development. Journal of Speech and Hearing Research, 34, 1169–1179.

Barsky, V. E., & Siegel, L. (1992). Predicting future cognitive, academic and behavioral outcomes for very-low-

birthweight infants. In S. L. Friedman, & M. D. Sigman (Eds.), The psychological development of low

birthweight children (pp. 275–298). Norwood, NJ: Ablex Publishing Corporation.

Bates, E., & Goodman, J. (1997). On the inseparability of grammar and the lexicon: Evidence from acquisition.

Language and Cognitive Processes, 12, 507–584.

Beek van, Y., Hopkins, B., & Hoeksma, J. B. (1994). Development of communicative behaviors in preterm infants:

The effects of birthweight status and gestational age. Infant Behavior and Development, 17, 107–117.

Bishop, D. V. M., North, T., & Donlan, C. (1996). Non word repetition as a behavioural marker for inherited

language impairment: Evidence from a twin study. Journal of Child Psychology and Psychiatry, 37,

391–403.

Bortolini, U., & De Gasperi, L. (2002). Diagnosi e trattamento differenziali dei bambini con disturbo specifico del

linguaggio. In S. Vicari, & M. C. Caselli (Eds.), I disturbi dello sviluppo (pp. 77–98). Bologna: Il Mulino.

Bozzo, M. T., & Mansueto Zecca, G. (1968). Scala di intelligenza Standford–Binet, forma L–M (III revisione di

Terman-Merrill). Firenze: Organizzazioni Speciali.

Briscoe, J., Gathercole, S. E., & Marlow, N. (1998). Short-term memory and language outcomes after extreme

prematurity at birth. Journal of Speech, Language and Hearing Research, 41, 654–666.

Byers-Brown, B., Bendersky, M., & Chapman, T. (1986). The early utterances of preterm infants. British Journal

of Communication Disorder, 21, 307–319.

Cheour-Luhtanene, M., Alho, K., Sainio, K., Rinne, T., Reinikainen, K., Pohjavuori, M., et al. (1996). The

ontogenetically earliest discriminative response of the human brain. Psychophysiology, 33, 478–481.

Chilosi, A. M., Cipriani, P., & Fapore, T. (2002). I disturbi specifici del linguaggio. In S. Vicari, & M. C. Caselli

(Eds.), I disturbi dello sviluppo (pp. 59–76). Bologna: Il Mulino.

Ciccarelli, L. (1998). Comprensione del linguaggio, processi di elaborazione e memoria di lavoro: uno studio in

eta prescolare. Unpublished doctoral dissertation. Italy: University of Padova.

A. Sansavini et al. / Journal of Communication Disorders 40 (2007) 239–256254

Counsell, S. J., Rutherford, M. A., Cowan, F. M., & Edwards, A. D. (2003). Magnetic resonance imaging of

preterm brain injury. Archives of Disease in Childhood Fetal and Neonatal Edition, 88, F269–F274.

DeCasper, A. J., Lecanuet, J. P., Busnel, M. C., Granier-Deferre, C., & Maugeais, R. (1994). Fetal reactions to

recurrent maternal speech. Infant Behavior and Development, 17, 159–164.

Devescovi, A., & Caselli, M. C. (2001). Una prova di ripetizione di frasi per la valutazione del primo sviluppo

grammaticale. Psicologia Clinica dello Sviluppo, 5, 341–364.

Dollaghan, C. A., Campbell, T. F., Paradise, J. L., Feldman, H. M., Janosky, J. E., Pitcairn, D. N., et al. (1999).

Maternal education and measures of early speech and language. Journal of Speech, Language and Hearing

Research, 42, 1432–1443.

Elman, J. L., Bates, E. A., Johnson, M. H., Karmiloff-Smith, A., Parisi, D., & Plunkett, K. (1996). Rethinking

innateness. Cambridge: The MIT Press.

Ferrari, P., Sturloni, N., & Cavazzuti, G. B. (1982). La maturazione dell’attivita bioelettrica e dei parametri

fisiologici (movimenti corporei, rapidi movimenti oculari, tono muscolare, attivita cardiaca e respiratoria) che

definiscono le fasi del sonno nel feto e nel neonato. In M. Bertolini (Ed.), La nascita psicologica e le sue

premesse neurobiologiche (pp. 57–113). Roma: YES Mercury.

Gallagher, T. M., & Watkin, K. L. (1998). Prematurity and language developmental risk: Too young or too small?

Topics in Language Disorders, 18(3), 15–25.

Gathercole, S. E., & Baddeley, A. D. (1990). Phonological memory deficits in language disordered children: Is

there a causal connection. Journal of Memory and Language, 9, 336–360.

Gathercole, S. E., & Baddeley, A. D. (1993). Working memory and language. Hove: Erlbaum Associates.

