Overt subjects in early Basque and other null subject languages

Abstract

This study focuses on person marking in early Basque and other null subject languages. From very early on, person marking on the verb and quite regular, adult-like, null subject rates are attested across early acquisition studies on genetically related and unrelated pro-drop languages. We survey several studies on bilingual children simultaneously acquiring two languages with the opposite value for the pro-drop parameter. The conclusion drawn is that children display a monolingual-like pattern in the production of person marking, overt subjects, and personal pronouns in the null subject language, whereas, in the non-null subject language, bilinguals evidence delayed target person marking and overt subject production. These data are compatible with the view that children correctly set the default parametric value at early stages and separate the languages being acquired. However, it is argued that accounts based on the lexical learning of features in the functional category T(ense) may better account for the crosslinguistic data. Moreover, the accuracy observed in overt person inflection production leads to the proposal that (the spelling out of) the [person] or [D] feature in T(ense) is the first subject feature available to the language acquirer, previous to other candidates such as number or case.

Keywords

agreement Basque cL2 2L1 ergative null subject overt subject person

Introduction

The study of null subjects (NSs) has received a great deal of attention in the literature on early monolingual and bilingual language acquisition in recent decades. However, less attention has been paid to the production of overt subjects (OSs) in the earliest stages of the acquisition of NS languages. This research aims to fill this gap by providing a summary of OS production data from several longitudinal case studies as well as some cross-sectional studies on the spontaneous production of Basque as only first language (L1), first language in simultaneous bilingualism (2L1), and as child L2 (henceforth cL2) following Meisel (2007; 2011). In addition, the results of several longitudinal studies on early acquisition of consistent and partial NS languages such as Catalan, Spanish, oral French, and Brazilian Portuguese (BP) are discussed in order to highlight the existence of different developmental patterns with regard to the production of OSs in NS (no or slight increase of OSs during the first years) and non-NS languages (noticeable increase of OSs). Additionally, data obtained from bilingual children growing up in different language contact situations offer clear evidence for the autonomous development of the production of OSs in each language. All the results suggest that the early identification of person (and D) features allows children acquiring NS languages to have a finite T(ense) containing a [+D] feature in their grammars from the very beginning, which licenses NSs (mostly referential). Furthermore, such data reflect an adult-like production of OSs in NS languages, which contrasts with the less target-like production of OSs in developing non-NS languages. Slight differences observed across NS languages seem to be related to the morphophonological specificities of particular languages, including the ones that distinguish consistent from partial NS languages.

NSs are unpronounced referential forms which, though abstract, are present in the syntactic positions that an OS occupies. Two kinds of languages are distinguished by the pro-drop parameter: pro-drop or NS languages, which license NSs, and non-pro-drop or non-NS languages, which do not allow NSs in root sentences.

Moreover, two mechanisms have been described in the literature for licensing NSs: rich or regular agreement and D-linking. Typically, the first mechanism has been evoked to explain the grammaticality of NSs in finite clauses in languages with rich person marking (Rizzi, 1982) and in languages without person marking (Jaeggli & Safir, 1989). The second served to explain the grammaticality of nonovert arguments in languages with verbs lacking person marking, namely, Chinese clauses (Huang, 1984) or nonfinite constructions in languages that otherwise have rich person marking, such as Basque (Elordieta, 2001). In this context, subject person marking on verbs is considered to be the main licensing mechanism of NSs in languages with person marking on verb. See the study by Zushi (2003) for an extended discussion on the topic.

A more recent approach to crosslinguistic variation in the production of NSs assumes that one of the parameters involved in the production of subject pronouns is related to “whether T does or does not host an inherent valued feature [D]” (Holmberg, Nayudu & Sheehan, 2009, p. 67). These authors reformulate Rizzi’s (1982) account of the NS parameter based on either having or not having a [+referential] feature in INFL(ection), as follows:

Instead of a valued D(efiniteness)-feature, T has an unvalued D-feature in consistent NS-languages, which is valued either by the subject, that is if the subject is a DP marked for (in-)definiteness, or by a null topic in SpecCP.

Three kinds of languages are distinguished under this approach: non-NS languages, partial NS languages, and consistent NS languages. According to Holmberg et al.’s (2009) classification, T hosts a D-feature only in consistent NS languages like Italian but not in partial NS languages like Finnish and Brazilian Portuguese or in non-NS languages like English.

Many longitudinal studies on the early production of subjects in different NS languages have concluded that children know of the possibility of subject omissions from a very early age. These results are compatible with both the Continuity Hypothesis, which sees children’s grammars as converging with adult grammars, and with Very Early Parameter Setting (Wexler, 1998). However, crosslinguistic differences have been observed between NS and non-NS languages in the sense that children acquiring non-NS languages need more time to learn to produce target-like OSs (Hyams, 1986). Comparatively, it is quite widely attested that learning the inflectional system of languages with poor or irregular person marking takes children quite a long time, while those acquiring languages with rich person marking start producing inflected verb forms and target-like NSs and OSs very early on. The differences observed between both kinds of languages were consistent with an explanation in terms of the default parameter setting, which assumes that children acquiring pro-drop languages do not need much exposure to set the correct (+) value for the pro-drop parameter, whereas children acquiring non-pro-drop languages (− value) must undergo an initial pro-drop-like stage characterized by several phenomena connected to the target-dislike functional category INFL (or T(ense), including the production of NSs, root infinitives (RIs), and verb final syntactic word orders (in verb-second (V2) languages) in finite contexts. Target-deviant productions are assumed to disappear in the stage following the consolidation of inflectional phrase (IP), coinciding with the age of target production of person marking on the verb inflection (Clahsen, 1986; Meisel, 1994a; Meisel & Müller, 1992). Parameter setting has been understood as a quasi-automatic process initiated by a (preferably unique) trigger detected in the grammatical structure assigned to the data. Under this account, the possibility of incorrect decisions was not excluded, in exceptional cases, and it was predicted that conflicting evidence concerning the value to be chosen could serve as a trigger to differentiate two grammatical systems (Meisel, 1995, p. 29).

In NS languages with uniform and rich person morphology (Jaeggli & Safir, 1989), two possibilities exist for subject expression in adult languages, since subjects can be expressed as overt arguments and/or by the production of person inflection on verbs. Similarly, child language studies report the presence of some OSs as well as quasi-errorless subject–verb agreement in many pro-drop corpora, where inflected verbs are attested to agree with their subject in person and number features from very early on — around the 22nd–30th month, immediately after or simultaneous with a short stage of production of pseudoverbs, rote-learned forms, and copulas (Barreña, 1995; Bel, 2002; Ezeizabarrena, 1996; Guasti, 1993/1994; Larrañaga, 2000). In general, very few errors are found in longitudinal corpora — most of which involve number features, while person errors are virtually absent — indicating that subject–verb agreement is active and that child sentences project a TP (Tense Phrase) before age 3. On the other hand, OSs are present from the very beginning, though their distribution in finite and nonfinite sentences varies across languages (Bel, 2002, 2003; Liceras, Bel, & Perales, 2006; Silva-Corvalán & Sánchez-Walker, 2007).¹

In addition to the nature of the “initial” or “earliest” stage and the (as)synchrony in setting the target value for the corresponding parameter in the two languages of the bilingual, the contact situation offers a clear opportunity for testing other hypotheses such as those concerned with possible crosslinguistic influence.

Following Hulk and Müller (2000) and Müller and Hulk (2001), the syntax-pragmatics interface is one of the domains in which crosslinguistic influence is expected, especially in structures where the languages in contact overlap at the surface level. The overproduction of OSs attested in several NS languages (OS optional) in contact with a non-NS language (OS compulsory) seems to confirm such a prediction. For instance, overproduction of OSs or higher acceptance rates of OSs in some comprehension tests and spontaneous production by bilinguals have been attested in different types of developing grammars, such as in early bilingual language acquisition (Haznedar, 2010; Serratrice, Sorace, & Paoli, 2004) and adult L2 acquisition (Sorace & Filiaci, 2006) of a NS language in contact with a non-NS language. In this context, the asymmetries observed between OSs (target deviant) and NSs (target like) have been explained in terms of the Interface Hypothesis (Sorace & Filiaci, 2006), which states that interface properties involving syntax and another cognitive domain, like pragmatics may not yet be acquired at the time when narrow syntactic properties are already in place. Under this account, learners exposed to two conflicting systems are expected to show a longer delay in the acquisition of peripheric phenomena than narrow-syntax phenomena.

But some controversy arises in the literature with regard to the interlinguistic influence of the non-NS on the NS language. Sorace (2005) reported on bilinguals of two NS languages who also had a higher acceptance of OSs than monolinguals and argued that overproduction may be better attributed to some extended default strategy in a situation of processing overload than to interlinguistic influence. One more piece of evidence in the ongoing debate is the data from early simultaneous English–Spanish bilingual twins studied by Liceras, Fernández-Fuertes and Pérez-Tattam (2008) and Liceras, Fernández-Fuertes & Alba de la Fuente (2011). These authors did not find any evidence of interlinguistic influence of the non-NS language on the NS one when it came to the production of OSs in Spanish. This conclusion was supported by the observations that OS production by the bilinguals in their study developed separately in both languages and that the developmental pattern for Spanish OSs was similar to the one observed in monolingual corpora. A more detailed analysis of the data led Liceras et al. (2011) to observe that the twins’ data showed the opposite directionality (Spanish on English) with respect to the one predicted by Hulk and Müller’s (2000) and Müller and Hulk’s (2001) account of interlinguistic directionality (the one option language on the two option language).

So far, this article has focused on the early production of subjects —rather than their omission—in Basque and several Romance languages. We will attempt to show that (a) children are very sensitive to the person feature from the very beginning; (b) the overt expression of person in early languages depends on the specific properties (rich verb inflection, rich pronominal system) of the language involved. Thus, consistent NS languages with rich person marking show a developmental pattern in the production of OSs (regular, quite constant, production of OSs), which differs from the consistent non-NS languages (increase in the production of pronominal subject); and (c) the pattern observed in NS languages is truly robust, regardless of whether the language is acquired alone or simultaneously with another (non)-NS language.

