Which elements of syntax do bilingual children tend to transfer from one of their languages to the other, and how long do these cross-linguistic effects persist in development? Although bilingual children can differentiate between the grammars of their languages from an early age (Meisel, 1989, 2001; De Houwer, 1990; Paradis and Genesee, 1997), cross-linguistic influence between the grammars of bilingual children has been found in many domains, including the expression of null vs. overt subjects, topic drop, word order, verbal inflection, and the use of determiners (Schlyter, 1994; Gawlitzek-Maiwald and Rosemary, 1996; Döpke, 1998; Yip and Matthews, 2000; Müller and Hulk, 2001; Paradis and Navarro, 2003; Kupisch, 2007). Furthermore, there is strong evidence that the two languages of a bilingual adult are highly interconnected and mutually influence each other during language processing (Kroll et al., 2006, inter alia), which presents the possibility that some cross-linguistic effects seen in early child bilingualism may last into adulthood. In this article, I address this possibility by examining the production of Differential Object Marking (DOM) and pre-verbal complementizers by child and adult Basque speakers, constructions which are considered non-standard in Basque. Specifically, I investigated the degree to which age and bilingualism affected the production of these variants in the speech of monolingual and bilingual children learning Basque.

Previous work found evidence of DOM in Basque-speaking adults (Austin, 2006; Rodríguez-Ordóñez, 2013), and some pre-verbal complementizers in children's speech (Austin, 2001). However, the possibility that pre-verbal complementizers are present in the adult input was not examined in previous research, nor were comparisons made of the production of DOM by children and adults. These questions were investigated in this paper. Before describing the methods used in this study, an overview of research into Spanish-Basque contact phenomena will be provided, followed by a description of DOM and pre-verbal complementizers in Basque.

Euskara, the Basque language, is a non-Indo-European language isolate with approximately 800,000 speakers in the Basque Country, a small region between Northern Spain and Southwestern France. In the 20th century, two important demographic changes affected how many people spoke Basque, as well as how proficient they were in speaking the language. First, workers from other parts of Spain moved to cities in the Spanish Basque Country seeking employment, and in doing so nearly quadrupled the population of the region. This influx of workers who were not Basque speakers, combined with Franco's suppression of the minority languages of Spain, turned Basque into a language spoken by only 25% of the population. Another important factor was the loss of Basque monolingualism; presently, all Basque speakers also speak either French or Spanish, with the exception of some young children who have not yet begun formal schooling, such as some of the participants in this study.

The recovery of Basque in the late 20th century was fostered by several developments, principally the end of Franco's dictatorship and the transition to democracy following his death, after which the Basque Country officially became bilingual in 1978. Additionally, in 1982 a new law mandated that all government services be offered in Basque, and also required all public schools to provide access to bilingual education in Basque and Spanish. These measures represented significant progress toward ensuring the survival of Basque because they introduced the language into prestigious spheres of use which had previously been limited to Spanish (Haddican, 2005). Another important factor in the maintenence of Basque was its standardization through the creation of a unified dialect, Euskara Batua (Standard Basque) (Amorrortu, 2000; Haddican, 2005).

Despite centuries of being in contact, the typological characteristics of Basque and Spanish have remained quite different, most notably in their case marking and head direction systems (Silva-Corvalán, 1997). Unlike Spanish, which is a head initial language with nominative/accusative case-marking, Basque has an ergative/absolutive case system and is a head-final language. Nonetheless, the fact that Basque and Spanish have been in contact for well over a 1000 years is readily apparent in both languages. Basque shows evidence of contact with Spanish (and also Latin and French) in its many loanwords and calques. In fact, by one estimate up to 40% of the modern Basque lexicon is comprised of borrowings from these Romance languages¹.

Basque is a triple-agreement language in which the verb is inflected with the person and number features of the subject, indirect object, and direct object. When there is triple agreement on the auxiliary (for the subject, indirect, and direct objects), the absolutive argument (direct object) argument can only be inflected for 3rd person.

A pro-drop language, Basque permits up to three arguments in a clause to be null, as shown in example (1). Most Basque verbs, like the verb in (1) are periphrastic, comprised of a participle and an inflected aux³.

