Input effects across domains: The case of Greek subjects in child heritage language

Abstract

A recurring question in the literature of heritage language acquisition, and more generally of bilingual acquisition, is whether all linguistic domains are sensitive to input reduction and to cross-linguistic influence and to what extent. According to the Interface Hypothesis, morphosyntactic phenomena regulated by discourse–pragmatic conditions are more likely to lead to non-native outcomes than strictly syntactic aspects of the language (Sorace, 2011). To test this hypothesis, we examined subject realization and placement in Greek–English bilingual children learning Greek as a heritage language in North America and investigated whether the amount of heritage language use can predict their performance in syntax–discourse and narrow syntactic contexts. Results indicated two deviations from the Interface Hypothesis: First, subject realization (a syntax–discourse phenomenon) was found to be largely unproblematic. Second, subject placement was affected not only in syntax–discourse structures but also in narrow syntactic structures, though to a lesser degree, suggesting that the association between the interface status of subject placement and its sensitivity to heritage language use among children heritage speakers is gradient rather than categorical.

Keywords

child heritage language acquisition heritage language use input and output effect Interface Hypothesis narrow syntax subject use in Greek syntax–discourse interface

I Introduction

Heritage speakers are early bilinguals whose first language, the heritage language they speak at home with their parents and siblings, is different from the majority/societal language of the community they live in (Valdés, 2000). Because they acquire their heritage language under reduced input conditions and under the influence of the dominant societal language, their performance in their heritage language is typically characterized by variability (Benmamoun et al., 2013; Kupisch and Rothman, 2018; Montrul, 2008, 2015; Polinsky and Kagan, 2007; Rothman, 2009b; Scontras et al., 2015). A recurring question in the literature of heritage language acquisition, and more generally of bilingual acquisition, is whether all linguistic domains are sensitive to input reduction and to cross-linguistic influence and to what extent (Benmamoun et al., 2013; Montrul, 2015; Rothman 2009b; Scontras et al., 2015). To address this question, Sorace and colleagues have proposed the Interface Hypothesis (IH), an influential account within generative approaches to language acquisition, which predicts that morphosyntactic phenomena regulated by discourse–pragmatic conditions are more likely to lead to non-native outcomes than strictly syntactic aspects of the language (see Sorace, 2006, 2011, 2012; Sorace and Filiaci, 2006; Sorace and Serratrice, 2009; Tsimpli and Sorace, 2006; for an overview and critical discussion of the IH, see Montrul, 2011; White, 2011). The predictions of this hypothesis have been tested against different types of bilingual populations (Sorace, 2011), including heritage speakers (Montrul and Polinsky, 2011), with some studies supporting the inherent complexity of the syntax–discourse domain (see, amongst others, Belletti et al., 2007; Flores, 2012; Montrul, 2004; Sorace and Filiaci, 2006; Tsimpli et al., 2004) and others indicating that syntax–discourse phenomena are not predetermined areas for fossilization (see, amongst others, Domínguez, 2013; Donaldson, 2012; Judy, 2015; Judy and Rothman, 2014; Leal et al., 2014; Montrul and Rodríguez-Louro, 2006; Rothman, 2009a).

However, the majority of these studies have been designed to determine the differential effect of cross-linguistic influence across domains. The potentially differential effect of input quantity has only been examined in an indirect way through comparisons between bilingual populations with different amounts of exposure to the target language. For instance, Argyri and Sorace (2007) and Sorace et al. (2009) compared bilinguals of the same language combination acquiring the target language in two different contexts: in a minority context, where the amount of exposure to the target language is arguably limited, and in a majority context, where the amount of exposure to the target language is arguably greater. Accordingly, Kupisch and colleagues carried out a series of studies with heritage speakers (HSs) of French (Kupisch et al., 2013, 2014) and heritage speakers (HSs) of Italian (Kupisch, 2012, 2014) residing in Germany. Their results showed that the French HSs outperformed the Italian HSs, an outcome that was attributed to the fact that the first group had attended a French school during childhood, as opposed to the second group, which had attended a German school. As a result, HSs of French had quantitatively and qualitatively more opportunities to converge with native standards in their heritage language (for an overview of these studies, see Kupisch and Rothman, 2016).

The findings of these studies are particularly significant as they demonstrate the role of input quantity in heritage language acquisition through between-groups comparisons. To our knowledge, though, no study to date has examined the differential effect of input quantity, as a continuous variable, within the same group of heritage speakers. Furthermore, there are fewer studies on child than on adult heritage language (Cuza, 2016; Jia and Paradis, 2015) and only a handful of studies has examined Greek as a heritage language (Dosi, 2017; Kaltsa et al., 2015; Zombolou, 2011). To address this gap, in our study we examined subject realization and placement in Greek–English bilingual children learning Greek as a heritage language in North America and investigated whether the amount of heritage language use can predict their performance in the syntax–discourse and narrow syntactic contexts.

II Background

1 The Interface Hypothesis

One of the most influential accounts of the vulnerability of syntax–discourse phenomena among bilingual speakers has been put forward by Sorace and colleagues and has come to be known as the Interface Hypothesis (Sorace and Filiaci, 2006). In its original formulation, the Interface Hypothesis assumed a bipartite division between syntax proper and interface syntax in general. Subsequently, Tsimpli and Sorace (2006) proposed a more fine-grained division between (1) internal interfaces, which involve operations between syntax and semantics, and (2) external interfaces, which involve the mapping between syntax and pragmatic conditions of contextual appropriateness. Both versions maintained that phenomena belonging to the syntax–discourse interface, due to the added difficulty involved in the integration of the multiple layers of information, are more vulnerable in situations of language contact than strictly syntactic phenomena.

2 Pronominal subject realization

Evidence in support of the vulnerability of the syntax–discourse interface has come primarily from studies on pronominal subject realization. According to generative approaches, the rich agreement morphology of the verbal paradigm in null subject languages syntactically licenses or allows for null (phonologically unexpressed) subjects (Rizzi, 1982). In these languages, the alternation between null and overt pronominal subjects is constrained by discourse factors, in that there is a preference for null pronouns to refer to topical subject antecedents and for overt pronouns to mark topic shift and contrastive focus (for Greek, see Kaltsa et al., 2015; Papadopoulou et al., 2015; Tsimpli et al., 2004; for Italian, see Belletti et al., 2007; Carminati, 2002; Sorace et al., 2009; for Spanish, see Alonso-Ovalle et al., 2002; see, also, for differences among Italian and Spanish, Filiaci, 2010). The Italian sentences in (1), taken from Carminati (2002), illustrate this point:

(1) a. Marta scrivera frequentemente a Piera quando era negli Stati Uniti

Marta wrote frequently to Piera when was in the United States

b. Marta_i scrivera frequentemente a Piera_j quando lei_*i/j era negli Stati Uniti

Marta wrote frequently to Piera when she was in the United States

Whereas the null pronominal subject of the embedded clause in (1a) is preferably associated with the subject/topic of the matrix sentence (i.e. with Marta), giving rise to a –Topic Shift reading, the overt pronominal subject lei ‘she’ in (1b) is preferably associated with the object of the matrix sentence (i.e. with Piera), giving rise to a +Topic Shift interpretation.

