Much research on bilingual first language acquisition has stressed the role of the dominant or preferred language when the two languages have some influence on one another. The present paper tries to look at transfer or interference from the perspective of the input the child is exposed to. Transfer will be argued to occur in those domains of the grammar where the language learner is confronted with ambiguous input. The bilingual child may, as a relief strategy, use parts of the analysis of one language in order to cope with ambiguous properties of the other. Ambiguity of input is crucial and will be evaluated through a comparison with monolingual language acquisition: if monolingual children have problems with the language material in question, it may be suggested that the input contains evidence for more than only one grammatical analysis. A quantitative difference between monolingual and bilingual language acquisition will be interpreted as evidence in favor of cross-linguistic influence in bilingual language development. The paper reviews longitudinal studies on the acquisition of word order in German subordinate clauses.

Natascha Müller's article concerns the acquisition of word order in German subclauses by young bilingual children. Müller's basic argument as I understand it runs as follows:

(1) In the acquisition of word order in German subclauses some children make errors and others do not;

(2) there are some monolingual and some bilingual children who do not make errors;

(3) there are some monolingual and some bilingual children who do make errors, and the errors are qualitatively similar for monolingual and bilingual children;

(4) when children do make errors, they occur more frequently in bilingual than in monolingual children;

(5) the errors in both monolingual and bilingual children are due to the misprojection of a separate functional category;

(6) in addition, the errors in bilingual children (but not in monolingual children) are due to influence from the other language they are simultaneously acquiring. The reason for positing this is that bilingual children's errors are more frequent than monolingual children's errors.

Natascha Müller's proposal to view transfer as a relief strategy used by bilingual learners to cope with problematic input is very interesting and has far-reaching consequences for theories of bilingual language acquisition. The author makes a strong case for the fact that bilingual children transfer parameter values from the language presenting unambiguous input to the other “puzzling language”. In what follows, I will not question the main thrust of her argument. Rather I will return to the definitions that have been proposed for transfer in the literature and show that they are usually too broad. I will then propose that anyone interested in studying transfer must take into account the language mode the language learner or bilingual subject is in when being studied, and I will end by showing the consequences that this may have if it is not done. Natascha Müller's main argument is not affected by the language mode factor but the quantitative difference she finds between monolingual and bilingual children could be.

The paper by Natascha Müller follows the new way of studying bilingual first language acquisition which has been advocated in the last few years by several researchers (Döpke, 1996, 1997; Gawlitzek-Maiwald & Tracy, 1996; Hulk & Van der Linden, 1996; Hulk, 1997). In this new approach, while it is taken as an established fact that bilingual children are capable of early language differentiation, at the same time a lot of attention is given to cross-linguistic influence as an important characteristic of bilingual first language acquisition. We will discuss the nature of differentiation and transfer here as presented by Müller's paper and the way we see it ourselves. We will illustrate our position with examples from the French-Dutch bilingual child we have been studying for some time.

The field of bilingual first language acquisition has focused on several important and interrelated issues: whether or not the young child acquiring two languages simultaneously differentiates his or her two languages from the onset of acquisition, what role the input plays in the acquisition of two languages, and whether the path of acquisition is similar to that of monolingual peers (see De Houwer, 1990). As a member of the DUFDE team, Natascha Müller has in previous work argued forcefully and convincingly for the bilingual child's separate development of his or her two languages, and hence how language-acquiring bilinguals behave like monolinguals. In her keynote article, Müller invokes the notion of transfer, a well-known term from research into second language acquisition, and proposes to consider transfer from the perspective of the input to which the young bilingual child is exposed. When this input is ambiguous, so that is there is variation in the input regarding one of the languages, the child will resort to transfer from the other language as a so-called relief strategy. In the following, I address the issue of cross-linguistic influence in language development and highlight the implications Müller's proposals may have for the field of bilingual first language acquisition. In conclusion I will relate these issues to the complexity of the notion of input in early bilingualism.

Natascha Müller discusses a question which in recent years – after a period of focus on cross-linguistic principles of language acquisition – has regained its position as a current topic, namely transfer, and here in bilingual children. One of her points is that for transfer to take place, the transferred construction must have some correspondence in the target language, a position similar to the “transfer to somewhere” principle often advocated in the L2 acquisition literature (cf. Gass & Selinker, 1983). The specific structure studied is the lack of Verb-End in the German of a bilingual German-French child.