In many languages, case marking is used to differentiate human and non-human direct objects, a pattern known as DOM. This phenomenon occurs in other languages too, such as the use of the “personal a” marker in Spanish that is required with specific, animate direct objects, as seen in (2a,b). DOM also occurs in Hindi, illustrated in example (2c,d) in which the post-position “ko” must be used with human, specific objects.

In Basque, DOM takes the form of using the dative case to mark animate direct objects in spoken Basque, a pattern which as been argued to result from convergence with Spanish (Austin, 2006; Rodríguez-Ordóñez, 2013). The Spanish spoken in the Basque Country is a dialect with DOM in which dative clitics rather than accusative ones are used to refer to animate direct objects (Fernández-Ordóñez, 1999), as seen in example (3):

DOM in spoken Basque can be seen in example (4)a. In this sentence, dative verbal inflection and case are used to mark an animate direct object, instead of the standard absolutive inflection as in (4)b.

The two forms shown in (4) a,b meaning “I have heard you” co-exist in the dialect of Basque spoken in the Spanish Basque Country, but not in the Basque spoken in France, where only the standard form (4)b is used. In both dialects utterance (4)b is grammatical only if there is a previously mentioned, unspecified direct object, meaning something like “I heard it from you.”

A corpus of contemporary adult spoken Basque collected in the Spanish Basque Country contained examples of DOM (Hualde et al., 1994). Similar examples appear in natural speech corpora from children acquiring Basque as an L1 or bilingually (Barreña, 1995; Ezeizabarrena, 1996; Austin, 2001). In these studies, children produced dative case with human direct objects when using verbs such as jo “to hit,” lotu “to tie up,” and harrapatu “to catch.” Barreña notes that examples of DOM are also present in adult speech, and Ezeizabarrena mentions that exchanges such as the one in example (5) are common, in which parents are teachers correct children's use of DOM in Basque:

Austin (2006) argued that DOM in Basque is a contact-induced change that is the result of convergence between Basque and Spanish leísmo, or the use of the dative case for animate, specific direct objects. In that paper, I suggested that DOM in Basque is a change in progress that has been accelerated by the bilingualism of Basque speakers, as well as by the large number of L2 speakers of Basque who are dominant in Spanish. Additionally, I suggested that several language-internal factors made the Basque case system susceptible to influence from Spanish in the transfer of DOM, including (1) a diachronic tendency to substitute dative for absolutive agreement in Spanish Basque; (2) language-internal trend toward syncretism or agreement simplification; and (3) the pro-drop nature of Basque, which may lead to reanalysis of agreement morphemes on the part of the learner for sentences such as (6) when it is not clear whether the 3rd person dative argument refers to the indirect or the direct object.

Rodríguez-Ordóñez (2013) examined the extent to which adult speakers of Gernika Basque found the use of DOM acceptable. Using a grammaticality judgment task, she asked 5 Spanish-dominant and 11 Basque-dominant participants to rate transitive sentences with and without DOM using a Likert scale. She found that both groups of speakers were significantly more likely to rate the non-DOM sentences higher than the DOM sentences. Although the Spanish-dominant bilinguals were more likely than the Basque-dominant ones to accept the DOM sentences, there was not a significant difference in judgments between the two groups. An analysis of verb types found that DOM was significantly more likely to appear with certain verbs (such as ikutu ′to touch) than others, and that it was favored in sentences with null dative objects.

In a second experiment, Rodríguez-Ordóñez asked adult speakers of the Gernika dialect of Basque from three different age groups (18–25, 30–45, and 50–65 years old) to rate speech samples from Gernika Basque, Standard Basque and Spanish for “likeability” and “Basqueness.” In each dialect or language, participants heard DOM and non-DOM transitive sentences. Across age groups, sentences from Gernika Basque (with DOM and without) were judged “more Basque” than comparable samples from Standard Basque or Spanish. However, Standard Basque sentences with DOM were judged “less Basque” than non-DOM sentences. In judging “likeability,” an interesting contrast emerged between older and younger participants; whereas the younger speakers judged Gernika Basque samples with DOM to be much less “likeable” than non-DOM samples, older participants did not distinguish between DOM and non-DOM samples from Gernika Basque. Rodríguez-Ordóñez interprets this finding to indicate that younger speakers of Gernika Basque stigmatize DOM to a greater extent than older ones, perhaps due to their receiving literacy instruction in Basque, and having been taught to avoid erderakadak or “Spanishisms.” This hypothesis was supported by data that Rodríguez-Ordóñez collected in interviews with adult speakers of Basque, in which she asked them whether using DOM in Basque was acceptable. She found that DOM has highly stigmatized by bilinguals who were Basque-dominant, whereas Spanish-dominant speakers admitted to using DOM in Basque and were more accepting of its use.