A large body of research has shown that, unlike monolinguals, who ultimately master both the syntactic and the discourse aspect of pronominal subject realization, Italian–English bilinguals (and, more generally, bilinguals learning a null subject language together with a non-null subject language) are often challenged by its pragmatic constraints. For example, Belletti et al. (2007), in line with Sorace and Filiaci (2006), found that near native speakers of Italian, under the influence of their first language (L1; English), were significantly more likely than native speakers of Italian to overextend the use and interpretation of overt pronouns in –Topic Shift contexts, where null pronouns are the pragmatically appropriate option. The pattern has been replicated with different types of bilingual populations speaking different combinations of null and non-null subject languages. These include L1 attriters (for Greek–English, see Tsimpli et al., 2004; for Greek–Swedish, see Kaltsa et al., 2015), bilingual children (for Spanish–English, see Paradis and Navarro, 2003; for Italian–English, see Serratrice, 2007; Serratrice et al., 2004; Sorace and Serratrice, 2009; for Hebrew-English, see Hacohen and Schaeffer, 2007; for Turkish–English, see Haznedar, 2007; for Italian–Dutch, see Pinto, 2006), L2 learners at the intermediate level (for Spanish–English, see Montrul and Rodríguez-Louro, 2006; Rothman, 2009a), and adult heritage speakers (for Greek–Swedish, see Kaltsa et al., 2015; for Russian–English, see Polinsky, 2007; for Spanish–English, see Montrul, 2004). Despite differences in age and language combinations, but not independently of proficiency, as demonstrated by Montrul and Rodríguez-Louro (2006) and Rothman (2009a), all these bilingual populations have been shown to exhibit non-native-like uses or judgements of overt pronouns in their null subject languages.

Most of these studies, though, have been designed to determine the impact of cross-linguistic influence across domains. Hence, they have focused on language combinations that differ systematically with respect to pronominal subject realization (i.e. null subject languages and non-null subject languages) and have shown that bilinguals tend to make pragmatically non-felicitous uses of the overlapping option (i.e. of overt subjects) in their null subject language. Interestingly, non-felicitous uses of null subjects have also been reported among these populations (Clements and Domínguez 2017; Domínguez, 2013; Montrul, 2004; Montrul and Rodríguez-Louro, 2006; Rothman, 2009a; Wolleb, 2013) suggesting that cross-linguistic transfer might not be the sole factor for non-convergence in the syntax–discourse domain. This possibility is further supported by studies showing that bilinguals speaking two null subject languages, with similar discourse conditions underlying pronominal subject realization, can also be challenged by the distribution of null and overt subjects in their L2 (for L1-Spanish–L2-Italian, see Bini, 1993; for L1-Farsi–L2-Spanish, see Judy 2015; Judy and Rothman, 2014; for L1-Greek–L2-Spanish, see Lozano, 2006; Margaza and Bel, 2006; for simultaneous Italian–Spanish bilinguals, see Sorace et al., 2009). In light of these results, other factors, including input quantity, have been hypothesized to modulate the vulnerability of the syntax–discourse domain (see, amongst others, Sorace, 2011; Sorace and Serratrice, 2009).

3 The role of input quantity on subject realization and placement

The effect of input quantity across domains has only been indirectly investigated through comparisons between groups of bilinguals of the same language combination, who, depending on their country of residency, acquired the target language either as a majority or as a minority language (Argyri and Sorace, 2007; Sorace et al., 2009).

Sorace et al. (2009), for instance, compared a group of Italian–English bilingual children living in the UK with a group of Italian–English bilingual children living in Italy. Focusing on pronominal subject realization in Italian (a phenomenon that, as discussed above, requires the integration of both narrow syntactic and discourse pragmatic knowledge), the authors tested the acceptability of subject pronouns (overt and null) in –Topic Shift and +Topic Shift contexts. Their results confirmed the sensitivity of syntax–discourse to input quantity, since bilinguals living in the UK, who arguably received more English input than Italian input, performed less accurately than bilinguals living in Italy, who received more input in Italian than in English.

In an earlier study, Argyri and Sorace (2007) had examined the role of input quantity by comparing Greek–English bilingual children born and raised in the UK with Greek–English bilingual children born and raised in Greece. Differently from Sorace et al. (2009), the authors tested the children’s use and comprehension of Greek subjects and objects in both syntax–discourse and narrow syntactic domains. This allowed them to investigate not only whether input reduction would affect the syntax–discourse interface (a domain that was hypothesized to be sensitive to input quantity), but also whether it would affect it more than the narrow syntactic context (a domain that was hypothesized to be relatively immune). The syntax–discourse domains included (1) pronominal subject realization in –Topic Shift contexts, where subjects are preferably null and (2) subject placement in wide focus contexts, where subjects are preferably post-verbal. The narrow syntactic domains concerned (1) subject placement in wh-embedded interrogatives, where subjects are always post-verbal (independently of discourse conditions), and (2) placement of object clitics, which are always preverbal (independently of discourse conditions). Their results confirmed the role of input quantity, since the bilinguals living in the UK (who had a greater exposure to English) performed less accurately in Greek than the bilinguals living in Greece (who had a greater exposure to Greek). Interestingly, though, input effects were attested in both discourse–syntax and narrow syntactic domains, in different degrees. Thus, whereas both embedded interrogatives and wide focus were affected, wide focus was affected to a greater extent. The findings of this study are particularly interesting because they appear to suggest that the difference between narrow syntactic and syntax–discourse phenomena with respect to their sensitivity to input quantity is gradient rather than categorical. In other words, instead of supporting a categorical dichotomy between an invulnerable syntactic domain and a vulnerable syntax–discourse domain, they indicate that both domains are vulnerable in different degrees.

4 Open issues

Summing up, while both Sorace et al. (2009) and Argyri and Sorace (2007) provide support for the differential impact of input quantity across domains, they do so by relying on between-groups comparisons. Less is known about the impact of input fluctuations among speakers that belong in the same group of bilinguals. The methodological advantage of this alternative approach is that by keeping the status of the target language/country of residence constant, we may better control for the impact of other experiential variables commonly differentiating majority from heritage/minority language learning. These include the richness of input (e.g. diversity of sources), the quality of input (e.g. amount of code switching in the target language input), as well as access to formal instruction (for an overview of the experiential variables associated with bilingual acquisition and development, see Unsworth, 2016, 2017). Additionally, by studying the amount of language use as a continuous predictor within the same group, we can investigate how it differentially affects the various language domains or structures within the same group of heritage speakers, and avoid setting arbitrary cut-offs to create distinct learner groups.

III Present study

To address this gap, in the present study, we collected production data from Greek–English bilingual children learning Greek as a heritage language in North America and examined whether the amount of heritage language use at home would be a predictor of the children’s performance in the domain of Greek subjects, in both discourse–syntax and narrow syntactic contexts.

The amount of heritage language use at home was calculated as the mean proportion of heritage language input and output to better capture the children’s experience with the heritage language, similarly to previous studies with bilingual children (Bedore et al., 2012; Paradis, 2011a). The domain of subjects was chosen because it has been shown to be problematic for other bilingual populations speaking Greek alongside a non-null subject language (for Greek–English simultaneous bilingual children living in the UK, see Argyri and Sorace, 2007; for adult heritage speakers of Greek living in Sweden, see Kaltsa et al., 2015; for Greek attriters, see Tsimpli et al., 2004). As to the selection of heritage speakers, it was motivated by the empirical observation that they differ significantly from each other in their daily amount of heritage language use (Montrul, 2015; Unsworth, 2017). As a result, they provide a good testing ground for hypotheses concerning how variation in input/output quantity affects different grammatical domains. Two research questions were addressed:

Do Greek heritage children differ from Greek monolingual children in their choices of subject form and subject placement as a function of syntactic and discourse factors?

Does the amount of heritage language use in the home affect children’s performance on subject form and placement? If so, are subjects at the syntax–discourse interface more sensitive to the amount of heritage language use than subjects constrained exclusively by syntactic factors?

To answer these questions, we replicated Argyri and Sorace’s (2007) design, in examining subject use in two syntax–discourse structures and one narrow syntactic structure. The syntax–discourse structures involved subject form, in –Topic Shift contexts, where subjects are preferably null, and subject placement in wide focus contexts, where subjects are preferably postverbal. The syntactic structure involved subject placement in wh-embedded interrogatives, where subjects are obligatorily postverbal and dissociated from discourse functions. In all three structures, subject use in Greek differs from subject use in English. These systematic differences enabled us to investigate the role of language use across domains, while keeping the likelihood of cross-linguistic transfer constant. In what follows, we will first present the syntactic literature and the properties of the target structures and we will then turn to our predictions.

