A sentence is not simply a set of words (e.g. cat, dog, bit), but also contains the grammatical information necessary to allow the listener to determine ‘who did what to whom’ (and also whether the utterance is a statement or a question, and so on). As we saw in the previous chapter, many languages convey this information by means of morphology (e.g. nominative and accusative noun markers). Many languages, with English being a prime example, instead rely primarily on word order. For example, basic declarative English sentences use SUBJECT–VERB–OBJECT word order to mark the AGENT, ACTION and PATIENT of an event respectively (e.g. The dog bit the cat means something different to The cat bit the dog). Although some languages (arguably) exhibit entirely ‘free’ word order (i.e. all word orders are possible, equally common and convey the same meaning), most retain a ‘default’ word order that is used for the majority of basic declarative sentences (and which guides interpretation when case marking is not present), even when case marking makes word-order information redundant.

Our main focus in this chapter is the question of how children learning word-order languages acquire the word-order ‘rules’ of their language. However, we will also discuss data from languages in which morphological cues to meaning override word-order information. We focus here on simple sentences: sentences with only one clause (e.g. The dog bit the cat).

In this introductory chapter, we briefly outline (1) the major theoretical approaches to child language acquisition research, (2) the domains and debates to be covered in the rest of the book and (3) the major methodological paradigms used in the field.

The major theoretical approaches

Although there are a few exceptions, for the majority of the debates that we will encounter in this book, each of the competing proposals will generally be aligned with one of the two major theoretical approaches to language acquisition. These are (a) the nativist, generativist, Universal Grammar (UG) approach and (b) the constructivist, emergentist, socio-pragmatic, functionalist, usage-based approach. As outlined below, which terms are most appropriate depends on the precise nature of each proposal, and the domain under investigation.

Nativist/generativist/Universal Grammar (UG) proposals

• A theoretical proposal that is nativist assumes that some important aspects of children's linguistic knowledge are not acquired, but innate (present from birth and, for at least some researchers, encoded in the genome).

• A theoretical proposal that is generativist assumes that children's knowledge of grammar (encompassing syntax, inflectional morphology and, in some approaches, phonology) consists of knowledge of formal ‘rules’ or operations that operate on abstract linguistic categories (e.g. VERB, NOUN) and phrases (e.g. VERB PHRASE, NOUN PHRASE). For example, oversimplifying somewhat, one operation for forming an English sentence combines a NOUN (e.g. John) and a VERB (e.g. danced) in that order (e.g. John danced not Danced John). This is an example of a syntactic operation (a ‘rule’ of syntax; see Chapters 4 and 6). An example of a morphological operation (a ‘rule’ of inflectional morphology) is the process by which the past-tense -ed inflectional morpheme is added to a regular English VERB (e.g. kiss → kissed; see Chapters 4 and 5).

• […]

How do children acquire their native language? This question has prompted a lively theoretical debate and a great deal of empirical research, much of which explicitly tests the predictions of the various competing accounts. In our view, it is therefore unfortunate that most previous child language textbooks and monographs barely cover this debate at all. Instead, most authors, whether implicitly or explicitly, adopt a particular theoretical position and largely discuss only research conducted within the relevant paradigm. If proposals that derive from an opposing theoretical viewpoint are discussed at all, it is generally only to dismiss them, with no serious consideration of their strengths and weaknesses.

We want to change all that with this textbook. Our goal is to identify the key debates in each of what we consider to be the ‘core’ domains in language acquisition (by monolingual, typically-developing speakers) and to outline the empirical evidence for and against each theoretical proposal, in an even-handed, systematic and (as far as space permits) comprehensive manner. We aim to hold every proposal to an equally high standard of proof – making no prior assumptions with regard to parsimony or psychological plausibility – and, in so doing, to provide detailed challenges to all researchers, of all theoretical standpoints.

This focus on theoretical debates has three important consequences. The first is that, although we seek to be comprehensive in covering all the major domains of acquisition (speech perception/segmentation and production, word learning, inflectional morphology and syntax), this textbook does not constitute a descriptive account of what children's language looks like at each age (there are many other textbooks that fulfil this function).

The defining characteristic of human language is that it allows speakers to produce entirely new utterances (i.e. utterances that the speaker has never heard in precisely that form). Clearly in order to do so, speakers must be in possession of a grammar: a set of some kind of abstract ‘rules’ or ‘schemas’ used for combining words into sentences (syntax) and adding markers to words to express features such as tense (morphology). Two central questions in child language acquisition research are therefore: (1) what does this grammar look like? (i.e. what is the nature of the underlying system)? and (2) where does it come from (i.e. how is it acquired)? In this chapter, we outline two competing answers to the first question – the theoretical accounts of the adult endstate assumed by the generativist and constructivist approach respectively – and the implications of these accounts with regard to the second question. Although, in principle, one could imagine a non-nativist generativist account, all the actual generativist accounts that will be discussed in this chapter (and the remainder of the book) assume that at least some linguistic categories and principles regarding utterance formation are innate (i.e. present from birth). Constructivist accounts assume that whilst the potential to acquire language is of course innate, children are not born with innate knowledge of grammatical categories or principles, and construct their grammars on the basis of the input to which they are exposed.