The article is organized as follows: subject marking in Basque will be described in section “Basque: A NS language”. Next two sections present data from early monolingual and bilingual children acquiring (at least) one NS language obtained in different studies on monolingual (L1) and bilingual (2L1, cL2) corpora: early Basque data in section “Subjects in child Basque: L1, 2L1 and cL2” data of other NS languages such as Spanish, Portuguese or Inuit in section “OSs in the early stages of acquisition of other languages”. Data Developmental differences attested across NS languages and are discussed together with the potential effect of interlinguistic influence in section “Discussion” preceding the concluding remarks.

Basque: A NS language

Basque is a head final language with subject–object–verb (SOV) as basic word order, and with a very rich agreement morphology on the verb, which shares person and number features with its subject, direct object (DO), and indirect object (IO) arguments. Traditionally, this richness has been taken as the licenser of null arguments (Rizzi, 1986) in each of the three corresponding syntactic positions. Furthermore, Basque is a language with ergative subject–verb morphology. This means that subjects of both (di)transitive (1a and c) and unergative verbs (1d) are marked with the overt -k suffix of E(rgative) case, in contrast to subjects of unaccusative verbs (1b) and DOs of transitive verbs (1a and c), which bear the phonologically null A(bsolutive) case. Subjects and objects can be omitted in both finite and nonfinite clauses (as indicated by the parentheses in the examples) without affecting the grammaticality of the sentence (1). Subjects are assumed to raise first to the Specifier position of the vP phrase (SpecvP) in order to check case and then to Tense Phrase (TP) in order to satisfy the extended projection principle (EPP) requirement, regardless of whether they are A or E marked morphologically. See also Artiagoitia (2000) for a summary of the different proposals on the derivation of subjects in Basque.

(1) a. (ume-a-k) (ni-Ø) ikus-i n-au-Ø

child-Det-E I-A see-PF A1s-have-E3s

the child saw me

b. (ume-a-Ø) etorr-i da

child-Det-A come-PF be.A3s

the child came

c. (ni-k) (txori-a-Ø) (ume-a-ri) erakuts-i di-o-t

I-E bird-Det-A child-Det-D show-PF have.A3s-D3s-E1s

I showed the bird to the child

d. (txori-a-k) etsi du-Ø

bird-Det-E resign-PF have.A3s-E3s

the bird resigned

e. (ni-k) (hanka-a-Ø) hautsi eta (ni-Ø) ezin ibili orain

I-E leg-Det-A break and I-A not.be.able walk now

I broke my leg and am not able to walk

Basque is a language with very rich overt person and number marking morphology on the verb. The verb agrees overtly with all its arguments: subjects can be case marked with A (when accompanying unaccusative verbs (1b and 1e)), or E (with a transitive (1a, 1c, and 1e), or an unergative verb (1d)); DO with A (1a and 1c); and IO with D(ative) (1c). Person marking is rich and regular in this language (in the sense of Jaeggli & Safir, 1989), with three different affixes expressing the three different grammatical persons (Table 1). Moreover, agreement in the singular is marked differently than the plural by the morphology: E1s -t versus E1pl -gu and E2s -zu versus E2pl -zue. Depending on the valence of the verb, tense, and so on, agreement affixes may be attached to the inflectional root as prefixes or suffixes, as indicated in Table 1.

Table 1.

Personal pronouns and subject affixes in Basque.

Pronouns	Person	Prefix	Suffix
Pronouns	Person	S of intransitive V and DO (all tenses); S of transitive V (past tense)	S of transitive V (present tense)
Ni	1s	n-	-t
Zu	2s	z-	-zu
—	3s	Ø-	-Ø
Gu	1pl	g … de/z	-gu
Zuek	2pl	z … te	-zue
—	3pl	Ø … de/te/z	-e

Basque lacks third-person pronouns, which means that person marking is typically expressed in the sentence by the four first and second overt personal pronouns —1s ni, 2s zu, 1pl gu, and 2 pl zuek — though demonstratives like hura “that” or the “quasi pronoun” bera “the same, himself, he” have been considered to act as third-person pronouns in some cases (Trask, 2003). See Iraola and Ezeizabarrena (2011) for a discussion of bera’s and hura’s anaphoric properties in adult and child language.

Moreover, agreement affixes are almost homophonous with their corresponding personal pronouns, which seems compatible with the idea that agreement affixes are pronominal, comparable to weak pronouns in languages like French (Kato, 1999). In what follows, the description of person marking and agreement forms will be limited to subject–verb agreement for the purpose of simplification, but it must be kept in mind that the person affixes on the verb are basically the same for agreement with the subject and the DOs (Table 1) and for IOs.

Since licensed empty categories appear in Basque in finite as well as in nonfinite sentences, two different mechanisms have been proposed for licensing dropped arguments: rich agreement for null pronominals in finite clauses and topic D-linking for null pronominals in nonfinite clauses (Elordieta, 2001). More recently, Duguine (2010) presents a unified account for both kinds of licensing of null arguments, proposing the Argument Ellipsis Condition based on the assumption that null arguments are determiner phrases (DPs) not spelled out at PF (2).

(2) The argument ellipsis condition (translated from Duguine, 2010, p. 271).

A DP can be elided iff it is linked (and if there are linking “traces” at PF,namely, p-features and indexation)

In fact, since the NS is the preferred option in neutral sentences (3a), the phenomenon cannot be considered an “idiosyncratic option.” Most OSs in Basque, especially pronominals, appear as (more or less) topicalized elements (3b), and very often, but not always, have a contrastive (3c) or a focusing interpretation (3d and 3e). This alternation can be observed with finite (4a, 4c, 5a and 5b) as well as with nonfinite verb forms (4b).

(3) a. ekarr-i du-t lore-a-Ø

bring-PF have.A3s-E1s flower-Det-A

(I) brought the flower

b. ni-k(,) ekarr-i du-t lore-a-Ø

I-E bring-PF have.A3s-E1s flower-Det-A

Me, I brought the flower

c. NI-K/__* ekarr-i du-t, ez Jonek

I-E bring-PF have.A3s-E1s no John-E

I brought it, not JOHN

d. NI-K/__* ekarr-i du-t

I-E bring-PF have.A3s-E1s

(It’s) me (who) brought it

(4) a. *__/ZUK bazenekien

you/YOU knew it

b. Nahi du-Ø __/ gu joa-te-a-Ø

want have.A3s-E3s we go-NOMINALIZATION-Det-A

He wants us or somebody else/US to go

c. Jon-ek_i esan-Ø du-Ø ___i/j etorr-i de-la

Jon-E say-PF have.A3s-E3s come-PF be.A3s-that

John said that he/somebody else came

Similarly to what has been observed for Spanish (Fernández Soriano, 1999), there are many contexts in Basque where null arguments are either the preferred or the only option. Examples include impersonal-like sentences (5c), the maintaining of unambiguous discourse (5d), and expletive contexts such as meteorological predicates (5e). The conclusion is that the occurrence of NSs does not always reduce to optionality.

(5) a. __ esan du petrolioa garestitzera doala

(s)he said that the price of oil is going to go up

b. berak/hark esan du petrolioa garestitzera doala

(s)he/that said that the price of oil is going to go up

c. __ esan dute petrolioa garestitzera doala

They (indefinite or referential) said that the price of oil is going to go up’

d. Jon bizilaguna dut. (__/*bera/?bizilaguna) matematikako irakaslea da, baina (__/*bera) filosofiarekin interesatuta dago, (__/*berak) neskalaguna filosofoa duelako

John is my neighbour. (__/*He/?my neighbour) is a teacher of mathematics but (__/*he) is interested in philosophy because (__/*he) has a girlfriend (who is a) philosopher

e. bart gauean __/*berak/*hark izotza bota du

It/*he/*that froze last night (= There was a frost)

The features of Basque appear to be compatible with those of consistent NS languages (Holmberg et al., 2009). For instance, the embedded NSs in (4c) can — but does not have to — be interpreted as coindexed with John, whereas in partial NS languages like BP or Finnish, the null pronoun could only be interpreted as John, and a topic shift would require an overt pronoun. Moreover, a NS like in 5a, which is interpreted as referential in Basque, would receive an indefinite (arbitrary) interpretation in partial NS languages. In contrast, the indefinite reading in Basque requires some extra morphological marking on verbs, as illustrated by the E3pl suffix -te in dute (5c). Thus, Basque fits in with the consistent NS languages (6).

(6) [Contrary to partial NS languages …] Consistent NS languages have an arbitrary null subject (null “they”) but to express a generic subject pronoun, they resort to some overt strategy. (Holmberg et al., 2009, p. 64)

Subjects in child Basque: L1, 2L1 and cL2

The data analyzed in this article come from two kinds of oral samples, all of them video recorded: (a) longitudinal case studies of spontaneous production and (b) elicited cross-sectional narratives. Only the non-imitative utterances containing a verb in both sets of samples were included in the study.

The data in the longitudinal case studies were obtained in sessions of spontaneous conversations in a home environment every 2–4 weeks from 18 months of age, up to age 4. The children interacted with their parents in natural situations such as having lunch, playing with toys, or talking about story books. In our discussion, the spontaneous Basque of several monolingual and early balanced bilingual children is described and analyzed with different levels of detail. Data from the monolingual boy (L1), Oitz (O), and one male early balanced Basque–Spanish bilingual child (2L1), Mikel (M), will represent the more detailed part of the empirical data. Moreover, results from previous studies regarding one additional bilingual boy (Jurgi (J)) and a monolingual female (Bianditz (B)) will be discussed for illustration (Barreña & Zubiri, 2000; Ezeizabarrena, 2002, 2003). The two 2L1 children were exposed to Basque and Spanish from birth at a similar rate (Almgren & Barreña, 2001; Barreña, 1995; Ezeizabarrena, 1996). Data and rates of OS marking in both L1 and 2L1 populations will be compared.