1 Target structures

a Subject form in –Topic Shift (–TS) contexts

Being a typical null subject language with rich verbal agreement, Greek can leave the subject of a finite sentence phonologically unexpressed. Thus, overt pronominal subjects may alternate with null pronominal subjects, as shown in (2):

(2) a. Aftos efije.

he left.3SING

‘He left.’

b. pro efije.

left.3SING

‘He left.’

However, as discussed above, the alternation between these two forms is constrained by discourse factors, in that null subjects are preferably interpreted as coreferential with the most salient linguistic antecedent (that is with the subject/topic of a previous sentence), whereas overt pronominal subjects typically mark topic shift (for Greek, see Kaltsa, et al., 2015; Papadopoulou et al., 2015; Tsimpli et al., 2004). This is illustrated in (3).

(3) a. Pote efije o Kostas_i?

when left.3SING the Kostas.NOM

‘When did Kostas leave?’

b. Otan pro_i teljose tin omilia tu.

when finished.3SING the talk his.GEN

‘When he finished his talk.’

c. Otan aftos_j teljose tin omilia tu.

when he finished.3SING the talk his.GEN

‘When he finished his talk.’

Whereas the null pronominal subject (pro) in (3b) is associated with the subject of the previous utterance (i.e. Kostas) leading to a –Topic Shift interpretation, the overt pronominal subject (aftos ‘he’) in (3c) is most naturally understood as referring to an extra-linguistic entity, leading to a +Topic Shift interpretation.

In English, on the other hand, subjects are typically overt, except for some restricted contexts (Haegeman, 1997), and are used regardless of whether there is a shift of topic (4):

(4) a. When did Kostas_i leave?

b. * When pro finished his talk.

c. When he_i/j finished his talk.

b Subject placement in Wide Focus (WF)

The second syntax–discourse property that we considered concerned subject placement in finite declarative clauses. In this context, subjects may appear both pre-verbally, as shown in (5a) and post-verbally, as shown in (5b):

(5) a. O Kostas chalase ti fotoghrafiki.

the Kostas.NOM broke.3SING the camera.ACC

b. Chalase o Kostas ti fotoghrafiki.

broke.3SING the Kostas.NOM the camera.ACC

‘Kostas broke the camera.’

The alternation between preverbal and postverbal subjects though, similarly to the alternation between null and overt subjects, rather than being random, is regulated by discourse factors (Alexopoulou, 1999). To simplify a rather complex picture, Verb-Subject-Object (VSO) is the most natural answer to wide focus questions of the ‘What happened with X?’ type (Alexopoulou, 1999). This is shown in (6):¹

(6) a. Ti ejine me tin fotoghrafiki tu Petru?

what happened.3SING with the camera.GEN the Peter.GEN

‘What happened with Peter’s camera?’

b. tin chalase o Kostas.

CL.ACC broke.3SING the Kostas.NOM

‘Kostas broke it.’

c. # O Kostas ti chalase.

the Kostas.NOM CL.ACC broke.3SING

‘Kostas broke it.’

Subject-Verb-Object (SVO), on the other hand, obtains higher felicity judgements than Verb-Subject-Object (VSO) in all new contexts (Keller and Alexopoulou, 2011), as well as in contexts where the preverbal subject receives a topic or a narrow/contrastive focus interpretation (Alexiadou, 1999; Alexopoulou, 1999; Tsimpli, 1990, 1995). The latter possibility is illustrated with the dialogue in (7):

(7) a. Pjos chalase ti fotoghrafiki, o Janis i o Kostas?

who broke.3SING the camera.ACC, the Janis.NOM or the Kostas.NOM

‘Who broke the camera? Janis or Kostas?’

b. O KOSTAS ti chalase.

the Kostas CL.ACC broke.3SING

‘Kostas broke it.’

c. # Ti chalase O KOSTAS.

CL.ACC broke.3SING the Kostas.NOM

‘Kostas broke it.’

As indicated by the grammaticality judgements, the most natural answer to a narrow focus question introducing a contrast between two entities that are realized as subjects (7a) involves a preverbal (7b) rather than a postverbal subject (7c) (Alexopoulou, 1999; Kapetangianni, 2011; Tsimpli, 1995).²

English differs from Greek in that it does not normally license postverbal subjects. Thus, preverbal subjects are licit, and actually obligatory, in both wide focus (8) and narrow focus contexts (9):

(8) a. What happened with Peter’s camera?

b. John broke the camera.

c. * Broke John the camera.

(9) a. Who broke the camera? John or Peter?

b. JOHN broke the camera.

c Subject placement in Embedded Interrogatives (EI)

Finally, we considered subject–verb inversion in Greek wh-interrogatives. Unlike the type of subject–verb inversion instantiated in Greek declaratives, subject–verb inversion in Greek interrogatives is dissociated from discourse functions and its violation leads to strong ungrammaticality. This is shown in (10) and (11) with matrix and embedded interrogatives, respectively:³

(10) a. Ti echase o Kostas?

what lost.3SING the Kostas.NOM

b. * ti o Kostas echase?

what the Kostas.NOM lost.3SING

‘What did Kostas lose?’

(11) a. Dhen ksero ti echase o Kostas.

NEG know.1SING what lost.3SING the Kostas.NOM

b. * Dhen ksero ti o Kostas echase.

NEG know.1SING what the Kostas.NOM lost.3SING

‘I don’t know what Kostas lost.’

Once again, English differs from Greek in that subject–verb inversion only takes place in matrix interrogatives. In embedded interrogatives, subjects are obligatorily preverbal (12):

(12) a. * I don’t know what lost Kostas.

b. I don’t know what Kostas lost.

2 Predictions

We made two sets of predictions based on the Interface Hypothesis. Under a strong/categorical interpretation of the Hypothesis, heritage children were predicted to be less accurate than monolingual children in the syntax–discourse contexts (i.e. in –Topic Shift and in Wide Focus) but not in the narrow syntactic context (i.e. in Embedded Interrogatives). Furthermore, heritage language use was expected to have an impact on heritage children’s accuracy in –Topic Shift and in Wide Focus, but not in Embedded Interrogatives.

Under a gradient/less categorical interpretation of the Interface Hypothesis, heritage children were predicted to be less accurate than their monolingual peers in both syntax–discourse and narrow syntactic contexts, but in different degrees. As to the impact of heritage language use on heritage children’s accuracy, it was expected to be stronger for the syntax discourse contexts than for the narrow syntactic one. In view of the results obtained in Argyri and Sorace (2007), we treated the second set of predictions as more likely than the first one.

IV Methods

1 Participants

Two groups of Greek-speaking children participated in our study: one group of heritage children and one group of monolingual children residing in Greece. The heritage children were tested in the same period in two different locations: (1) a group of 43 six-to-seventeen-year-old Greek heritage children was tested in Western Canada, and (2) a group of 35 six-to-twelve-year-old Greek heritage children was tested in New York City (NYC). The Greek heritage children were recruited from Greek language schools (Saturday and weekday afternoon schools) operating in Western Canada (Alberta, Saskatchewan, and British Columbia) and New York City. A group of 45 six-to-fourteen-year-old monolingual children residing in Greece served as the control group.