Giovanelli, G., Sansavini, A., & Farneti, A. (1999). Perception of sound, rhythm and speech from pre-natal to post-

natal life. In A. F. Kalverboer, M. L. Genta, & J. B. Hopkins (Eds.), Current issues in developmental

psychology. Biopsychological perspectives (pp. 137–159). Dordrecht: Kluwer Academic Publishers.

Grant, J., Karmiloff-Smith, A., Gathercole, S. A., Paterson, S., Howlin, P., Davies, M., et al. (1997). Phonological

short-tem memory and its relationship to language in Williams Syndrome. Cognitive Neuropsychiatry, 2, 81–89.

Grunau, R. V., Kearney, S. M., & Whitfield, M. F. (1990). Language development at 3 years in pre-term children of

birth weight below 1000 g. British Journal of Disorders of Communication, 25, 173–182.

Hindmarsh, G. J., O’Callaghan, M. J., Mohay, H. A., & Rogers, Y. M. (2000). Gender differences in cognitive

abilities at 2 years in ELBW infants. Early Human Development, 60, 115–122.

Kisilevsky, B. S., Hains, S. M. J., & Low, J. A. (1999). Differential maturation of fetal responses to vibroacoustic

stimulation in a high risk population. Developmental Science, 2, 234–245.

Largo, R. H., Molinari, L., Comenale Pinto, L., Weber, M., & Duc, G. (1986). Language development of term and

preterm children during the first five years of life. Developmental Medicine and Child Neurology, 28, 333–350.

Le-Normand, M. T., & Cohen, H. (1999). The delayed emergence of lexical morphology in preterm children: The

case of verbs. Journal of Neurolinguistics, 12, 235–246.

Maalouf, E. F., Duggan, P. J., Rutherford, M. A., Counsell, S. J., Fletcher, A. M., Battin, M., et al. (1999).

Magnetic resonance imaging of the brain in a cohort of extremely preterm infants. Journal of Pediatrics, 135,

351–357.

Marton, K., & Schwartz, R. G. (2003). Working memory capacity and language processes in children with specific

language impairment. Journal of Speech, Language and Hearing Research, 46, 1138–1153.

Menyuk, P., Liebergott, J. W., & Schultz, M. S. (1995). Early language development in full-term and premature

infants. Hillsdale, NJ, Hove: Lawrence Erlbaum Associates.

Molfese, V. J., Holcomb, L., & Helwig, S. (1994). Biomedical and social–environmental influences on cognitive

and verbal abilities in children 1 to 3 years of age. International Journal of Behavioral Development, 17, 271–

287.

Oller, D. K., Eilers, R. E., Steffens, M. L., & Lynch, M. P. (1994). Speech-like vocalizations in infancy: An

evaluation of potential risk factors. Journal of Child Language, 21, 33–58.

Sansavini, A., Guarini, A., Alessandroni, R., Faldella, G., Giovanelli, G., & Salvioli, G. P. (2006). Early relations

between lexical and grammatical development in very immature Italian preterms. Journal of Child Language,

33, 199–216.

Sansavini, A., Guarini, A., Ruffilli, F., Alessandroni, R., Giovanelli, G., & Salvioli, G. P. (2004). Fattori di rischio

associati alla nascita pretermine e prime competenze linguistiche rilevate con il MacArthur. Psicologia Clinica

dello Sviluppo, 8, 47–67.

A. Sansavini et al. / Journal of Communication Disorders 40 (2007) 239–256 255

Sansavini, A., Rizzardi, M., Alessandroni, R., & Giovanelli, G. (1996). The development of Italian low and very-

low-birthweight from birth to five years: The role of biological and social risks. International Journal of

Behavioral Development, 19, 533–547.

Speidel, G. E. (1993). Phonological short-term memory and individual differences in learning to speak. A

bilingual case study. First Language, 13, 69–91.

Taylor, H. G., Klein, N., Minich, N. M., & Hack, M. (2000). Middle-school-age outcomes in children with very

low birthweight. Child Development, 71, 1495–1511.

Vicari, S., Caravale, B., Carlesimo, G. A., Casadei, A. M., & Allemand, F. (2004). Spatial working memory

deficits in children at ages 3–4 who were low birthweight preterm infants. Neuropsychology, 18, 673–678.

Vohr, B. R., Garcia Coll, C. T., & Oh, W. (1988). Language development of low-birthweight infants at two years.

Developmental Medicine and Child Neurology, 30, 608–615.

Volpe, J. J. (2001). Neurology of the newborn. Philadelphia: WB Saunders.

Wolke, D., & Meyer, R. (1999). Cognitive status, language attainment, and prereading skills of 6-year-old very

preterm children and their peers: The Bavarian Longitudinal Study. Developmental Medicine and Child

Neurology, 41, 94–109.

A. Sansavini et al. / Journal of Communication Disorders 40 (2007) 239–256256