The cross-sectional data were collected at two ikastola-s or Basque immersion schools. The first one is found in a locality where Basque is the dominant language in social interactions in the neighborhood, whereas the second is located in a mostly monolingual Spanish-speaking area in which the use of Basque is generally restricted to the school context. Two groups of Basque–Spanish bilingual children participated in the study (L1 and cL2). The 12 children in the L1 group are being raised in Basque-speaking families, living in a Basque-dominant sociolinguistic environment. The 12 children composing the second group were exposed to Basque very early through attending a Basque immersion school from ages 2 to 3, but they are members of monolingual Spanish-speaking families living in a monolingual Spanish area. The latter are a subset of the sample collected by Beloki and Manterola (2007). Since their exposure to and use of Basque are mostly restricted to the school environment, they are considered to exemplify child L2 acquisition, abbreviated as cL2 (Meisel, 2007; 2011).

The same procedure was used with both groups in the cross-sectional study. First, an adult told the story called Mattin Zaku to a group of children with the support of pictures. The cross-sectional data were obtained in recordings made during the narrative task in which each child retold the story to another classmate, with the support of the pictures used by the adult. This procedure was followed during two or three sessions with each child. The first session took place when the children were 5 years old, the second session (only for the cL2 group) took place 6 months later, and the last one took place at age 8. See Manterola (2011) for details.

OSs in longitudinal case studies

Instances of overt nominal subjects are present from the earliest productions in monolingual corpora in both finite (7a and 7c) and nonfinite sentences (7b), as shown in the examples produced by the female Bianditz and the male Oitz monolinguals, respectively. Though ergative case is frequently omitted during the earliest stages, instances of zero marked A (7a and 7c) and E case -k (7b) are both attested very early on.

(7) a. Hemen da kutxare-a-Ø (B 1;07) (Barreña & Zubiri, 2000)

Here is spoon-Det-A

Here is the spoon

b. Bian(d)itz-ek bota (B 1;08) (Barreña & Zubiri, 2000)

Bianditz-E throw

Bianditz throw it

c. ez dau zezen-a-Ø (O 2;00) (Barreña & Zubiri, 2000)

no is bull-Det-A

the bull is not there

Note that such examples are attested even some months before the productive marking of person agreement morphology on the verb (8). Age of productive marking is established as the age at which at least two different person markings are produced attached to the same verb root and/or the same accurate overt person marking is attached to two different roots.

(8) Age of productive person marking on V by monolinguals

Oitz 2;03 MLU 2.36

Bianditz 1;09 MLU 1.77

OSs are attested from the first recording onward (see numbers under the columns in Figures 1 and 2).² After age 1;11 (mean length of utterance (MLU) = 1.7), a rate of around 40% remains quite constant throughout the following studied period (MLU = 1.7–3.8), and they are even attested in Oitz’s earlier recordings from 1;07 (MLU = 1.2), with the exception of age 1;10, at which time only nonfinite verbs are attested (n = 8) in the sample, none containing OSs. Thus, no clear developmental pattern is seen in this respect in the longitudinal data of the monolingual male child Oitz.

Figure 1.

Rates of OS in Basque sentences produced by the monolingual child Oitz.

Figure 2.

Rates of OS in finite and nonfinite Basque sentences produced by the monolingual child Oitz.

The variation in frequencies of OS in Figure 2 is clearly related to the increasing production of finite verb forms rather than to the increase in the production of OSs. Similar mean rates of OSs were observed with finite verbs (mean = 44.2%) and with nonfinite verbs (mean = 43.5%) throughout the studied period from age 2 onward (Figure 2). OSs are attested in Oitz’s corpus earlier (1;11 in (10)) than the productive use of person marking on the verb (2;03) and simultaneously with — or even earlier than — the production of the first finite forms at age 2;00 (Figure 2). Early production of subjects has also been reported for other monolingual and bilingual children as well.

Looking at the data in Table 2, it must be pointed out that the bilingual data (Mikel and Jurgi) correspond to a longer longitudinal study (extending to age 4) than that reported for the monolingual Oitz (1;07–2;06) in Figures 1 and 2. This could be one of the reasons for a less proportional distribution of finite verbs (>95%) and nonfinite (target + target deviant) verbs forms (<5%) in the bilingual corpora (Ezeizabarrena, 2002). But regardless of the age differences among the three corpora, mean rates of OSs do not vary very much (range = 34%–44%). Interindividual differences between the two bilinguals appear to be more visible than those between each bilingual and the monolingual Oitz, suggesting that they are more an effect of age than of linguistic profile.

Table 2.

Rates of OSs in finite and nonfinite clauses in Basque. Data of one monolingual (Oitz) and two bilingual children (Mikel and Jurgi).

	Oitz (1;07–2;06)		Mikel (1;07–3;11)		Jurgi (1;10–4;01)
	Verbs	OS	Verbs	OS	Verbs	OS
	n	%	n	%	n	%
Subtotal Finite clauses	368	44	1845	39.95	1068	48.98
*RI			148	14.2	56	26.8
Target n			316	10.5	143	2.8
Subtotal Nonfinite clauses	349	43.83	464	11.7	199	9.6
Total	717	43.9	2309	34.3	1267	42.78

RI: root infinitive; OV: overt subject.

So far, it can be concluded that overt production of subjects is attested during the earliest productions of two word utterances and that their rate remains quite constant during the first years. Although some individual differences in rates can be observed, overt production of subject arguments in finite clauses does not distinguish the simultaneous bilingual children (2L1) from the monolingual (L1).

Rates of OSs with infinitives (around 10%) are considerably lower in the bilingual corpora than in the monolingual one (44%) (Table 2). This difference might be related to the fact that the monolingual data belong to an earlier period than the bilinguals’ and that in this period many of the nonfinite verb forms are target-deviant productions of RIs in contexts of finite verbs.³ In fact, many of Oitz’s early verbs (38.6%) were target-deviant root infinitives (*RIs) up to age 2;02. In general, the decrease of *RIs to rates below 10% is a developmental step occurring in monolingual and bilingual corpora (Mikel: 31 months, Jurgi: 37 months, and Oitz: 31 months) after the 30th month of age (Ezeizabarrena, 2003). Though monolingual data with the necessary detail are not available, bilingual data in Table 2 show that OSs in early Basque are more frequent with target-deviant *RIs than with target-like infinitives and still more frequent with finite verbs, which is compatible with the idea that the earliest *RIs are intended finite forms lacking morphological person marking, in most cases, lacking the expected auxiliary (Ezeizabarrena, 2002, 2003).

Figure 3 shows that OSs are produced in every recording from the first production of a finite verb containing utterance in both bilingual corpora and reach a mean rate of around 45%. Children produce OSs even earlier with some nonfinite verbs, which indicates that OS production is quite constant from the very beginning since its rate varies with a range of around 30% throughout the over 30 months of longitudinal data collection (range, Mikel: 28%–57% and Jurgi: 26%–64% in recordings with more than four instances).

Figure 3.

Rates of Basque OS vis à vis total finite verb produced by two bilingual children (Mikel and Jurgi).

Thus, the three children — the monolingual and the two bilinguals — show a similar pattern of regular production of OSs in finite sentences — ranging between 20% and 60% across recordings from the earliest productions of inflected verb forms. Though the reduced number of participants in the longitudinal comparison prevents any statistical analysis, frequency rates in the production of OSs in early Basque suggest that the interindividual differences observed are not attributable to the L1/2L1 variable.

Three types of clauses containing OSs have been reported in early monolingual and bilingual Basque corpora (Almgren & Barreña, 2001; Barreña, 1995; Barreña & Zubiri, 2000): (a) utterances without any verb (10a), (b) utterances containing a nonfinite verb (10b and11b), or (c) utterances with finite verb forms (10c and 11a). Larrañaga (2000, 2009) remarks that most of the OSs are preverbal as in (10b and c) and (11a and b).

(10) a. Andoni, hemen ni-Ø (O 1;08)

Andoni here I-A

Andoni, I (=me) here

b. ni-Ø igon (O 1;11)

I-A go up

I go up

c. ni-Ø ni-Ø ni-k kertsi-a-Ø ba-d-auka-(t) (O 2;02)

I-A I-A I-E pullover-Det-A yes-A3s-have-(E1s)

I, I, I have a pullover

The distribution of OSs in Basque shows that demonstrative pronouns are the most frequent type of OSs (31.5%), followed by one-word DPs —proper or definite common nouns (29.5%) — and by personal pronouns (25.4%) (Table 3). Thus, about a third of the subjects are truly lexical arguments (35.8%) in both corpora (Oitz: 32.6%, Mikel: 38.1), while pronominals of different kinds are the most frequent option (nearly 60%). Among them, demonstratives (Oitz: 25.4%, Mikel: 31.5%) and first- and second-person pronouns (Oitz: 29.2%, Mikel: 22%) are followed by interrogatives (Oitz = 10.7%, Mikel = 3.2%) (Table 3). Demonstratives are the most frequent pronominals in finite clauses. However, in nonfinite clauses, demonstratives are a less frequent option (Oitz = 10.6%, Mikel = 14.2%), following lexical DPs (Oitz = 37%, Mikel = 51%) and first- and second-person pronouns (Oitz = 40.5%, Mikel = 32%). In contrast, no differences in pattern are observed in the typology of lexical subjects with finite and nonfinite utterances.

Table 3.

Distribution of subject types across finite and nonfinite clauses in Oitz’s and Mikel’s production in Basque.

		Total OSs	Personal pronouns n (%)	Demonstrative n (%)	WH n (%)	One-word DPn (%)	More complex DP n (%)
Oitz (1;06–2;06)	Finite	163	30 (18.4%)	64 (39.2%)	17 (10.4%)	33 (20.2%)	14 (8.6%)
	Nonfinite	152	62 (40.5%)	16 (10.6%)	17 (11.2%)	55 (36.4%)	1 (0.6%)
	Total Oitz	315	92 (29.2%)	80 (25.4%)	34 (10.7%)	88 (27.9%)	15 (4.7%)
Mikel (1;07–3;06)	Finite	434	97 (22.3%)	160 (36.8%)	14 (3.2%)	133 (30.6%)	30 (6.9%)
	Nonfinite	25	8 (32%)	4 (14.2%)	1 (4.7%)	8 (32%)	4 (19.0%)
	Total Mikel	459	105 (22.8%)	164 (35.7%)	15 (3.26%)	141 (30.7%)	34 (7.4%)
	Total Finite	597	127 (21.2%)	224 (37.5%)	31 (5.1%)	166 (27.8%)	44 (7.3%)
	Total Nonfinite	177	70 (39.5%)	20 (11.2%)	18 (10.1%)	63 (35.6%)	5 (2.8%)
Total		774	197 (25.4%)	244 (31.5%)	49 (6.3%)	229 (29.5%)	49 (6.3%)

OS: overt subject; DP: determiner phrase; WH: interrogative pronoun.