To collect information on the timing and length of exposure to the two languages, parental concerns about the child’s language development, the child’s and the guardian’s place of birth and year of immigration, we used a parental questionnaire that was based on the Alberta Language Environment Questionnaire (ALEQ; Paradis, 2011a). This information was used to include and exclude participants from the group. Children from the two heritage groups were excluded from the study if they spoke languages other than Greek and English or if they failed to answer any of the questions in Greek during informal interaction. Children were included in the group if they were exposed to Greek from birth, if they had all immigrated to North America by the age of three years in the case of first generation immigrant children, and had started sustained and consistent exposure to English in a daycare, or a preschool programme by the age of five years. This resulted in eleven children from the Canadian group and five from the NYC group being excluded. The final numbers of children in each group are reported in Table 1. In terms of immigration generation, the Canadian heritage children were predominantly third generation immigrants. There was only one child that belonged to the second generation and thirteen children whose one parent was first and the other parent was second generation. In the NYC group, two children had migrated to the USA before the age of three years and were first generation immigrants. In addition, there were eleven children whose one parent was first and the other second generation immigrant, fifteen children, who were second generation immigrants, and three children, who belonged to third generation heritage speakers.

Table 1.

Mean age, Greek expressive vocabulary, English receptive vocabulary, current use of Greek and use of Greek in early childhood in the North American (Canadian and NYC) and the monolingual Greek-speaking groups.

Background variables		Canadian group (n = 32 )	NYC group (n = 30)	North American group (n = 62)	Greek monolingual children (n = 45)
Age (in months)	Mean	136.3	113.4	125.1	130.2
	Range	72–203	78–134	72–203	75–177
	SD	30.5	15.1	26.7	24
GR-EXPR (RS)	Mean	17.3	24.4	20.7	45.6
	Range	4–32	8–45	4–45	39–49
	SD	8.7	9.5	9.7	2.3
PPVT-4 (SS)	Mean	106.3	109.3	107.7
	Range	85–136	87–135	85–136
	SD	13.7	11.4	12.7
CurGrUse	Mean	.42	.50	.46
	Range	.25–90	.25–1	.25–1
	SD	.21	.24	.22
GrUseEC	Mean	.83	.74	.79
	Range	.33–1	.33–1	.33–1
	SD	.23	.29	.25
SES	Mean	16.77	18.04	17.4	17.3
	Range	12–24	12–20	12–24	6–24
	SD	2.95	3.05	3	4.7

Notes. GR-EXPR = Greek expressive vocabulary (Vogindroukas et al, 2009); RS = Raw scores; PPVT-4 = English receptive vocabulary (Dunn and Dunn, 2007); SS = standard scores; CurGrUse = Current Use of Greek; GrUseEC = Greek-use-Early Childhood is the proportion of Greek spoken to the child by family members in early childhood (0 to 36 months); SES = socioeconomic status as measured by years of maternal education.

Independent samples t-tests showed that the heritage group did not differ from the monolingual control group in terms of age (t(100) = −.88, p = .38), even though the NYC group was the youngest of both groups (vs. Canadian children: p < .001; vs. monolingual children: p = .016), whereas the Canadian group did not differ from the monolingual control group (p = .28), as pairwise comparisons with Bonferroni correction revealed. The Canadian children did not differ from the NYC children in terms of early exposure to Greek (t(49.1) = −1.37, p = .17), current use of Greek (t(48.2) = −1.41, p = .16), or socioeconomic background (SES) as measured through years of maternal education, and all three groups having mid-to-high SES levels. The heritage children were also matched on SES with the monolingual control children (Table 1).

2 Materials

a Baseline tasks

Proficiency in English: To assess children’s proficiency in English, we used the Peabody Picture Verification Task, 4th edition (PPVT-IV; Dunn and Dunn, 2007), which is a receptive vocabulary task standardized with English-speaking monolingual children in the United States and Canada. In this task, children are shown a four-picture panel and are asked to point to the picture that corresponds to the word spoken by the experimenter. Raw scores were converted to standard scores (Table 1). All bilingual heritage children met monolingual age-appropriate norms, confirming their good competence in English. The two groups of heritage children did not differ in their proficiency in English (F(1,60) = 160.6, p = 1).

Proficiency in Greek: Children’s proficiency in Greek was assessed using the only currently available task for this age range (Vogindroukas et al., 2009). This is an expressive vocabulary test, which is an adaptation of Renfrew’s (1997) picture naming task. In this task, children are presented with a total of 50 black-and-white flashcards and are asked to name the object depicted on the flashcard. There is no discontinue rule and all items need to be administered (Table 1). A one-way ANOVA with the raw vocabulary scores as the within participants factor and Group as the between groups factor revealed a main effect of Group (F(2, 104) = 169.4, p < .001, partial η² = .69) Post hoc tests with Bonferroni correction revealed that the NYC heritage children had higher accuracy on the Greek expressive vocabulary task than their Canadian counterparts (p < .001). Both groups had a statistically significant lower accuracy than the monolingual controls (p < .001 in both cases), even when collapsed into the combined heritage group (p < .001). For the analysis, we converted the raw scores into proportions with the maximum score of 50 as the denominator and the number of correct items as the numerator.

Language background questionnaire: To collect information regarding Greek language use with family members at home (parents and siblings, both at the time of testing and in early childhood), children’s input and output in Greek, time of immigration to the USA or Canada, and parental education, we administered a parental questionnaire that was based on the Alberta Language Environment Questionnaire (ALEQ; Paradis, 2011a). Whereas the original ALEQ was designed to measure the current English language use (input and output) in the child’s environment, the adaptation that we used (ALEQ Heritage) measured the current heritage (i.e. Greek) language use.

Information about the child’s input was measured using questions about how frequently the parents, guardians (including grandparents) and other siblings spoke Greek to the child on a scale from 0 (Greek almost never/English almost always) to 4 (Greek almost always/English almost never). Output was measured as the frequency with which the child spoke Greek to the same family members and guardians. Greek language use at home was then calculated as the mean proportion of Greek input and output that the child received from and directed to other family members (mother, father, siblings and grandparents).

Additionally, we calculated Greek language use in early childhood, that is before the age of three years, by asking about the frequency of interaction in Greek (same scale as above for input). Children’s Age of Onset (AoO) and length of systematic exposure (LoE) to English was also measured and this usually coincided with the child’s attendance of daycare/preschool/school. The Canadian heritage children were exposed to English at the mean age of 41.1 months (range: 12–60, SD: 11.68) and had a mean LoE to English of 95.5 months (range: 36–155, SD: 29.35), and the NYC children were exposed to English at the mean age of 41.8 months (range: 12–96, SD: 14.7) and had a mean LoE to English of 70.3 months (range: 33–115, SD: 19). Finally, information about the socioeconomic status of the family was calculated based on years of maternal education. All heritage children belonged to mid-to-high SES families.

b Experimental task

To examine children’s use of subjects in syntax–discourse and narrow syntactic contexts, we adapted Argyri and Sorace’s (2007) elicited production task that was designed to target: (1) subject form in –Topic Shift contexts, as in (13), (2) subject placement in Wide Focus (WF) contexts, as in (14), and (3) subject placement in embedded wh-interrogative (EI) contexts, as in (15). For each context, participants were shown a picture of animated characters on a laptop screen, and were, subsequently, asked a question that was meant to prompt the production of the target structure (for a detailed presentation of the task, see Appendix 1).

In the –Topic Shift condition (13), participants were shown a picture of a character involved in a certain activity (e.g. a woman going to the bakery) and were prompted with a lead-in question to give a reason for the character’s activity. They were then instructed to begin their response with the word epidhi ‘because’. The felicitous response in this condition involved the use of a null subject pronoun to refer to the already prominently introduced topic/discourse antecedent.

(13) Experimenter: Jati pije sto furno i kiria Sofia?

why went.3SING to the bakery the miss Sofia.NOM

‘Why did Miss Sofia go to the bakery?’

Expected response: Epidhi ithele na agorasi psomi.

Because wanted to buy.3SING bread

‘Because she wanted to buy bread.’