Compared with the wealth of investigations on monolingual language acquisition, research on early bilingualism still has some catching up to do. Papers such as the one by Natascha Müller show us what may be gained if results from both areas are integrated within one theoretical framework, in this case current generative theory.

Natascha Müller presents a very interesting analysis of the structure of the German subordinate clause in bilingual first language acquisition. The main issue in this paper is to explain the fact that some bilingual children – but not all of them – display non-target language structures in German subordinate clauses. That is, the finite verb does not appear in the final position of the subordinate clause. These non-target structures can in part be explained by transfer, but this cannot be the whole story, because some monolingual children produce these structures as well. Bilingual children, though, appear to have problems with the German subordinate clause more frequently than monolingual children. Interestingly, acquiring the target structures is a slow process for children who produce non-target structures. Ivar, the French-German child Müller discusses in most detail, for example, needs two years to acquire the correct position of the finite verb in German subordinate clauses. Müller argues that the problems arise because the input children receive is ambiguous: the finite verb does not always appear in the finite position in German subordinate clauses in adult speech. The ambiguity of this situation opens the way to transfer. In their uncertainty, the children turn to their other language as a relief strategy.

From the beginning of this century fascinating work has been carried out on the simultaneous acquisition of two languages from birth in the form of diary studies for example, that by Ronjat (1913). Many aspects of monolingual language acquisition have been studied in the same way, whereby researchers generally observed their own children, for example, Stern and Stern (1928). The modern research strategy is to audiotape or videotape the speech of young children at fixed intervals. Of course, a quantitative analysis can be carried out for both diary studies and more recent types of longitudinal study. However, researchers such as Ronjat (1913) do not provide us with exact quantitative results. Notwithstanding, these studies deserve to be taken seriously and I think it is possible to interpret some of the observations they contain as good and “reliable” evidence and compare these with results from modern longitudinal studies, where exact quantification is an important issue. I therefore think that the quantitative results of my review of existing studies on the acquisition of German subordinate clauses in bilingual children can be taken seriously, although I agree with François Grosjean and Elizabeth Lanza, who object that the quantitative differences between monolingual and bilingual children's errors are perhaps artefacts of the setting in which the children have been recorded. That I considered diary studies as well disappointed one commentator, Annick De Houwer, who would have preferred a comparison of modern studies only.

A word association experiment examined conceptual representation in bilingual memory. Dutch-English bilinguals associated twice to nouns and verbs that varied on concreteness and cognate status, once in the language of the stimuli (within-language), and once in the other language (between-language). Within- and between-language associations for concrete words and for cognates were more often translations of one another than those for abstract words and noncognates, and nouns evoked more translations than verbs. In both within- and between-language association, retrieving an associate was easier to concrete than to abstract words, to cognates than to noncognates, and to nouns than to verbs. These findings suggest that conceptual representation in bilingual memory depends on word-type and grammatical class: concrete translations, cognates, and noun translations more often share, or share larger parts of, a conceptual representation than abstract translations, noncognates, and verb translations. The results are discussed within the framework of distributed memory representation.

Two picture-word interference experiments were conducted to investigate whether or not words from a first and more dominant language are activated during lexical access in a foreign and less dominant language. Native speakers of Dutch were instructed to name pictures in their foreign language English. Our experiments show that the Dutch name of a picture is activated during initial stages of the process of lexical in English as a foreign language. We conclude that bilingual speakers cannot suppress activation from their first language while naming pictures in a foreign language. The implications for bilingual speech production theories are discussed.

This paper identifies two versions of the question as to whether there is a single initial system in the syntax of developing bilinguals. Version 1 asks whether there are early mixed utterances, and if so, attributes this to a single initial system. Version 2 asks whether the utterances containing words from one of the child's languages exhibit the same syntax as the utterances containing words from the child's other language. We argue with reference to our own data (from an English-Spanish bilingual from ages 1;7 to 1;9) that Version 1 is not tenable because of the paucity of lexical resources when the child begins to produce two-word utterances. However, we argue that the early two-word utterances in our data do seem to exhibit a single rudimentary syntax, based on a predicate-argument structure found in all utterance types, mixed and non-mixed. We then argue in relation to Version 2 of the question, that it can only be answered once the child's utterances can be identified as language-specific in the two languages, and that this is not possible before the emergence of morphological marking. This is illustrated by an analysis of our data from ages 1;8 to 2;3. We argue that language-specific morphology allows us to identify the language of the utterances in our data and to see evidence for the appearance of two differentiated morphosyntactic systems.