As seen before, first- and second-person pronouns (Oitz: 29.2%, Mikel: 22.8%) are not the most frequent kind of OSs, but, interestingly, their frequency does not seem to vary very much developmentally. As Table 4 shows, personal pronoun rates oscillate between 15% and 53% out of all OSs in the 2-year period studied. Since their first appearance in verb-containing utterances at age 2;01, some first- and second-person pronouns are attested in every recording. Moreover, some personal pronouns are produced in nonfinite clauses, though a couple of them are imperative forms (11b). It must be noted that imperative forms are also periphrastic in this language, though the auxiliary-less option (11b) is the most frequent in oral speech.

Table 4.

OSs and pronouns in Basque by a monolingual and a bilingual child.

Age	Oitz (monolingual)			Mikel (bilingual)
	OSn	Pronounsn	Pronoun/OS%	OSn	Pronounsn	Pronoun/OS%
1;07	2	0	0
1;08	9	0	0
1;09	—	0	0	12	0	0
1;10	0	0	0	4	0	0
1;11	15	2	13.3	4	0	0
2;00	22	7	31.8	21	0	0
2;01	18	10	55.5	4	3	75
2;02	34	8	23.5	14	1	7.14
2;03	30	10	33.3	16	4	25
2;04	83	22	26.5	42	5	11.90
2;05	39	9	23	26	3	11.54
2;06	63	24	38.1	7	2	28.57
2;07				20	0	0
2;08				26	5	19.2
2;09				23	9	39.1
2;10				14	3	21.4
2;11				15	8	53.3
3;00				36	8	22.2
3;01				27	4	14.8
3;02				25	10	40
3;03				22	5	22.7
3;04				24	11	45.8
3;05				22	10	45.4
3;06				30	6	20
Total	309	92	29.2	434	97	22.35

OS: overt subject.

(11) a. Ni-Ø tu(s)tau in-Ø na(iz) (M 2;01)

I-A frighten do-PF be.A1s

I’m frightened

b. Zu-(k) ipini! (M 2;00)

You-E put

put it on!

OSs in the cross-sectional study

NSs are assumed to be frequent in both child and adult Basque, though, to our knowledge, no empirical studies on adult OS and NS production have yet been conducted. The only data available that can be brought to bear on this question is the frequency of OSs in the narrative of one of the adults participating in the cross-sectional experiment, in charge of telling the story to one of the groups of children. This adult produced 46 OSs out of 109 finite clauses (42.2%).

The frequency of OSs in Basque was also tracked in the oral corpora of L1 and cL2 children at ages 5 and 8. The cross-sectional data obtained from the two sessions of narratives produced by 12 early Basque–Spanish bilingual children, for whom Basque is their only first language (L1), or one of their first two (2L1), show rates of OSs below 30% of the total of finite clauses in any of the sessions (Figure 4). Unexpectedly, subjects dropped more frequently in the first session at age 5 (OS = 18%) than in the last one (OS = near 30%) at age 8. A standard two-tailed paired-samples (dependent) t-test was conducted. Based on the resulting statistic (t(11) = 3.29, p = 0.007), the null hypothesis of no difference between the age-5 and age-8 means was rejected. Thus, a between-session development is attested in the production of OSs by this group. This development in the production of OSs is not surprising as it may be related to the increasing complexity of the narratives from ages 5 to 8 (to adulthood), in which the production of lexical OSs will be required for cohesive purposes as well as to disambiguate between different third-person characters.

Figure 4.

Rates of OS in narratives produced by Basque (2)L1 children in two different sessions.

Figure 5 includes the number and rates of OSs in finite sentences produced in three successive sessions by the group of children whose use of Basque is restricted to the school environment (cL2). Contrary to what was observed in the L1 group, rates of OS do not vary in the cL2 group, ranging between 32% and 35%. The repeated-measure analysis of variance (ANOVA) confirms this fact (F(2,22) = 0.23, p = 0.79).

Figure 5.

Rates of OSs in narratives produced by Basque cL2 children in three different sessions.

In the between-group statistical analysis, the Kormogorov–Smirnof test confirmed that the paired L1/cL2 scores were normally distributed at age 5 (Z = 0.673, p = 0.75) and at age 8 (Z = 0.53, p = 0.94). The t-test revealed significance in the between-group difference of means (L1 = 17.9%, cL2 = 34.08%) during the first session at age 5 (t(16,14) = −3.61, p = 0.002). In contrast, differences observed at age 8 between L1 (29; 55%) and cL2 (34; 92%) are only marginally significant (t(22) = −1.73, p = 0.098).

Overall results from longitudinal and cross-sectional studies

Summing up, subject person marking is attested very early on in early Basque and — contrary to what is observed in some non-NS languages — no clear increase is observed in the overt production of OSs (a steady rate around 45%) from age 2 to 3;6 in this language. Neither did the frequency of production of personal (first and second person) pronouns (mean = 22% for all OSs) show any development in the period studied, that is, from 2;01 onward, an age that shortly follows the explosion in the productive use of person marking on the verb.

These results are compatible with the assumption that T contains person and number features from very early on, that T agrees with the subject, and that L1 and 2L1 children’s OSs occupy high positions such as SpecTP or even higher in the syntactic structure at the same period. In contrast, more (age- and profile-related) variability seems to exist in the narrative corpora of older children. The narrative data of the L1 children at ages 5 and 8 show that OSs are less frequent (from 18% to 35%) than in both spontaneous production at earlier ages below age 4 (over 40%) and the adult narrative (40%). Moreover, significant differences are observed between L1 and cL2 in the session at age 5 (p < 0.05), which disappear in the last session at age 8 (p > 0.05).

Data on OS production contrast with those concerning case marking within the same populations. Contrary to the quite consistent pattern observed in the production of OSs, a comparison between the same participants discussed in the present study shows that (2)L1 children acquiring Basque pass through an initial optional ergative case marking stage (OECS) until approximately age 3. This stage differentiates L1 children from cL2 children, as L1 children overcome it by age 3;6, while the productions of cL2 children remain target deviant even after 4 years of (partial) exposure to the language (ages 5 through 8) (Ezeizabarrena, 2012).

OSs in the early stages of acquisition of other languages

The data analyzed in the previous section suggest that no change is observed in the types and rates of subject production in Basque during the first years of language development. Results obtained in monolingual and bilingual language corpora from various NS and non-NS languages will be presented in the following subsections.

OSs in monolingual acquisition studies

This section includes several studies on early monolingual acquisition of NS (Catalan, Spanish, Italian) and also non-NS languages (English, French).

Catalan and Spanish

Catalan and Spanish are NS languages with rich person marking in the verb morphology. The finite verb agrees obligatorily in person and number with its subject. NSs are very frequent in adult Spanish and Catalan (around 60%–80% NSs), and OSs have a contrastive or focusing function in most cases. Following the study by Bel (2003), OSs are scarce in both languages during the earliest stages (around 20%). Their rate increases during later recordings and parallels that of adult language from age 2;03 on for Catalan (Julia) and age 1;10 on for Spanish (María) (Figure 6). At this age, MLU reaches values equal to or higher than 2.

Figure 6.

NS rates in early Spanish and Catalan.

As with Spanish, the production of stressed pronouns in Catalan (12) is very scarce. This is especially so for third-person pronouns, whose first and only production occurs at 2;05, and the plural forms, which are not yet attested at 2;06 (Table 5).

Table 5.

Personal pronouns in early Spanish and Catalan.

Child; language	First person	Second person	Third person	Plural pronouns	Total pronouns	Studied production period

	First production age (n)	First production age (n)	First production age (n)
Emilio; Spanish	2;1 (20)	2;1 (1)	—	—	21	1;10–2;6
Juan; Spanish	2;03 (4)	2;3 (45)	—	—	79	1;9–2;6
María; Spanish	1;9 (137)	2;2 (4)	—	—	141	1;07–2;06
Mean Rate in Spanish	79.25%	20.75%	0	0	100%
Gisela; Catalan	1;10 (22)	2;2 (9)	—	—	31	1;10–2;06
Júlia; Catalan	2;5 (11)	2;5 (2)	—	—	13	1;9–2;06
Pep; Catalan	1;10 (27)	2;3 (12)	2;5 (1)	—	40	1;6–2;06
Mean Rate in Catalan	71.43%	27.38%	1.19%	0%	100%
Total	251 (77.23%)	473 (22.46%)	1 (0.31%)		325 (100%)

Source: Adapted from Bel (2003).

(12) a. No, jo em vull treure els patins (Gisela 2;6)

No, I want to loosen my skates

b. Tu plorarás (Júlia 2;5)

You will cry

Bel (2003) reports a low production rate (12%) of personal subject pronouns in both Romance languages (Spanish and Catalan), which increases gradually until reaching adult values of around 25%. Among the personal pronouns, the first-person singular tends to appear earlier than the rest and is clearly produced as the dominant one (77%), followed by the second-person singular (22.5%). Only one form of the third-person pronoun is attested in her Catalan corpora and none in the Spanish, which reflects the marked character of this personal pronoun in both early Romance languages (Table 5).