In the WF condition (14), participants were presented with a picture depicting an activity between two animated characters (e.g. a dog playing with a ball and a little girl, Maria, who was looking at the dog, clearly upset). They were then asked a wide focus question about the activity. This context in Greek was expected to induce a postverbal subject, as in (14):

(14) Experimenter: Ti ejine i bala tis Marias?

what happened.3Sing the ball.Nom the Maria.Gen

‘What happened to Maria’s ball?’

Expected response: Tin pire o skilos.

It CL.ACC took.3SING the dog.NOM

‘The dog took it.’

In the EI condition (15), participants were presented with a picture depicting a forgetful grandparent, who complained about not remembering his/her grandchild’s activities. After each picture, participants were prompted to complete a lead-in sentence of the sort i jaja / o papus den thimate … (‘Grandmother/grandpa doesn’t remember …’), which again elicits postverbal subjects in Greek:

(15) Experimenter: I egoni mu i Maria mu ipe

the granddaughter.NOM my the Maria.NOM me told.3SING

ti forese alla dhen thimame tora.

what put on.3SING but NEG remember.1SING now

‘My granddaughter Maria, told me what she put on but I don’t remember.’

Experimenter: Ti dhen thimate i jaja?

what NEG remember.3SING the grandmother.NOM

‘What doesn’t the grandmother remember?’

Expected Response: dhen thimate ti forese i egoni tis.

NEG remember.3SING what put on.3SING the granddaughter her.GEN

‘She doesn’t remember what her granddaughter put on.’

There were eight items per condition, which resulted in twenty-four items. The items were the same as the ones developed by Argyri and Sorace (2007), but were increased in number from six to eight per structure. The added items were comparable to the ones borrowed from Argyri and Sorace (2007).

3 Coding and scoring

In the –Topic Shift condition, all responses with a null subject were coded as correct and given a value of ‘1’, and all responses with an overt subject, either a pronoun or a full lexical DP, were coded as incorrect and were given a value of ‘0’. In the WF and the EI conditions, all responses with postverbal subjects were coded as correct and were given a value of ‘1’ and responses with preverbal subjects as incorrect and were given a value of ‘0’. Responses containing intersentential code-switching, responses with missing verbs or null responses were excluded from the calculation. This amounted to 11.02% of the data for the Canadian group, 9.6% of the data for the NYC group. There were no such responses in the monolingual Greek group.

4 Procedure

Children were tested in their homes (21 out of the 62 children) or at their Greek schools (41 out of the 62 children). Each child participated in an hourly session that included a battery of other tasks on a laptop and in a shorter session where the PPVT was administered. All tasks were administered by a Greek–English bilingual researcher. Parents were administered the questionnaire over the phone, through face-to-face interviews or were asked to complete it on their own.

V Results

We statistically analysed the results using a generalized linear mixed model with a logistic link function given the binomial nature of the dependent variable: correct (VS in the WF and EI conditions, null subject in the –TS condition) and incorrect responses (SV in the WF and EI conditions and overt subject in the –TS condition). Statistical analysis was carried out using the lme4 package in R (Baayen, 2008). We opted for mixed-effects logistic regression models over repeated-measures ANOVAs because we wanted to consider whether individual variability on the participant and item level affected the results (Baayen, 2008). The mixed effects logistic regression allows using a larger number of data-points to include a more complete and theoretically-motivated random effects structure: specifically, one that specifies a random intercept for subject and a random slope for condition by subject (Barr et al., 2013; Dixon, 2008). This random effects structure follows the hypothesis that performance not only varies between individuals, but also individual performance varies across conditions. Visualization of the results was carried out using ggplot and visualization of the regression results using the effects package (Fox, 2016).

1 Accuracy

Figure 1 presents the accuracy results on the –TS, the EI and the WF conditions. In the –TS condition, the dependent variable is the proportion of null pronouns, whereas in the EI and the WF conditions the dependent variable was the proportion of VS structures that the children produced.

Figure 1.

Mean accuracy per group and condition.

To address the first research question, we fitted a mixed-effects regression model with Group (heritage, monolingual) and Condition (–TS, EI and WF) as the fixed-effects predictors. The models were built up incrementally with Group and Condition included as the main effects in the first model, followed by a model with the interaction between Condition and Group. Random intercepts for subjects and items were also included in the models. The optimal model included the maximal random effect structure justified by the data, and this resulted in including the by-subject intercept and a condition by subject slope. Model comparisons using chi square tests were carried out to ascertain whether the inclusion of a main effect or an interaction improved the fit of the model. In the first model, the embedded interrogative condition was used as the reference level since this is the only purely syntactic condition out of the three, and we wanted to establish whether the two discourse conditions differed from the strictly syntactic condition. Pairwise comparisons between the conditions were carried out by changing the reference level. To judge the fit of the logistic model, we calculated the Concordance Index (C; Chatterjee and Hadi, 2006) and Somers’ Dxy rank correlation between the predicted probabilities and the observed responses (Harrell, 2015). A C-index and a Somers’ Dxy rank correlation of above 0.8 indicates that the model is good.

The optimal model was the one with the main effects of Group and Condition and the Condition by participant slope and participant intercept. This model provided a good fit (C = .91, D_xy = .83) and is presented in Table 2.⁴ The optimal model revealed that the monolingual children were at ceiling across conditions and differed overall from the heritage children. The WF condition had the lowest accuracy and differed from both the –TS (Table 2) and the EI (< .001) conditions, whereas the –TS condition had higher accuracy than the EI condition (Table 2). In the –TS condition, the heritage children produced primarily full lexical DPs when not producing a null subject. These responses amounted to 8.9% off all responses (34/381 total items).

Table 2.

Optimal model with group, condition and use as predictors.

	Estimate	Standard error	t	p
(Intercept)	−1.38	0.51	2.67**	.007
Monolingual	10.28	1.32	7.78***	< .001
Topic shift	1.81	0.65	2.78**	.0055
Wide focus	−4.17	0.65	−6.43***	< .001

Notes. * p < .05, ** p < .01, *** p < .001.

2 Effects of language use on interface and narrow-syntactic conditions in the heritage group

To examine whether heritage language use affected the performance of the heritage group across the three conditions, we ran a mixed-effects logistic regression with Condition and Use as fixed effects in the heritage group only. Random effects included the participant intercept and the Condition by Participant slope. Language Use was entered in the model as a continuous variable. The optimal model is presented in Table 3 and it provided a good fit (C = .84, Dxy = .69).⁵

Table 3.

Optimal model with heritage language use as predictor for the heritage children’s accuracy across the three conditions.

	Estimate	Standard error	t	p
(Intercept)	−1.46	0.65	−2.24*	.025
Topic shift	1.63	0.51	3.17**	.0016
Wide focus	−4.05	0.59	−6.83***	< .001
Use	5.88	1.19	4.91***	< .001

Notes. * p < .05, ** p < 01, *** p < .001.

Across all three conditions, performance in the heritage group improved with increased language use (Figure 2).⁶ However, the three conditions differed in terms of how much heritage language use is required for the heritage children’s accuracy to improve across the three conditions. In the –TS condition, the heritage children with very little input (less than 25%) had approximately 75% accuracy and reached ceiling performance with approx. 40% input, whereas heritage language use seems to matter more in the case of the two conditions tapping into word order, EI (approximately 35% accuracy when use is less than 25%) and WF (less than 10% accuracy when use is less than 25%). Heritage children reached ceiling performance on EI with approximately 75% of language use, whereas they failed to reach ceiling accuracy on the WF condition even when their heritage language use was above 75%.

Figure 2.

Interaction effect of condition and use on accuracy.

VI Discussion

The main goal of this study was to investigate whether the amount of language use in heritage contexts equally affects narrow syntactic and interface phenomena. To this end, we investigated the relationship between the amount of heritage language use at home and heritage children’s accuracy across two different grammatical domains: syntax–discourse and narrow syntax. More specifically, we tested Greek–English bilingual children learning Greek as a heritage language in two syntax–discourse (subject realization in –TS; subject placement in WF) and one narrow syntactic structure (subject placement in EI). Our research questions were aimed at determining (1) differences between monolingual and heritage children, as well as (2) differences between heritage children associated with their daily amount of heritage language use at home.