A second syntactic characteristic which distinguishes Basque from Spanish is the order of constituents in a sentence. Basque is a head-final language in which verbs, complementizers such as relative pronouns, determiners and post-positions always appear on the right side of the phrase or the sentence. For example, Basque sentences have an unmarked subject-object-verb order such as the example in (7):

Notice that in addition to the object preceding the verb, the verb precedes the auxiliary. In Spanish, the order of these constituents is the opposite; the direct object follows the verb, and the verb follows the auxiliary, as in (8):

This pattern of the head of a phrase on the right for Basque and on the left for Spanish is repeated recursively to form all the elements of a sentence. The different position of the inflection node (INFL) in each language reflects the fact that the Basque verb is inflected on the right, and the Spanish one on the left, as seen in examples (9)a, b:

Similarly, embedded clauses in Basque are situated to the left of the embedding element, whereas in Spanish, they are on the right, as shown in (10):

Basque complementizers are also head-final, such as the morphemes -ela “that” and -elako “because” that attach to the final verb of a clause, as seen in examples (11) a and b:

In Austin (2001), I found examples of children producing head-initial complementizers in Basque such as (12) and (13)a and b:

I interpreted examples like these to be these to be instances of transfer from Spanish, which has preverbal complementizers, such as those in (14).

However, to prove that these examples stem from cross-linguistic transfer from Spanish, one would need to see whether monolingual children acquiring Basque also produce them, although they are unattested in previous research (Kintana, 2011). If they did, then the production of preverbal complementizers would presumably be a stage that children pass through while learning Basque, rather than the result of bilingual influence from Spanish.

This paper addresses how age and bilingualism affect the production of DOM and preverbal complementizers in Basque-speaking children and adults, phenomena that have been argued to result from contact with Spanish (Austin, 2001, 2006; Rodríguez-Ordóñez, 2013). The research questions under investigation include:

The experimental methods and information about the participants are described in the following section.

The data for this paper were collected from 20 bilingual children learning Basque and Spanish simultaneously and from 11 monolingual children acquiring Basque. All the participants were living in the Spanish Basque Country at the time of data collection. Children whose parents signed consent forms were selected to participate in the study; the adult participants also signed consent forms. All the consent forms and the study protocol were reviewed and approved by the Cornell University Institutional Review Board. The children who participated in this study spoke Standard (Unified) Basque, but some participants also spoke Bizkaian or Gipuzkoan Basque dialects, and some features from these dialects appeared in their speech. All the bilingual children spoke standard Northern Peninsular Spanish, and were recruited through their schools or through acquaintances of the author. In parental and teacher reports the children were identified as bilingual, simultaneous acquirers of Spanish and Basque. Most children produced many more utterances in Basque than in Spanish. This is most likely due to the fact that most children were interviewed at school, a primarily Basque-speaking environment for many of them, which led them to feel more comfortable speaking Basque there.

The bilingual children who participated in this study were recruited from schools which taught in Basque and Spanish equally (model B) or for which Basque was used exclusively by teachers in talking to children (model D). Table 2 provides information about the children's linguistic backgrounds, including the school models that they attended and their parents' language use.

The children's ages were between two and three and a half years old. This age range was chosen because it is the point at which verbal inflection and case develop rapidly, and also it allowed for comparison with studies conducted previously (Barreña, 1995; Ezeizabarrena and Larrañaga, 1996).

The results from adult speakers of Basque analyzed in Austin (2006) are shown in Table 5. Overall, adult participants produced DOM in 18.3% production of possible contexts with human objects, bilingual children produced DOM in 33% of possible contexts, and monolingual children produced utterances with DOM in 43% of possible contexts.

As seen in Table 5, there was individual variation in the use of DOM by the adult speakers.