French

Oral standard French is an exception among the Romance languages. While many of them distinguish first, second, and third person in both the singular and the plural verb inflection, person marking on the French verb is very poor and does not allow NSs. However, colloquial French has been considered a pro-drop language in which NSs are identified by prefixed verb morphology (clitics). A very detailed longitudinal and cross-sectional study on early colloquial French by Palasis (2010) has shown very early production of adult-like person marking in this language.⁴ On her account, and similar to Meisel’s (1994a) proposal, evidence for subject person marking is not restricted to strong personal pronouns or to the suffixes of verb forms with rich morphology (i.e. auxiliaries) since prefixed weak clitics (argumental il and nonargumental weak pronouns following an overt or nonovert DP) act with the lexical verbs (which has a very poor phonological realization of person morphology) in a way similar to rich person marking in typical pro-drop languages. In Palasis’ (2010) corpus, 3- to 4-year-old children —mostly exposed to colloquial French — produce clitics associated with a finite verb in 80.9% of the cases, with some additional instances of NS utterances (13d) but at a low rate (3.16%), which decreases throughout the period studied. “Doubled” subjects, like in (13a–c), occur at a rate of 26%, similar to what has been observed in Spanish and Catalan OSs. Moreover, Palasis (2010) finds that most early subjects are placed preverbally (78.53%) (13a).

(13) a. Mon papa i(l) vient me chercher

my father he comes to pick me up

b. elle est là ma photo

there is my picture

c. on voit Lucie nous

we see Lucie

d. * __veux pas manger

I don’t want to eat

On her account, children know that the morphosyntactic properties of subject marking in colloquial French allow the verbal morphology to be expressed as an affix (the universally unmarked option) and as a prefixed clitic (the marked option), which allows NSs. The low rates of “doubled” OSs (DP + clitic) suggest that their production may correspond to a contrastive or a focused meaning. In her corpus, doubled weak personal pronouns show the following distribution: 1s (36.89%), 2s (3.38%), 3s (including the impersonal and 1pl inclusive pronoun on: 57%), 2pl (0.2%), and 3pl (2.45%). Thus, her subjects produce consistent person marking (suffixes and weak subject clitics) very early, which identify NSs and allow (doubled) OSs.

English

English pronominals are assumed to have a purely grammatical motivation. For this reason, the realization of overt and dropped pronouns is not expected to follow the same pattern as the one observed in the NS languages, though NSs are observed in some contexts such as coordinated structures, RIs, and question–answer pairs (Austin, Blume, Parkinson, Núñez del Prado, & Lust, 1997).

(14) a. Adult What is she doing now?

Child ___ Playin game

Since they appear in RI constructions, these kinds of NS — which are also found in child speech — have been labeled as instances of PRO and are supposed to lack the functional category T (Grinstead, 1998). Austin et al. (1997) observe that NSs exist in adult English and in child English, too, although their rates are quite low (33.3% maximum at the beginning). Since the proportion does not decrease suddenly but gradually and is never 0%, which is similar to the adult system, they conclude that it is not parametric resetting that causes their decrease in development but rather some pragmatic learning of the discourse constraints. When compared with Italian children (Valian, 1991), rates of OSs are very high in the linguistic production of American and English children. Valian suggests that NSs and overt pronouns have a complementary distribution in the sense that “pronouns are the least taxing NPs one can produce” (Valian, 1991, p. 32). This assumption led to the prediction that the sum of the NSs and the overtly realized subjects divided by the total number of subjects will remain constant across the MLU range, whereas the percentage of pronouns out of the sum total may increase. This is exactly what she found. Despite some interindividual variation, their rate is relatively high (an average of 69%–84% OSs by children aged 1;10–2;5 with the lowest/highest mean length of utterance measured in words (MLUw). Pronoun rates increase from a mean of 73% of the OSs at time I to 86% at time II. Similarly, Austin et al. (1997) observe that 60% of the OSs in their study were pronouns. Moreover, parental data of subject use was also attested — about 97% OSs (excluding imperatives) and 92% OSs (including imperatives), yielding a mean rate of about 75% — which confirms the similarity between child and adult performance in this regard. Furthermore, Italian children aged 1;10–2;6 produce OSs (mostly postverbal) at around 22%, and this rate remains constant from time I (up to 1;10) to time II (up to 2;5), whereas rates of pronouns increase from 22% at time I to 35% at time II in this language. With these data, Valian (1991) concludes that children know the value of the pro-drop parameter for their language from very early on, as rates of OSs are almost 70% during the first stage (MLU below 2) and never lower than 84% after MLU ≤ 2 (reaching 95% around MLU = 4.00). There certainly are changes in children’s performance afterward, but these changes are more gradual than abrupt, and the errors are due more to performance than competence. Assuming that all elements in the skeletal tree are innate, this author concludes that “the developing child has to learn how to lexicalize the nodes, what the proper phrase orders are in the particular language and what empty categories exist in her language” (Valian, 1991, p. 78).

OSs in bilingual acquisition studies

Spanish–English

This section includes several studies on early and simultaneous bilingual acquisition of a NS and a non-NS language. Paradis (2001) analyzes subject production in the data of three monolingual Spanish-speaking children and one Spanish–English bilingual girl, Manuela. Manuela is exposed to Spanish by her father, a native speaker of Cuban Spanish, and her mother, a native speaker of English who speaks Spanish (L2), and exposed to English through interaction with her caregiver and her grandmother. Two different periods are distinguished in the development of the four children: a very early one (MLU up to 1.75), in which OSs are scarce (average 5% to 30%), followed by a second stage (MLU over 2), in which OSs become more frequent (over 24% for each child). Interestingly, rates of OS in Table 6 do not vary dramatically from stage I (30%) to II (39%) in the bilingual, Manuela, though they do for the monolingual, Emilio⁵ (5%–24%).

Table 6.

Child production of OSs in Spanish by three monolinguals and one bilingual.

Child	% OS	Range (%)	MLU
Stage I (MLU < 1.7)
Manuela (bilingual)	30.8	14–71	1.1–1.4
Emilio (monolingual)	5	0–6	1.2–1.5
Eduard (monolingual)	7	0–18	1.0–1.7
Stage II (MLU > 1.8)
Manuela (bilingual)	39	32–57	2.1–2.5
Emilio (monolingual)	24	12–30	1.9–2.3
Maria (monolingual)	44	33–61	1.8–2

Source: Adapted from Paradis (2001, pp. 70–71).

OS: overt subject; MLU: mean length of utterance.

The bilingual child, Manuela, shows a different kind of linguistic behavior: (a) OSs are frequent during the whole period of study (1;7–2;6), even from the beginning, though at a similar rate to those in the other monolingual studies mentioned in section “OSs in longitudinal case studies”; (b) many of her subjects are pronouns (30%), in contrast with the Spanish monolingual children (16%–25%); and (c) she produces more pragmatically deviant subjects than her peers (Paradis, 2001; Paradis & Navarro, 2003). A more detailed observation of her linguistic input revealed that Manuela’s parents also produce OSs more frequently (54%–58%) than the parents of the monolingual children (below 46%). As Caribbean varieties of Spanish are considered to have more subject pronouns than other varieties, Manuela’s rates of OSs are attributable to her linguistic input. Paradis and Navarro (2003) do not discard either explanation, namely, that both internal and external mechanisms may be responsible for crosslinguistic influence in bilingual acquisition.

Another study on bilingual Spanish–English children is the one carried out by Liceras et al. (2008), who studied two Spanish–English bilingual twins growing up in a bilingual family. Their main conclusion in their longitudinal study is that the children follow two different developmental patterns in the production of OSs in both languages (See also Silva-Corvalán and Sánchez-Walker (2007)).

Table 7.

Rates of OS in Spanish and English by two bilingual twins.

	Spanish		English
	MLU	% OS (n of items)	MLU	% OS (n of items)
Stage I (2;04–2;06)	1.43–1.48	16.2 (n = 206 OS)	1.16–1.31	32 (n = 50 OS)
Stage II (3;01–3;09)	3.35–3.77	26.37 (n = 952 OS)	3.11–3.45	95.7 (n = 278 OS)
Stage III (4;04–4;11)	3.88–4.28	27 (n = 957 OS)	4.17–4.48	95.6 (n = 876 OS)

Source: Adapted from Liceras et al. (2008).

MLU: mean length of utterance; OS: overt subject.

On the one hand, OSs in Spanish remain quite constant during the three stages (15%–30%). Such a rate is very similar to the ones found in monolingual corpora of early Spanish (Table 7). In contrast, rates of OSs show a very different picture in the English corpus of these bilingual twins. NSs are highly frequent during the earliest stage (68%), whereas their rate decreases very strongly in the following stages (to 4%). Note that the MLU values during the first stage are lower than 1.5, which indicates that the frequency of constructions of more than two words is still quite low. When the length of utterance becomes long enough (MLU > 3), NSs almost disappear.

Furthermore, these authors show that there is a trade-off between NSs and pronouns in English after stage I, which is not observed in the Spanish data. In Spanish, overt pronouns are scarce throughout the whole period of study, below 20% in Table 8, whereas their rate increases dramatically in the English productions of the bilingual twins at stage II. In fact NSs virtually disappear from this age onward (5% NSs).

Table 8.

Spanish pronouns in the corpus of two bilingual twins.

	Spanish		English
	Total pronouns (null + overt)	% Overt pronouns (n) out of total	Total pronouns (null + overt)	% Overt pronouns (n) out of total
Stage I (2;04–2;06)	188	8% (15)	47	17% (13)
Stage II (3;01–3;09)	785	11% (84)	249	95% (237)
Stage III (4;04–4;11)	833	16% (135)	810	95% (771)

Source: Adapted from Liceras et al. (2008).

Following Robert’s (2001) markedness proposal, Liceras et al. (2008) conclude that Spanish is the unmarked option of the NS parameter, whereas English appears as the marked one. Furthermore, in their view, the special role of bound/free morphology characterizes first-language acquisition. Since the Spanish data seem very similar to the Italian data analyzed by Valian (1991), these results would appear to indicate that consistent NS languages have similar morphological properties, that is, licensing mechanisms that constitute the unmarked option of the parameter. Therefore, it appears that the morphological properties of each NS language may play a role in the developmental pattern for setting the correct [+/−] value of the NS parameter. With regard to a possible crosslinguistic influence on bilingual acquisition, the data presented by Liceras et al. (2008; 2011), in which no transfer from either language has been attested, provide one more piece of evidence for the separation hypothesis (Meisel, 1994a, 1994b).