Regarding the comparison between heritage children and the monolingual controls, heritage children showed variable performance in all three structures, although in different degrees, unlike their monolingual peers, who performed at ceiling across the board. More precisely, heritage children opted for:

preverbal subjects in WF (a context, where post-verbal subjects are preferred);

preverbal subjects in EI (a context, where post-verbal subjects are required);

(to a much lesser degree) overt lexical subjects in –TS (a context, where null subjects are preferred).

Importantly for our purposes, heritage children’s accuracy varied depending on the structure. The –TS structure was the least challenging of the three, with heritage children performing almost at ceiling. Of the remaining two subject placement structures, the EI was shown to be less challenging than the WF.

Turning to our second research question, results revealed a non-linear relationship between the amount of heritage language use in the home and all three structures. More specifically, the structures differed in terms of how much heritage language use was required for the heritage children’s performance to improve and what the starting point in terms of accuracy for each structure was. In the –TS structure, heritage children even with limited language use (less than 25%) achieved high performance (approx. 75%). The structures tapping into subject placement, on the other hand, displayed a different degree of dependence on language use. Thus, with less than 25% heritage language use, children had 35% accuracy in the EI structure and less than 10% accuracy in the WF structure. This suggests that when language use is limited, the WF structure (a syntax–discourse structure) is more vulnerable compared to the EI structure (a narrow-syntactic structure). Interestingly, the accuracy difference between the two subject placement conditions remained as the proportion of language use increased. Namely, heritage children reached almost ceiling accuracy on the EI condition when the proportion of language use was approximately 60%. Conversely, even with 100% heritage language use in the home, heritage children did not reach more than 75% accuracy on the WF condition. A possible interpretation for these results is that the WF condition, being dependent on the coordination of both syntactic and discourse knowledge, is inherently more challenging and, consequently, requires language use beyond the home environment to be fully acquired. It is also possible that it is more vulnerable to cross linguistic influence as in this context, the overlapping option between Greek and English (i.e. preverbal subjects) is merely dispreferred, not disallowed.

What are the implications of these findings for the Interface Hypothesis? Overall, our study – in line with Argyri and Sorace’s (2007) study on simultaneous Greek–English bilingual children raised in the UK – suggests two deviations from the strong version of the Interface Hypothesis. First, one of the two syntax–discourse structures (i.e. distribution of overt and null pronominal subjects in –TS contexts) was found to be largely unproblematic. Even children with a minimum of heritage language use (less than 25%) made appropriate uses of null subjects in –TS contexts (approx. 75%). There were very few uses of overt lexical DPs (lower than 10%) and no uses of overt pronominal subjects. Second, subject placement in EI (a narrow syntactic context) was also found to be challenging for heritage children and dependent on language use though to a lesser degree than subject placement in WF.

The observed dissociation between the two syntax–discourse contexts is in line with Tsimpli et al.’s (2004) study on Greek attriters and Argyri and Sorace’s (2007) study on simultaneous Greek–English bilingual children. In both studies, Greek bilinguals were shown to differ significantly from their monolingual peers in the rate of non-felicitous preverbal subjects, but not in the rate of non-felicitous overt subject pronouns in –TS. At the same time, these results set Greek bilinguals apart from bilingual learners of other null subject languages such as Italian and Spanish, who, with the exception of high proficiency speakers (Montrul and Rodríguez-Louro, 2006; Rothman, 2009a) and the near natives of Judy (2015) and Judy and Rothman (2014), have been repeatedly shown to overextend the use of overt pronouns into pragmatically infelicitous contexts (for Italian–English bilinguals, see, amongst others, Belletti et al., 2007; Serratrice et al., 2004; for Spanish–English bilinguals, see Montrul, 2004; Paradis and Navarro, 2003).

It is possible that the observed discrepancies are related to the lexical properties of third person personal pronouns in Greek (a possibility also discussed in Tsimpli and Sorace, 2006 to account for the fact that Russian near native speakers of Greek generally avoid the use of third person overt pronouns in –TS contexts). Unlike the Italian lui and the Spanish él, the Greek aftos is a demonstrative that may assume an anaphoric interpretation. Thus, its use in the –TS contexts of our study would introduce an ambiguity between an anaphoric reading (establishing co-reference with the subject antecedent) and a deictic reading (establishing an association with an extra-linguistic referent). In other words, it could be the case that due to the salience of its deictic reading, the Greek aftos is resilient to input reduction (and cross-linguistic transfer). If this explanation is on the right track, then we expect to find differences even among monolingual speakers of Greek, Italian, and Spanish in their use of third subject personal pronouns.

It is also possible that the reason why Greek heritage children are more likely to use non-felicitous preverbal than non-felicitous overt subjects is related to frequency of use. More precisely, while preverbal subjects (of transitive verbs) in Greek are more frequent than postverbal subjects (Skopeteas, 2016), overt subjects are less frequent than null subjects (Kapetangianni, 2011). If this explanation is on the right track, then we might expect that even young monolingual learners of Greek will be more challenged by subject placement than by subject realization.

Setting aside, therefore, the complication raised from pronominal subject realization, which as suggested above could be due to frequency effects and/or to the lexical semantics of the Greek pronouns, the present study indicates that, at least in the case of heritage speakers, who have experienced a drastic reduction in exposure to their heritage language in early childhood, the amount of language use has an impact on subject placement not only in the syntax–discourse domain but also in the narrow syntactic domain (in different degrees). In this respect, our results align with Cuza’s (2012, 2016) studies on heritage Spanish, showing that heritage speakers are challenged by subject placement in the narrow syntactic context of interrogatives.

Last but not least, in this study, we have only considered the effects of quantitative aspects of the input. Heritage children, though, differ from monolinguals and from each other, not only in terms of the quantity of input that they have received, but also in terms of its type (Paradis and Navarro, 2003; Paradis, 2011b; Kupisch and Rothman, 2018; Sorace, 2004, 2011; Unsworth, 2016, 2017). This is because the main source of the heritage language comes from their parents and grandparents, who, depending on the generation to which they belong, might show different degrees of attrition. Whether different degrees of attrition have an effect on heritage language acquisition and whether this effect is relevant for both narrow syntactic and syntax–discourse properties is a question we aim to address in a follow up study by analysing the parental input of our participants.

Supplemental Material

Daskalaki_et_al._Appendix_(1)_(1) – Supplemental material for Input effects across domains: The case of Greek subjects in child heritage language

Supplemental material, Daskalaki_et_al._Appendix_(1)_(1) for Input effects across domains: The case of Greek subjects in child heritage language by Evangelia Daskalaki, Vasiliki Chondrogianni, Elma Blom, Froso Argyri and Johanne Paradis in Second Language Research

Footnotes

Acknowledgements

We thank EFF-SAS Research Fund (University of Alberta) and Advanced Research Collaborative Fellowship (City University of New York) for funding this project. We also thank St George’s Hellenic School of Edmonton, the Hellenic Community of Edmonton, St Paul’s Greek Orthodox Community of Regina, the Hellenic Community of Calgary, the Greek Orthodox Community of East Vancouver, the Academy of Hellenic Paideia and Hellenic Education, and Language Programme in New York City. Many thanks to Eliana Kavgadoulis and Katerina Katehis for assisting with the data collection in Western Canada and New York City. We are furthermore grateful to all the children and their parents for their participation and enthusiasm.

Declaration of Conflicting Interest

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

Notes

Supplemental Material

Supplemental material for this article is available online.

References

Alexiadou

(1999) On the properties of some Greek word order patterns. In: Alexiadou

Horrocks

Stavrou

(eds) Studies in Greek syntax: Studies in natural language and linguistic theory (pp. 45–65). Dordrecht: Springer.