Introduction

When we hear speech, a stream of sound enters the auditory system. From this stream, any adult who is a native speaker of that language can virtually instantly identify sound sequences that allow them to understand what is being said. They can also produce these sounds to convey meaning. The question of how this is done is by no means resolved for adult speech perception. But explaining how an infant learning a first language identifies and produces the units of that language presents even more difficulties. This issue divides into two main questions. First, how does the infant work out what are the meaningful sounds of the language and second, how does the infant connect these same meaningful sounds to the movements of the articulators (e.g. tongue, lips) needed to produce them? In this chapter, we first outline the basic acoustic and articulatory features of speech and then, in Section 2.3, children's learning of the sounds of their language. Here, the main debate we consider is whether or not children build the phonemic inventories on the basis of innate distinctive features – a clearly nativist position. The constructivist alternative is that the acoustic variability between phonemes that are analysed as having the same features means that the phonemic inventory, and the articulatory contrasts that underpin it, have to be built bottom-up from the input. Section 2.4 considers how children segment the speech stream into words, phrases and clauses.

Introduction

At the same time as they are parsing the continuous speech stream into individual words (Chapter 2), children must learn the meanings of these words. This chapter investigates different proposals for how they do this.

Initially, it may seem as if there is no question to answer. One might imagine that the child's caregiver simply produces a word (e.g. rabbit) and points out the object that this word denotes (e.g. a rabbit that is running past). A problem with this account is that such ‘teaching’ is virtually impossible for word types such as abstract nouns (e.g. happiness), non-actional verbs (e.g. watching), prepositions (e.g. in) and determiners (e.g. the) (Tomasello, 2003). Another problem (Quine, 1960) is that if a child's mother (or a speaker attempting to teach his or her own language to a foreign linguist) produces a word (e.g. gavagai) as a rabbit runs past, the range of potential meanings is virtually infinite. Of course, gavagai could mean rabbit, but it could equally mean leg, white, furry (a part or property of the rabbit), animal, rodent, Flopsy (an object name, but at a different level of specificity), running, fleeing, chasing (the action of the rabbit), pointing, looking, catch it! (actions involving the speakers) or even grass, field or Today is Tuesday (something not connected to the rabbit at all).

The previous chapter dealt with children's acquisition of sentences that use canonical (basic) declarative word order and that are ‘simple’, in the sense that they contain a single clause expressing a single proposition or idea (e.g. The boy kicked the ball). The present chapter investigates children's acquisition of sentences that use different word orders (passives and questions; e.g. The ball was kicked by the boy; What did the boy kick?; Section 7.1) and multiple-clause sentences (e.g. The boy who is smoking is crazy; Section 7.2).

The four debates covered here – on the acquisition of passives, questions, relative clauses and sentential complement clauses – are each a debate between generativist accounts on the one hand, and constructivist accounts on the other. On the generativist side, although we cover a number of different proposals, most modern approaches share an important similarity in assuming that children have the competence necessary to produce and comprehend these structures from birth. Thus generativist research has generally attempted to show that, when various task demands are minimized, children show adultlike performance with these structures from a very early age. Constructivist approaches assume that children are not born with the necessary syntactic knowledge, but must acquire these constructions on the basis of the input. Thus, as for the acquisition of basic word order (Chapter 6), constructivist research has generally attempted to show that early knowledge of these constructions is restricted to low-scope frames built around certain items, that gradually increase in abstraction with age.

In Chapters 1–8, we have looked at the leading debates in the core areas of first language acquisition by monolingual, typically developing children. Our goal in this final chapter, before summarizing our previous discussions and drawing some conclusions, is to briefly outline some relevant issues which we have not been able to consider elsewhere. All bear in one way or another on the debates between (broadly speaking) generativist–nativist and constructivist approaches (even if only to suggest that there may be other ways of framing the debate). We do not attempt to adjudicate between the competing proposals in detail, as we have in the rest of the book, but simply to make readers aware of the more recent research on these issues.

Related debates

Modularity, domain specificity and brain localization

A traditional nativist view is that children are born with an innate language module. A module, as defined by Fodor (1983), is a closed system (encapsulation) that is not available to conscious awareness (unconscious), is mediated by a dedicated neural system (localization) and – crucially for our purposes here – deals exclusively with one particular information type (domain specificity); in this case, language. Although it is perfectly possible, in principle, for a modular system to be acquired as opposed to innate (e.g. Bates, 1993; Elman et al., 1996), the traditional view (e.g. Chomsky, 1957, 1965, 1988; Fodor, 1983) is that evidence for the domain specificity and localization of language constitutes evidence for an innate language module.