In natural speech transcripts from 4 adult speakers (9 h total), DOM was produced in 13/71 cases (18%). There were fewer cases of DOM produced by the two adult participants whose first language was Basque than the two participants who had learned Basque as a second language, as seen in Table 4.

Adults used DOM with the following verbs: entzun “to hear,” jo “to hit,” jarri “to put,” utzi “to allow” and molestatu “to bother.” Examples are provided in (16) a–c.

Table 6 shows the verbs which appeared with DOM in the adults' speech.

In most cases, adults' utterances with transitive verbs that take human objects did not trigger DOM; some of the verbs consistently used with standard transitive agreement (non-DOM) included ikusi “to see,” ezagutu “to know,” maite “to love,” harrapatu “to catch” and salbatu “to rescue.” It seems that animacy triggers the use of Basque DOM, like Spanish leísmo, rather than reflecting a particular verb's subcategorization for a lexically-specified quirky case.

The bilingual children produced utterances in Basque with DOM in 16/48 possible contexts, or 33% of the time on average, as seen in Table 7.

The verbs used by bilingual children with DOM are shown in Table 8.

Examples of utterances with DOM produced by bilingual children are provided in (17) a,b.

In Spanish, instances of DOM were considered to be those in which the child used either the differential object marker “a” or a dative clitic with animate, direct objects. Bilingual children began using DOM in Spanish at around 2;06 years, the same time at which it emerged in Basque, as seen in Table 9.

An example of the use of DOM in Spanish can be seen in example (18):

The monolingual children used DOM in 16/37 or 43% of possible contexts, as shown in Table 10.

The verbs used by monolingual children with DOM can be seen in Table 11.

While the monolingual children used DOM with many of the same verbs as the adults and bilingual children, it is interesting to note that they used DOM with a greater number of different verbs than either of the other participant groups. Some examples of DOM produced by the monolingual children can be seen in (19):

Preverbal complementizers in Basque were produced by 4 bilingual children, as seen in Table 12.

Preverbal complemetizers were used by 4 children between the ages of 2;08 and 3;02, as shown in the following examples in (20) and (21) (non-target utterances are marked with an asterisk):

One child, (IA, 3;02) was able to use a target-like post-verbal morpheme for one complementizer (ela “that”) as seen in (22)a, but used a non-target preverbal complementizer for -elako “because,” as shown in (22)b.

All the adults and the monolingual Basque-speaking children used standard post-verbal complementizers exclusively, as seen in example (23).

These examples were rare in the adult input. In 1488 utterances containing clauses from Basque-speaking adults, I found 7 uses of the -elako “because” morpheme, and 8 examples of the -ela “that” complementizer.

In Spanish, pre-verbal complementizers were produced by adults and two of the bilingual children (AR and IA), whose utterances with preverbal complentizers can be seen in (24):

While there is good reason to think that both DOM and pre-verbal complementizers in Basque are morphosyntactic outcomes of contact with Spanish, children's production of the two phenomena is quite different, in part reflecting their presence or absence in the adult input. DOM is used by adults as well as children, supporting the hypothesis that it is a change in progress occurring in Basque which results from morphosyntactic convergence with Spanish (Austin, 2006), consistent with the predictions of Sánchez's Functional Convergence Hypothesis (2003, 2004). The finding that DOM is produced more by Spanish-dominant bilingual adults and that those bilinguals find its use more acceptable than Basque-dominant ones is also consistent with this analysis (Rodríguez-Ordóñez, 2013).

Monolingual children produced utterances with DOM to a greater degree (43%) than bilingual children (33%) or adults (18%), suggesting the possibility that both groups of children are not merely matching the patterns present in the adult input. Rather, their behavior may reflect a tendency to generalize or regularize inconsistent patterns, as has been found to be the case for children acquiring an artificial grammar (Kam and Newport, 2005). Of the monolinguals, 4/11 children over the age of 3:00 used DOM in Basque, compared to 7/20 bilinguals. The bilingual children began using DOM earlier, at age 2;06, the same age at which DOM emerged in their Spanish. Because the number of productions of DOM in this study is quite small, more data would have to be collected to evaluate this hypothesis more carefully. It is possible, for example, that bilingual and monolingual children receive input in Basque that differs with respect to the presence of DOM; this could be true, for instance, if bilingual children receive greater exposure to Basque at school than home, and teachers at school use less DOM than the parents of monolingual Basque-speaking children do⁴.