Alexiadou

Anagnostopoulou

(1998) Parametrizing AGR: Word order, V-movement, and EPP-checking. Natural Language and Linguistic Theory 16: 491–539.

Alexopoulou

(1999) The syntax of discourse functions in Greek: A non configurational approach. Unpublished PhD dissertation, University of Edinburgh, Edinburgh, UK.

Alonso-Ovalle

Fernandez-Solera

Frazier

Clinton

(2002) Null vs. overt pronouns and the topic–focus articulation in Spanish. Journal of Italian Linguistics 14: 151–69.

Argyri

Sorace

(2007) Crosslinguistic influence and language dominance in older bilingual children. Bilingualism: Language and Cognition 10: 79–99.

Baayen

(2008) Analysing linguistic data: A practical introduction to statistics using R. Cambridge: Cambridge University Press.

Barr

Levy

Scheepers

Tily

(2013) Random effects structure for confirmatory hypothesis testing: Keep it maximal. Journal of Memory and Language 68: 255–78.

Bedore

Peña

Summers

et al . (2012) The measure matters: Language dominance profiles across measures in Spanish–English bilingual children. Bilingualism: Language and Cognition 15: 616–29.

Belletti

Bennati

Sorace

(2007) Theoretical and developmental issues in the syntax of subjects: Evidence from near-native Italian. Natural Language and Linguistic Theory 25: 657–89.

10.

Benmamoun

Montrul

Polinsky

(2013) Heritage languages and their speakers: Opportunities and challenges for linguistics. Theoretical Linguistics 39: 129–81.

11.

Bini

(1993) La adquisición del italiano: Más allá de las propiedades sintácticas del parámetro pro-drop [The acquisition of Italian: Beyond the syntactic properties of the pro-drop parameter]. In Liceras

(ed.) La lingüística y el análisis de los sistemas no natives (pp. 126–39). Ottawa, ON: Dovehouse.

12.

Carminati

(2002) The processing of Italian subject pronouns. Unpublished PhD dissertation, University of Massachusetts, Amherst, MA, USA.

13.

Chatterjee

Hadi

(2006) Regression Analysis by example. JFourth Edition. John Wiley & sons, Inc., Hoboken, New Jersey.

14.

Clements

Domínguez

(2017) Reexamining the acquisition of null subject pronouns in a second language: Focus on referential and pragmatic constraints. Linguistic Approaches to Bilingualism 7: 33–62.

15.

Cuza

(2012) Cross-linguistic influence at the syntax proper: Interrogative subject–verb inversion in heritage Spanish. International Journal of Bilingualism 17: 71–96.

16.

Cuza

(2016) The status of interrogative subject–verb inversion in Spanish–English bilingual children. Lingua 180: 124–38.

17.

Dixon

(2008) Models of accuracy in repeated-measures designs. Journal of Memory and Language 59: 447–56.

18.

Domínguez

(2013) Understanding interfaces: Second language acquisition and first language attrition of Spanish subject realization and word order variation. Amsterdam: John Benjamins.

19.

Donaldson

(2012) Syntax and discourse in near-native French: Clefts and focus. Language Learning 62: 902–30.

20.

Dosi

(2017) Asymmetries in grammatical aspect in (non-)heritage Greek–English bilingual children: Evidence from comprehension and production tasks. 37th Annual Meeting of the Department of Linguistics 37: 213–26.

21.

Dunn

(2007) Peabody picture vocabulary test. 4th edition. San Antonio, TX: Pearson.

22.

Filiaci

(2010) Null and overt subject biases in Spanish and Italian: A cross-linguistic comparison. In Borgonovo

Espanol-Echevarria

Prevost

(eds) Selected proceedings of the 12th Hispanic Linguistics Symposium (pp.171–82). Somerville, MA: Cascadilla Press.

23.

Flores

(2012) Differential effects of language attrition in the domains of verb placement and object expression. Bilingualism: Language and Cognition 15(3): 550–567.

24.

Fox

(2016) Applied Regression Analysis and Generalized Linear Models. Third Edition. SAGE Publications, Inc. LA.

25.

Hacohen

Schaeffer

(2007) Subject realization in early Hebrew/English bilingual acquisition: The role of crosslinguistic influence. Bilingualism: Language and Cognition 10: 333–44.

26.

Haegeman

(1997) Register variation, truncation and subject omission in English and in French. English Language and Linguistics 1: 233–70.

27.

Harrell

(2015) Regression modeling strategies: With applications to linear models, logistic and ordinal regression, and survival analysis. Heidelberg: Springer.

28.

Haznedar

(2007) Crosslinguistic influence in Turkish–English bilingual first language acquisition: The overuse of subjects in Turkish. In: Belikova

Meroni

Umeda

(eds) Proceedings of the 2nd Conference on Generative Approaches to Language Acquisition North America (GALANA) (pp. 124–34). Somerville, MA: Cascadilla Proceedings Project.

29.

Jia

Paradis

(2015) The use of referring expressions in narratives by Mandarin heritage language children and the role of language environment factors in predicting individual differences. Bilingualism: Language and Cognition 18: 737–52.

30.

Judy

(2015) Knowledge and processing of subject-related discourse properties in L2 near-native speakers of Spanish, L1 Farsi. In: Judy

Perpiñán

(eds) The acquisition of Spanish in understudied language pairings (pp. 169–200). Amsterdam: John Benjamins.

31.

Judy

Rothman

(2014) The processing of pronominal subjects by near-native L1 Persian-L2 Spanish adult bilinguals. In: Orman

Vallea

(eds) BUCLD 38: Proceedings of the 38th annual Boston University Conference on Language Development (pp. 192–204). Somerville, MA: Cascadilla Press.

32.

Kaltsa

Tsimpli

I-M

Rothman

(2015) Exploring the source of differences and similarities in L1 attrition and heritage speaker competence: Evidence from pronominal resolution. Lingua 164: 266–88.

33.

Kapetangianni

(2011) Variable word order in child Greek. In: Anderssen

Bentzen

Westergaard

(eds) Variation in the input: Studies in theoretical psycholinguistics (pp. 179–205). Dordrecht: Springer.

34.

Keller

Alexopoulou

(2001) Phonology competes with Syntax: Experimental evidence for the interaction of word order and accent placement in the realisation of Information Structure. Cognition 79: 301–72.

35.

Kotzoglou

(2006) Subject–verb inversion in Greek: Implications for head movement and typology. Journal of Universal Language 7: 91–137.

36.

Kupisch

(2012) Generic subjects in the Italian of early German–Italian bilinguals and German learners of Italian as a second language. Bilingualism: Language and Cognition. 15: 736–56.

37.

Kupisch

(2014) Adjective placement in simultaneous bilinguals (German–Italian) and the concept of cross-linguistic overcorrection. Bilingualism: Language and Cognition 17: 222–33.

38.

Kupisch

Rothman

(2018) Terminology matters! Why difference is not incompleteness and how early child bilinguals are heritage speakers. International Journal of Bilingualism 22(5): 564–582.

39.

Kupisch

Akpinar

Stöhr

(2013) Gender assignment and gender agreement in adult bilingual and second language speakers of French. Linguistic Approaches to Bilingualism 3: 150–79.

40.

Kupisch

Lein

Barton

et al . (2014) Acquisition outcomes across domains in adult simultaneous bilinguals with French as weaker and stronger language. Journal of French Language Studies 24: 1–30.

41.

Leal

Rothman

Slabakova

(2014) A rare structure at the syntax–discourse interface: heritage and Spanish-dominant native speakers weigh in. Language Acquisition 21: 411–29.

42.

Lozano

(2006) The development of the syntax–information structure interface: Greek learners of Spanish. In: Torrens

Escobar

(eds) The acquisition of syntax in Romance languages (pp. 371–99). Amsterdam: John Benjamins

43.