German–Portuguese bilingualism

Hinzelin (2003) reports different rates of OSs during the whole period studied in the production of German–Portuguese children. This researcher distinguishes two stages in every language based on the acquisition of finiteness (production of contrastive first- and third-person endings): (a) stage I in Portuguese is described as children producing 3s marking on the verb (present, imperfect, perfect) and 3s pronouns; (b) stage II in Portuguese is described as children producing 1s, subject pronouns, modals, and auxiliaries; (c) stage I in German is described as children producing agreement errors, few OSs, and few pronouns; and (d) stage II in German is described as children producing different person markings, increase of OSs, and frequent pronouns, which suggests that production of OS increases with the consistency of person marking morphology (Daniel at age 2;07 and Luis at age 3;02). Interestingly, children seem to produce more OSs in Portuguese than in German initially, but the pattern changes in the second stage.

Moreover, differences are observed in both varieties of Portuguese, European Portuguese (EP), and Brazilian Portuguese (BP) with regard to OSs, as they are less frequent in BP during the initial (19%) and later stages (40%) than in EP (quite constant on mean rates of 50%). In contrast, pronouns — almost absent at stage I — become quite frequent (30%) at stage II in both varieties.

In German, the initial production of overt pronouns is random in the period previous to the productive marking of person, but pronominal subjects are produced at rates of over 80% after this developmental point. Among them, the expletive pronoun es “it” is scarce even later, contrary to what Hyams (1986) observed in English, where the emergence of expletives was considered to mark the end of the optional subject stage.

Hinzelin’s (2003) results confirm bilingual children’s high sensitivity to the frequency of production of OSs and personal pronouns in each of the languages. Interindividual age differences are observed, as Daniel produces productive person marking earlier in German than in BP (2;07), whereas Luis shows the opposite pattern. However, rates of OSs contrast in both languages of both bilinguals. In the non-NS language, children produce few OSs in the first stage (around 20%), and their rate increases dramatically in the second stage (over 70%), similar to what was observed for personal pronouns (0% to around 60%). In contrast, a quite steady increase in rates of OSs and pronouns is attested in Portuguese, despite some dialectal variation observed with regard to OSs. Both kinds of languages show an early presence of OSs, even before the productive person inflection at stage II (Table 9).

Table 9.

Productions of OSs and personal pronouns by bilingual children in German and Portuguese.

	Daniel (Brazilian Portuguese)	Luis (European Portuguese)	Daniel (German)	Luis (German)
Stage I (up to 2;08)	19% OS	45% OS	27% OS	20% OS
	0% pronoun/OS	1.8% pronoun/OS	4.2% pronoun/OS	0% pronoun/OS
Stage II (2;09 onward)	40% OS	58% OS	84% OS	72% OS
	33% pronoun/OS	28% pronoun/OS	69.8% pronoun/OS	55.4% pronoun/OS
Age of acquisition of finiteness	2;08	2;08	2;07	2;10/3;02

Source: Adapted from Hinzelin (2003).

OS: overt subject.

So far, the longitudinal bilingual data are in line with those of the monolinguals studied by Valian and Eisenberg (1996) who reported that 20 monolingual children acquiring Portuguese (mean age 2;06) showed values of OSs that varied from 28% in the lowest MLU group to 57% in the highest MLUw group. Rates of OSs and pronouns vary across the samples and studies reported by Valian and Eisenberg (1996), but data suggest that rates of overt pronouns are bound to person features (1s pronouns around 80%, 3s pronouns around 65%). Note that both varieties of Portuguese, BP and EP, are NS languages though their morphological system of person marking varies as well as the properties of their NSs (Holmberg et al., 2009). Valian and Eisenberg do not report any differences between the EP- and the BP-speaking children in their study. These authors find a correlation between higher NS rates and the lowest MLU values and conclude that the lowest production of OSs during the earliest stages, or during those with the lowest MLU values, is an indicator of low performance in both NS and non-NS languages. Though their rate may vary across individuals and across varieties, the early presence of subject marking seems to be confirmed in early monolingual and bilingual Portuguese.

Inuktitut–English bilingualism

One more study of early bilingual acquisition of a NS (Inuktitut) with a non-NS language (English) is that by Zwazinger, Allen, and Genesee (2005). The acquisition pattern in the longitudinal English data of the six simultaneous bilingual children aged 1;8–3;9 is very similar to the one in Valian’s study. From the very beginning, children in their study produce high rates of OSs in contexts of obligatory subject (75%), which reach adult-like rates (95%) by age 2;6. As for Inuktitut, monolingual and bilingual children produce a target-like rate of OSs (mean = 14%), showing a development between the first (0% OSs) and the last stage (around 18% OSs). However, it should be pointed out that, in this language in which first- and second-person pronouns cannot appear in the argument position, child OSs are dominantly demonstratives, followed by lexical arguments.

Discussion

Spontaneous monolingual and bilingual L1 Basque data have shown an early production of person inflection on verbs at early stages in children’s development: age ranging from 2;00 to 2;08 and MLU —measured in morphemes — ranging from 1.7 to 2.4. Moreover, OSs are present from the very beginning in their production at a rate that remains quite constant throughout the period studied (mean = 34%–44% OSs, across corpora), especially after the acquisition of finiteness, and replicates rates attested in cross-sectional studies of older children’s narrative texts at ages 5 (mean range = 18%–35% OSs) and 8 (~30% OSs). Though rates vary across ages and children’s corpora, OSs are attested throughout the whole studied period, and no development can be observed in their rates, contrary to what is observed in non-NS languages. More specifically, personal pronouns are attested in almost every recording in M’s bilingual corpus between ages 2;01 and 3;06, although no clear increase is observed in their frequency, contrary to what is observed in non-NS languages (Hinzelin, 2003; Liceras et al., 2008; Valian, 1991).

Moreover, RI stages have been reported in many languages, though they vary across languages both in terms of their frequency (less in Catalan and Spanish than in Basque, Dutch, English, or German) and form (roots in English and stems or inflected forms in German, Spanish, and Catalan). Regardless of these factors, the data illustrate a different behavior of NSs with finite and RI verbs crosslinguistically. OSs are found with both finite and nonfinite verb (RIs and *RIs) forms in Basque, contrary to Spanish and Catalan (Liceras et al., 2006). With regard to their distribution, lexical DPs (35.1%) and demonstratives (37.5%) are the most frequent option with Basque finite verbs, followed by first- and second-person pronouns (21.2%). Likewise in nonfinite contexts, lexical DPs (38.4%) are very frequent, as they are personal pronouns (39.5%), both followed by demonstratives (11.2%). These results seem to contradict Guasti’s (1993/1994) conclusion that children distinguish finite and nonfinite verbs from very early on; however, some additional data should be mentioned. The similarity observed in finite and nonfinite verbs in child Basque is consistent with the assumption that a NS in Basque is a licensed silent DP, regardless of the finite or nonfinite character of the clauses (Duguine, 2010; Elordieta, 2001).

Early crosslinguistic production data from NS languages converge providing strong evidence for the robustness of the child’s knowledge of subjects by age 3 and from MLU 2 onward (mostly measured in words), regardless of the language typology. In spite of the shortness of utterances and the defective overt morphology in early productions, some general results are attested across a variety of related and unrelated NS languages. See Barreña (1995), Bel (2002, 2003), Ezeizabarrena (1996, 2002), Hinzelin (2003), Guasti (1993/1994), Larrañaga (2009), Meisel (1994b), Rojas (2003), Serrat and Aparici (2001), Valian (1991) and Zwazinger et al. (2005), among many others. On the one hand, rich and regular subject–verb agreement languages like Basque, Spanish, Catalan, (probably) EP, French, Italian, and Inuktitut show:

Early contrastive production of overt person morphology in the verb inflection — affixes or clitics — as the most frequent way of subject marking. Person error commission is virtually nonexistent, though some omissions are attested.⁶

Early production of OSs, at quite a constant rate over time up to age 4, which approaches adult rates: Basque, Catalan, French, Italian,⁷ and Spanish (~30%) and Inuktitut (<20%).

Early production of some personal pronouns, whose low frequency (<30% of OSs) does not increase in development and is close to adults’ use. The OS/pronoun distribution is clearly dependent on language-specific features of the pronominal system.

On the other hand, studies on the early acquisition of “irregular” subject–verb agreement languages such as English (Valian, 1991), German (Hinzelin, 2003; Meisel, 1994a), and (probably) BP (Lopes, 2003), report:

Quite early production of (some) contrastive person marking but a delayed target-like production of person inflection.

Very early presence and production of OSs, which increases developmentally.

Early production of personal pronouns (~20% out of OSs), which increases developmentally.

In this regard, the absence of argument personal pronouns (Inuktitut), or of third-person pronouns (Basque), as well as the marked use of third-person pronouns in languages such as Catalan and Spanish have a clear effect in the frequency of pronouns and consequently in the total rate of OSs in adult and child speech, especially in spontaneous data. As for overt pronouns, first and second-person pronouns show a similar pattern in early Spanish, Catalan, and Basque, regardless of the existence/absence of a 3s personal pronoun in adult Catalan–Spanish/Basque, respectively. In child Spanish and Catalan, this pronoun is almost absent before age 2;6, in contrast with French and English. In this regard, developmental accounts, which state that the speaker will use a null NP rather than a pronoun NP, seem to fit in the data, but statements concluding that “the subjects that are expressed should be primarily lexical rather than pronominal” (Valian, 1991, p. 36) need to be reformulated in order to account uniformly for all NS languages.

In addition to the overt marking of the person feature, children’s early knowledge of subjects extends to their placement. Despite the low frequency in Catalan, Spanish, and Basque, lexical subjects of unaccusative verbs are attested to appear postverbally in contrast to those of (di)transitive verbs and unergative verbs in the early production of 2- to 4-year-old children (Bel, 2002; Larrañaga, 2000, 2009).⁸ Nevertheless, children show some delay in the target marking of some other features, such as the plural and ergative case in Basque (Barreña, 1995; Barreña & Zubiri, 2000; Elosegi, 1997; Ezeizabarrena, 2012; Ezeizabarrena & Larrañaga, 1996).