While the finding that monolingual children produced more DOM than bilingual children was unexpected, a Chi-square test indicated that this difference between the groups of children was not significant. The monolingual children's greater use of DOM was a surprise, given than use of DOM is correlated with dominance in Spanish in bilingual adults. Perhaps monolingual children's use of DOM reflects that they are more sensitive to case-marking as a syntactic cue in Basque than bilingual children; exploiting case as a cue in understanding grammatical relations has been argued to develop over time in multilingual children acquiring Light Warlpiri and Lajamanu Warlpiri, for example (O'Shannessy, 2011). Of the 7 children who produced DOM in Basque, 6 attended Basque-only schools (model D, as seen in Table 2), suggesting that for children, more exposure to Basque may correlate with greater use of DOM.

The use of pre-verbal complementizers presents a very different developmental pattern. These forms are used exclusively by four bilingual children between the ages 2;08 and 3;02, and were never produced by monolingual children or adults. Five bilingual children in this age range never used them at all, and their production does not seem to be correlated with their MLU in Basque. I interpret this variability in their use to indicate that children may use pre-verbal complementizers in Basque as a relief strategy until they have acquired the post-verbal complementizers in Basque, which are low-frequency morphemes. However, the variability in the use of pre-verbal complementizers between children suggests that the transfer of this feature from Spanish to Basque is not a universal stage in bilingual Basque/Spanish development. Furthermore, these preverbal complementizers do not seem to reflect a more fundamental type of cross-linguistic influence such as the wholesale transfer of the head-initial properties from Spanish to Basque, since bilingual children's grammars in other ways are consistent with the head-final characteristics of Basque. Rather, I understand these utterances to be a type of temporary relief strategy which may be used by some bilingual children when they are confronted with a construction that they have not yet acquired, following proposals by Gawlitzek-Maiwald and Rosemary (1996) and Bernardini and Schlyter (2004). In Gawlitzek-Maiwald and Rosemary (1996) longitudinal study of a bilingual child learning German and English, they proposed that the child was transferring a German IP to English because she was more grammatically advanced in German than English, as seen in example (25):

The child produced mixed-IP utterances such as the one in (26) until she was 2;09 years, at which point the authors claimed she acquired an IP in English too. Gawlitzek-Maiwald and Rosemary referred to the use of the L1 functors in the L2 as bilingual bootstrapping, which they argued was a kind of relief strategy for filling in syntactic gaps in the bilingual child's weaker language. Bernardini and Schlyter (2004) came up with a similar proposal, the Ivy Hypothesis, in which they suggested that child bilinguals can use their stronger language as a kind of syntactic scaffolding for the weaker one. They also claimed that syntactic transfer from the stronger to the weaker language is possible in all areas of grammar.

The results from this study, in particular, the finding that preverbal complementizers are only used by bilingual children as well as the fact that they disappear as target-like post-verbal forms begin to be produced supports the hypothesis that preverbal complementizers are used temporarily in bilingual development as a relief strategy. However, while the result from this study are suggestive, there are only a few data points from a few children with regard to pre-verbal complementizers (which were only used by 4 out of 20 bilingual children and none of the monolingual children). Longitudinal data would be needed to determine whether the use of pre-verbal complementizers in Basque is in fact a type of short-term transfer similar to the temporary cross-linguistic influence that has been found in bilingual children acquiring many different language pairs (Döpke, 1998; Yip and Matthews, 2000; Müller and Hulk, 2001; Paradis and Navarro, 2003, inter alia). In contrast, the use of DOM seems to be a change in progress that is spreading through the Basque language, as evidenced by the finding that both monolingual and bilingual children use DOM more often than adults and with a greater range of verbs. It is possible that if adult speech is inconsistent with regard to the use of DOM in verbs such as ulertu “to understand” that children may adopt a strategy of agreement simplification, such as using dative agreement with all human direct objects, in line with the findings of Kam and Newport (2005). In this regard, child learners could play as important a role as second language speakers of Basque in promoting DOM.