Margaza

Bel

(2006) Null subjects at the syntax–pragmatics interface: Evidence from Spanish interlanguage of Greek speakers. In: O’Brien

Shea

Archibald

(eds) Proceedings of the 8th Generative Approaches to Second Language Acquisition Conference (GASLA 2006) (pp. 88–97). Somerville, MA: Cascadilla Press.

44.

Montrul

(2004) Subject and object expression in Spanish heritage speakers: A case of morphosyntactic convergence. Bilingualism: Language and Cognition 7: 125–42.

45.

Montrul

(2008) Incomplete acquisition in bilingualism: Re-examining the age factor. Amsterdam: John Benjamins.

46.

Montrul

(2011) Interfaces and incomplete acquisition. Lingua 121: 591–604.

47.

Montrul

(2015) The acquisition of heritage languages. Cambridge: Cambridge University Press.

48.

Montrul

Polinsky

(2011) ‘Why not heritage speakers? ‘. Linguistic Approaches to Bilingualism 1: 58– 62.

49.

Montrul

Rodríguez-Louro

(2006) Beyond the syntax of the null subject parameter: A look at the discourse–pragmatic distribution of null and overt subjects by L2 learners of Spanish. In: Torrens

Escobar

(eds) The acquisition of syntax in Romance languages (pp. 401–18). John Benjamins: Amsterdam.

50.

Panagiotidis

Tsiplakou

(2004) ‘What I said this time?’ Wh-questions in the acquisition of English L2. In: Catsimali

Kalokairinos

Anagnostopoulou

Kappa

(eds) Proceedings of the 6th International Conference on Greek Linguistics. Rethymno: University of Crete.

51.

Papadopoulou

Peristeri

Plemenou

Marinis

Tsimpli

(2015) Pronoun ambiguity resolution in Greek: Evidence from monolingual adults and children. Lingua 155: 98–120.

52.

Paradis

(2011a) Individual differences in child English second language acquisition: Comparing child-internal and child-external factors. Linguistic Approaches to Bilingualism 1: 213–37.

53.

Paradis

(2011b) The impact of input factors on bilingual development: Quantity vs. quality. Linguistic Approaches to Bilingualism 1: 67–70.

54.

Paradis

Navarro

(2003) Subject realization and crosslinguistic interference in the bilingual acquisition of Spanish and English: what is the role of the input? Journal of Child Language 30: 371–93.

55.

Pinto

(2006) Subject pronouns in bilinguals: Interference or maturation? In: Torrens

Escobar

(eds) The acquisition of syntax in Romance languages (pp. 331–350). Amsterdam/Philadelphia, PA: John Benjamins.

56.

Polinsky

(2007) Incomplete acquisition: American Russian. Journal of Slavic Linguistics 14: 191–262.

57.

Polinsky

Kagan

(2007) Heritage languages: In the ‘wild’ and in the classroom. Language and Linguistics Compass 1: 368–95.

58.

Rothman

(2009a) Pragmatic deficits with syntactic consequences?: L2 pronominal subjects and the syntax–pragmatics interface. Journal of Pragmatics 41: 951–73.

59.

Rothman

(2009b) Understanding the nature and outcomes of early bilingualism: Romance languages as heritage languages. International Journal of Bilingualism 13: 155–63.

60.

Rizzi

(1982) Issues in Italian syntax. Dordrecht: Foris.

61.

Scontras

Fuchs

Polinsky

(2015) Heritage language and linguistic theory. Frontiers in Psychology 6: 1545.

62.

Serratrice

(2007) Cross-linguistic influence in the interpretation of anaphoric and cataphoric pronouns in English-Italian bilingual children. Bilingualism 10(03): 225–238.

63.

Serratrice

Sorace

Paoli

(2004) Crosslinguistic influence at the syntax–pragmatics interface: Subjects and objects in English–Italian bilingual and monolingual acquisition. Bilingualism: Language and Cognition 7: 183–205.

64.

Serratrice

Sorace

Francesca

Baldo

(2009) Bilingual children’s sensitivity to specificity and genericity: Evidence from metalinguistic awareness. Bilingualism: Language and Cognition 12: 239–57.

65.

Skopeteas

(2016) Information structure in Modern Greek. In: Féry

Ishihara

(eds) The Oxford handbook of information. Oxford: Oxford University Press. Available at: http://doi.org/10.1093/oxfordhb/9780199642670.013.15 (accessed June 2018).

66.

Sorace

(2004) Native language attrition and developmental instability at the syntax-discourse interface: data, interpretations and methods. Bilingualism: Language and Cognition 7: 143–145.

67.

Sorace

(2011) Pinning down the concept of ‘interface’ in bilingualism. Linguistic Approaches to Bilingualism 1: 1–33.

68.

Sorace

(2012) Pinning down the concept of ‘interface’ in bilingualism: A reply to peer commentaries. Linguistic Approaches to Bilingualism 2: 209–17.

69.

Sorace

Filiaci

(2006) Anaphora resolution in near-native speakers of Italian. Second Language Research 22: 339–68.

70.

Sorace

Serratrice

(2009) Internal and external interfaces in bilingual language development: Beyond structural overlap. International Journal of Bilingualism 13: 195–210.

71.

Sorace

Serratrice

Filiaci

Baldo

(2009) Discourse conditions on subject pronoun realization: Testing the linguistic intuitions of older bilingual children. Lingua 119: 460–77.

72.

Spyropoulos

Revithiadou

( 2009) Subject chains in Greek and PF processing. MIT Working Papers in Linguistics 57: 293–309.

73.

Tsimpli

(1990) The clause structure and word order of Modern Greek. UCL Working Papers in Linguistics 2: 226–55.

74.

Tsimpli

(1995) Focusing in Modern Greek. In: Kiss

(ed.) Discourse configurational languages (pp. 176–207). Oxford: Oxford University Press.

75.

Tsimpli

Sorace

(2006) Differentiating interfaces: L2 performance in syntax–semantics and syntax–discourse phenomena. In: Bamman

Magnitskaia

Zaller

(eds) Proceedings of the 30th Boston University Conference on Language Development (pp. 653–64). Somerville, MA: Cascadilla Press.

76.

Tsimpli

Sorace

Heycock

Filiaci

(2004) First language attrition and syntactic subjects: A study of Greek and Italian near-native speakers of English. International Journal of Bilingualism 8: 257–77.

77.

Unsworth

(2016) Quantity and quality of language input in bilingual language development. In: Nicoladis

Montanari

(eds) Lifespan perspectives on bilingualism (pp. 136–96). Berlin: Mouton de Gruyter; Washington, DC: American Psychological Association.

78.

Unsworth

(2017) Quantifying language experience in HL development. In: Schmid

Köpke

(eds) The Oxford handbook of first language attrition. Oxford: Oxford University Press.

79.

Valdes

(2000).The teaching of heritage languages: An introduction for Slavic-teaching professionals. In: Kagan

Olga

Rifkin

Benjamin

(eds) The learning and teaching of Slavic languages and cultures (pp. 375–403). Bloomington, IN: Slavica.

80.

Vogindroukas

Protopapas

Sideridis

(2009) Experiment on the expressive vocabulary (Greek Version of Renfrew Word Finding Vocabulary Test). Chania: Glafki.

81.

White

(2011) Second language acquisition at the interfaces. Lingua 121: 577–90.

82.

Wolleb

(2013) Referring expressions in the narratives of Italian–English bilingual children. Studia Linguistica 67: 28–46.

83.

Zombolou

(2011) Attrition in Greek Diaspora. In: Ihemere

(ed.) Language contact and language shift: Grammatical and sociolinguistic perspectives: Studies in Language Typology 17 (chapter 3). Munich: Lincom Europa.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.34 MB