The licensing conditions of NSs in NS languages have traditionally been related to core syntax (Rizzi, 1982). In contrast, the production of OSs in NS languages, more specifically pronouns, has been commonly related to the pragmatic domain (Serratrice et al., 2004; Sorace, 2005). However, as both overt and null third-person pronouns in NS languages may refer to a person or object introduced as a topic in the discourse, it sounds plausible to assume that not only the identification but also their licensing may be regulated by conditions that apply beyond narrow syntactic structural conditions such as c-command, at least in some NS languages (Zushi, 2003).

This dissociation between the narrow syntax and the syntax-pragmatics interface in the C domain has served as an explanatory basis for the results of many studies which reported overproduction of OSs in NS languages in contact situations following the Interface Hypothesis (Sorace & Filiaci, 2006). As mentioned above, comparative research on monolingual corpora of NS/non-NS languages revealed different patterns in the production of OSs and of pronouns in both kinds of languages (Austin et al., 1997; Bel, 2002, 2003; Valian, 1991; Valian & Eisenberg, 1996). Moreover, the overproduction of pronouns attested in developing grammars of NS languages in contact with a non-NS language and in early simultaneous bilingualism has led to the conclusion that the overuse of overt pronouns is due to the influence of the non-NS language on the NS language (Haznedar, 2010; Müller & Patuto, 2009; Serratrice et al., 2004; Sorace, 2005; Sorace & Filiaci, 2006). However, no homogeneous evidence for such interlinguistic influence has been found, as no overproduction of OSs is reported in the studies on bilingual Spanish (Liceras et al., 2008; Liceras et al., 2011; Silva-Corvalán and Sánchez-Walker, 2007), Portuguese (Hinzelin, 2003), or Inuktitut (Zwazinger et al., 2005) surveyed here. Moreover, different developmental patterns are observed in the two languages with the opposite option acquired by the simultaneous bilinguals, which is compatible with the separation hypothesis (Meisel, 1994a, 1994b). Other aspects of subject marking have also been reported as evidence for potential vulnerability to crosslinguistic influence, such as ergative case marking in Basque. See Austin (2007) and Ezeizabarrena (2012) for discussion on this topic.

So far, data of children acquiring NS and non-NS languages simultaneously confirm a robust knowledge of the syntactic and pragmatic rules governing subjects, which is in line with the early knowledge of finiteness. Moreover, the target-like OSs point to the existence of similar licensing mechanisms in child and adult languages, which supports the continuity position (Valian, 1991). The conglomeration of the empirical results mentioned suggests that monolingual and bilingual children are extremely sensitive to the interpretation of the person feature, regardless of whether it is expressed by bound person morphology (affixes), by free morphemes (clitics, person pronouns), or by nonovert mechanisms. From the results discussed so far, robust evidence exists for the knowledge of subjects’ syntax (feature checking, subject placement) from age 2 or 2;06 onward. It is this sensitivity that makes children’s conservatism in spontaneous production possible, in Snyder’s (2007) sense, since they rarely produce instances of options not permitted in the grammar of the target language such as mismatches in overt person marking. This same sensitivity is also compatible with a very early knowledge of Duguine’s Argument Ellipsis Condition. As the argument ellipsis condition is nonparametric, it seems plausible to assume that children do not need to learn or set any value for it, thus they will know that an argument can be elided “iff it is linked (and if there are linking ‘traces’ at PF –namely, p-features and indexation” (Duguine, 2010, p. 271).

Finally, children acquiring a consistent NS language will need to learn, for instance, the lexical specificities of the functional category T, and among them, T has an uninterpretable [D] feature in consistent NS languages, which could receive a value (a referential index) from a null A-topic (Holmberg et al., 2009).

Conclusion

The early production of OSs analyzed in different NS languages leads to the conclusion that there is no need to assume some defective value of the pro-drop parameter at stage 0. Assuming that the EPP and knowledge about the interpretable person feature are innate, it can be hypothesized that what children need to learn is the way this feature is checked (or valued) in the relevant language(s). Moreover, the asymmetry observed in the mastering of subject–verb agreement inflection and that of case marking on the subject in both L1 and cL2, suggests that the feature specification of case and the overt marking of person agreement in verb inflection are not acquired simultaneously. What children acquiring a (consistent or partial) pro-drop language with rich person inflection must learn is not whether NSs are allowed in the NS language but when and how the subject person marking already expressed in the verb morphology (T) should appear doubled as an overt argument or pronoun.

Footnotes

Acknowledgements

I am very grateful to Andoni Barreña, who provided me with some of the data on Oitz; to all the children participating in the studies mentioned; and to the institutions Lizarrako Ikastola, Maria eta Jose Ikastetxea, as well as to Iñaki Garcia for his help with the statistical analysis. I also appreciate comments by Katerina Palasis and Esther Torrego and colleagues at ELEBILAB on earlier versions of this article. Furthermore, suggestions made by the two anonymous reviewers, by the editors, as well as by a native speaker of English lead me to improve this final version considerably. The remaining mistakes are all my own.

Funding

This research was funded by the University of the Basque Country (GIU09-39), the Ministry of Science and Technology (CSD2007-00012), the Ministry of Science and Innovation (FFI2009-13956-C02-01), and the Basque Government (PI2009-22).

Notes

Author biography

Maria-José Ezeizabarrena is Associate Professor of General Linguistics at the University of the Basque Country (Spain). She obtained her Ph.D. at University of Hamburg (Germany). Her research on early monolingual and bilingual acquisition of morphosyntax and vocabulary is carried out within research projects funded by diverse national and international institutions such as the ESF. Part of her individual and collaborative work has been published in journals like Lingua, Linguistics, Infancia y Aprendizaje, IJB as well as as chapters in books published by John Benjamins, Kluwer, Cascadilla, Multilingual Matters, etc.

References

Aguado-Orea

Pine

J. M.

(2002). There is no evidence for a “no overt subject” stage in early child Spanish: A note on Grinstead (2000). Journal of Child Language, 29, 865–874.

Almgren

Barreña

(2001). Bilingual acquisition and separation of linguistic codes: Ergativity in Basque versus accusativity in Spanish. In Nelson

K. E.

Aksu-Koç

Johnson

C. E.

(Eds.), Children’s language: Interactional contributions to language development (pp. 27–48). Mahwah, NJ: Erlbaum.

Artiagoitia

(2000). Hatsarreak eta parametroak lantzen [Working on principles and parameters]. Vitoria/Gasteiz, Spain: Diputación Foral de Alava-Universidad del País Vasco.

Austin

(2007). Grammatical interference and the acquisition of ergative case in bilingual children learning Basque and Spanish. Bilingualism, Language and Cognition, 10, 315–331.

Austin

Blume

Parkinson

Núñez del Prado

Lust

(1997). The status of pro-drop in the initial stage: Results from new analyses of Spanish. In Pérez-Leroux

A. T.

Glass

W. R.

(Eds.), Contemporary perspectives on the acquisition of Spanish: Developing grammars (Vol. I, pp. 37–54). Somerville, MA: Cascadilla.

Barreña

(1995). Haur euskaldunen hizkuntzaren jabekuntza-garapena: INFL eta KONP Funtzio-kategorien erabileraz (Published PhD). Universidad del País Vasco, Bilbao, Spain.

Barreña

Zubiri

J. J.

(2000). Subjektu eta objektu markak euskaraz eta espainieraz haur elebakar eta elebidunengan [Subject and object marking in Basque and Spanish by monolingual and bilingual children]. In Barreña

Ezeizabarrena

M. J.

(Eds.), Kode bereizketa eta bateraketa eremu urriko hizkuntzarekiko elebitasun goiztiarrean (Ikastaria Vol. 12, pp. 69–106). San Sebastian, Spain: Eusko Ikaskuntza.

Bel

(2002). Sujetos nulos y sujetos explícitos en las gramáticas iniciales del castellano y del catalán [Null and overt subjects in early Spanish and Catalan grammars]. Revista Española de Lingüística, 31, 537–561.

Bel

(2003). The syntax of subjects in the acquisition of Spanish and Catalan. Probus, 15, 1–26.

10.

Beloki

Manterola

(2007). Ahozko ipuia haur-hezkuntzan: Ikerketa esperimentaletik didaktikara [Oral narratives in child education: From experimental research to didactics]. In Idiazabal

Garcia

(Eds.), Ahozko euskara. Azterketak eta irakaskuntza. Bilbao, Spain: Universidad del País Vasco. Retrieved from http://testubiltegia.ehu.es/

11.

Clahsen

(1986). Verb inflections in German child language: Acquisition of agreement markings and the functions they encode. Linguistics, 24, 79–121.

12.

Duguine

(2010). Argumentu isilak eta hizkuntz aldakortasuna [Silent arguments and linguistic variation]. Bilbao, Spain: Udako Euskal Unibertsitatea.

13.

Elordieta

(2001). The role of verbal agreement in licensing null arguments. In Fernández

Albizu

(Eds.), Kasu eta komunztaduraren gainean/On case and agreement (pp. 113–132). Bilbao, Spain: Universidad del País Vasco.

14.

Elosegi

(1997). Kasu eta preposizioen jabekuntza-garapena haur elebidun batengan (Unpublished master’s dissertation). University of the Basque Country, Bilbao, Spain.

15.

Ezeizabarrena

M. J.

(1996). Adquisición de la morfología verbal en euskera y castellano por niños bilingües (Unpublished master’s dissertation). Universidad del País Vasco, Bilbao, Spain.

16.

Ezeizabarrena

M. J.

(2002). Root infinitives in two pro-drop languages. In Pérez-Leroux

A. T.

Liceras

J. M.

(Eds.), The acquisition of Spanish Morphosyntax (pp. 35–65). Dordrecht, Netherlands: Kluwer.

17.

Ezeizabarrena

M. J.

(2003). Null subjects and optional infinitives in Basque. In Müller

(Ed.), (In)vulnerable domains in multilingualism (Hamburg Series on Multilingualism) (pp. 83–106). Amsterdam, Netherlands: Benjamins.

18.

Ezeizabarrena

M. J.

(2012). The (in)consistent ergative marking in early Basque: L1 versus cL2. In B. Fernández & I. Laka (Eds.), Special issue accounting for ergativity, Lingua, 122(3), 303–317.

19.

Ezeizabarrena

M. J.

Larrañaga

M. P.

(1996). Ergativity in Basque: A problem for language acquisition? Linguistics, 34, 955–991.

20.

Fernández Soriano

(1999). El pronombre personal. Formas y distribuciones. Pronombres átonos y tónicos [The personal pronoun. Forms and distributions. Stressed and unstressed pronouns]. In Bosque

Demonte

(Eds.), Gramática descriptiva de la Lengua Española (pp. 1209–1273). Madrid, Spain: Espasa.

21.

Grinstead

(1998). Subjects, sentential negation and imperatives in child Spanish and Catalan (Unpublished master’s dissertation). University of California, Berkeley, CA.

22.

Guasti

M. T.

(1993/1994). Verb syntax in Italian child grammars: Finite and non-finite verbs. Language Acquisition, 3, 1–40.

23.

Haznedar

(2010). Transfer at the syntax-pragmatic interface: Pronominal subjects in bilingual Turkish. Second Language Research, 26, 355–378.

24.

Hinzelin

M. O.

(2003). The acquisition of subjects in bilingual children: Pronoun use in Portuguese-German children. In Müller

(Ed.), (In)vulnerable domains in multilingualism (Hamburg Studies on Multilingualism) (pp. 107–138). Amsterdam, Netherlands: Benjamins.

25.

Holmberg

Nayudu

Sheehan

(2009). Three partial null-subject languages: A comparison of Brazilian Portuguese, Finnish and Marathi. Studia Linguistica, 63, 59–97.

26.

Huang

(1984). On the distribution and reference of empty pronouns. Linguistic Inquiry, 15, 531–574.

27.

Hulk

Müller

(2000). Bilingual first language acquisition at the interface between syntax and pragmatics. Bilingualism: Language and Cognition, 3, 227–244.

28.

Hyams

(1986). Language acquisition and the theory of parameters. Dordrecht, Netherlands: Foris.

29.

Iraola

Ezeizabarrena

M. J.

(2011). Anaphora resolution in Basque: Null vs. overt subject hura. In Danis

Mesh

Sung

(Eds.), BUCLD 35 online proceedings supplement. Retrieved from http://www.bu.edu/bucld/proceedings/supplement/vol35/

30.

Jaeggli

Safir

(1989). The null subject parameter and parametric theory. In Jaeggli

Safir

(Eds.), The null subject parameter (pp. 1–44). Dordrecht, Netherlands: Kluwer.

31.

Kato

(1999). Strong and weak pronominals in the null subject parameter. Probus, 11, 1–37.

32.

Larrañaga

M. P.

(2000). Ergative Sprachen, akkusative Sprachen. Der Erwerb des Kasus bei bilingualen Kindern [Ergative languages and accusative languages. The acquisition of case by bilingual children]. Frankfurt am Main, Germany: Vervuert.

33.

Larrañaga

M. P.

(2009). The subject of unaccusative verbs in bilingual Basque/Spanish children. In Guijarro-Fuentes

Larrañaga

M. P.

Clibbens

(Eds.), First language acquisition of morphology and syntax (pp. 183–208). Amsterdam, Netherlands: Benjamins.

34.

Liceras

J. M.

Bel

Perales

(2006). Living with optionality: Root infinitives, bare forms and inflected forms in child null subject languages. In Sagarra

Toribio

A. J.

(Eds.), Selected proceedings of the 9th Hispanic linguistics symposium (pp. 203–216). Somerville, MA: Cascadilla.

35.

Liceras

J. M.

Fernández-Fuertes

Alba de la Fuente

(2011). Overt subjects and copula omission in the Spanish and the English grammar of English-Spanish bilinguals: On the locus and directionality of interlinguistica influence. First Language. doi:10.1177/0142723711403980

36.

Liceras

J. M.

Fernández-Fuertes

Pérez-Tattam

(2008). Null and overt subjects in the developing grammars (L1 English/L1 Spanish) of two bilingual twins. In Pérez-Vidal

Juan-Garau

Bel

(Eds.), A portrait of the young in the new multilingual Spain (child language and child development) (Vol. 9, pp. 111–134). Clevedon, UK: Multilingual Matters.

37.

Lopes

R. V.

(2003). The production of subject and object in Brazilian Portuguese by a young child. Probus, 15, 123–146.

38.

Manterola

(2011). Euskarazko murgilketa haur hezkuntzan euskara-gaztelania elebitasuna ikergai. Ahozko ipuinen azterketa diskurtsiboa eta didaktikoa [Research on the Basque-Spanish bilingualism in the Basque inmersion educational system. Discoursive and didactic analysis of oral narratives]. Universidad del País Vasco. Retrieved from http://www.argitalpenak.ehu.es/p291-content/eu/contenidos/informacion/se_indice_tescspdf/eu_tescspdf/adjuntos/IBON%20MANTEROLA.pdf

39.

Meisel

J. M.

(2011). First and second language acquisitin. Cambridge: Cambridge Textbooks in Linguistics.

40.

Meisel

J. M.

(1994a). Getting FAT: Finiteness, agreement and tense in early grammars. In Meisel

J. M.

(Ed.), Bilingual first language acquisition: French and German grammatical development (pp. 89–129). Amsterdam, Netherlands: Benjamins.

41.

Meisel

J. M.

(Ed.). (1994b). La adquisición del vasco y del castellano [The acquisition of Basque and Spanish languages]. Frankfurt am Main, Germany: Vervuert.

42.

Meisel

J. M.

(1995). Parameters in acquisition. In Fletcher

MacWhinney

(Eds.), The handbook of child language (pp. 10–35). Cambridge, UK: Blackwell.

43.

Meisel

J. M.

(2007). Child second language acquisition or successive first language acquisition or successive first language acquisition? In Meisel

J. M.

(Ed.), Exploring the limits of the LAD. Arbeiten zur Mehrsprachigkeit [Working papers in multilingualism] (Vol. 80, pp. 65–94). Hamburg, Germany: Hamburg University.

44.

Meisel

J. M.

Müller

(1992). Finiteness and verb placement in early child grammars: Evidence from simultaneous acquisition of French and German in bilinguals. In Meisel

J. M.

(Ed.), The acquisition of verb movement (pp. 109–138). Dordrecht, Netherlands: Kluwer.

45.

Müller

Hulk

(2001). Crosslinguistic influence in bilingual language acquisition: Italian and French as recipient languages. Bilingualism: Language and Cognition, 4, 1–21.

46.

Müller

Patuto

(2009). Really competence-driven cross-linguistic influence in bilingual first language acquisition? The role of the language combination. In: Bernardini

Egerland

Granfeld

(Eds.), Mélanges plurilingues offerts à Suzanne Schlyter à l’occasion de son 65ème anniversaire (Études Romanes de Lund) (Vol. 85, pp. 299–319). Lund, Sweden: Lunds Universitet.

47.

Palasis

(2010). Syntaxe Générative et Acquisition: Le sujet dans le developpement du système linguistique du jeune enfant [Generative Syntax and acquisition: The subject in the development of young children’s linguistic system]. Villeneuve d’Ascq, France: ANRT Diffusion.

48.

Paradis

(2001). The use of subjects by a Spanish-English bilingual child. In Cantone

K. F.

Hinzelin

M. O.

(Eds.), Proceedings of the colloquium on structure, acquisition, and change grammars: Phonological and syntactic aspects. Arbeiten zur Mehrsprachigkeit [Working papers in multilingualism] (Vol. 27, pp. 67–77). Hamburg, Germany: Hamburg University.

49.

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, 371–393.

50.

Rizzi

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

51.

Rizzi

(1986). Null objects in Italian and the theory of pro. Linguistic Inquiry, 17, 501–557.

52.

Roberts

(2001). Language change and learnability. In Bertolo

(Ed.), Language acquisition and learnability (pp. 81–125). Cambridge, UK: Cambridge University Press.

53.

Rojas

(2003). Early acquisition of Spanish verb inflection: A usage based account. Psychology of Language and Communication, 7, 17–36.

54.

Serrat

Aparici

(2001). Morphological errors in early language acquisition: Evidence from Catalan and Spanish. In Almgren

Barreña

Ezeizabarrena

M. J.

Idiazabal

MacWhinney

(Eds.), Research on child language acquisition: Proceedings of the 8th conference of the international association for the study of child language (pp. 1260–1277). Somerville, MA: Cascadilla.

55.

Serratrice

Sorace

Paoli

(2004). Cross-linguistic influence at the syntax-pragmatic interface: Subjects and objects in English-Italian bilingual and monolingual acquisition. Bilingualism: Language and Cognition, 7, 183–205.

56.

Snyder

(2007). Child language: The parametric approach. Oxford, UK: Oxford University Press.

57.

Sorace

(2005). Selective optionality in language development. In Cornips

Corrigan

K. P.

(Eds.), Syntax and variation. Reconciling the Biological and the Social (pp. 55–80). Amsterdam, Netherlands: Benjamins.

58.

Sorace

Filiaci

(2006). Anaphora resolution in near-native speakers of Italian. Second Language Research, 22, 339–368.

59.

Trask

(2003). The noun phrase: Nouns, determiners and modifiers; pronouns and names. In Hualde

J. I.

Ortiz

Urbina

(Eds.), A grammar of Basque (pp. 119–358). Berlin, Germany: Mouton de Gruyter.

60.

Valian

(1991). Syntactic subjects in early speech. Cognition, 40, 21–81.

61.

Valian

Eisenberg

(1996). The development of syntactic subjects in Portuguese-speaking children. Journal of Child Language, 23, 103–128.

62.

Wexler

(1998). Very early parameter setting and the unique checking constrain: A new explanation of the optional infinitive stage. Lingua, 106, 23–79.

63.

Zushi

(2003). Null arguments: The case of Japanese and Romance. Lingua, 113, 599–604.

64.

Zwazinger

Allen

S. H.

Genesee

(2005). Crosslinguistic influence in bilingual acquisition: Subject omission in learners of Inuktitut and English. Journal of Child Language, 32, 893–909.