Language Development Over the Lifespan

Kees de Bot, Robert W. Schrauf

• Publisher: Routledge
• Year: 2010
• eBook ISBN: 9781135839383
• Topic: Languages & Linguistics
• Subtopic: Linguistics
• Index: Languages & Linguistics

Language Development Over the Lifespan is a reference resource for those conducting research on language development and the aging process, and a supplementary textbook for courses in applied linguistics/bilingualism programs that focus on language attrition/aging and adult literacy development in second languages. It offers an integrative approach to language development that examines changes in language over a lifetime, organized by different theoretical perspectives, which are presented by well-known international scholars.

Part I
THEORETICAL APPROACHES TO LANGUAGE DEVELOPMENT

1
FIRST AND SECOND LANGUAGE DEVELOPMENT FROM A UG PERSPECTIVE

Sharon Unsworth

1 Motivating UG: L1 Acquisition

The central tenet of the generative approach (Chomsky 1959, 1965, 1975, 1981, 1986, 1995, 1999) is that humans are predetermined to acquire language; this biological endowment ensures that all (non-impaired) children acquire the language (or languages) to which they are exposed in their environment, and it distinguishes humans from all other species. The argument for such a language acquisition device is derived from the observation that the linguistic knowledge which children acquire goes far beyond the input to which they are exposed. Before considering what this predetermined knowledge involves, let us examine the argument for its existence in more detail.

Children’s linguistic development progresses at an astonishing rate. At the age of 6–8 months already, when children are only just starting to crawl, their linguistic abilities are so sensitive that they are able to discriminate between the sounds of what will become their native language and between sounds in other languages (e.g. Werker & Tees, 1984). At around age 1 year, they start to produce their first words, and approximately 6 months later they start to combine words with each other, adding grammatical elements such as articles, plural forms and prepositions, to form increasingly complex sentences such that at around age

when they might be learning how to ride a tricycle, they start to produce complex sentences such as relative clauses.¹ What is more remarkable is that these stages are consistent across children and that they have been observed for various languages, including sign languages (see e.g. Lillo-Martin, 1999). The rapidity and uniformity of language acquisition across different learning conditions also suggests (though of course does not necessarily entail) that children must come equipped for the task with innate knowledge about language.

Perhaps the most crucial argument for the existence of an innate language faculty comes from the observation that children go beyond the input they hear, that is, they come to know more than the totality of that to which they are exposed. How children progress beyond the input, a problem which has come to be known as the logical problem of language acquisition or the poverty of the stimulus, has to be addressed by any account of language acquisition. The generativist’s solution to this problem is to posit that certain parts of linguistic knowledge are innate. It is claimed that the child cannot solely rely on the input because not only does this vary from child to child, but more importantly, it is also degenerate, finite and restricted in scope (Chomsky, 1965). It is degenerate because it includes the incomplete/unfinished sentences which are characteristic of everyday speech. It also only contains a finite number of sentences, yet the knowledge which children acquire allows them to produce a potentially infinite number of sentences. Finally, the input is restricted in scope in that it only contains positive evidence, that is, evidence of what is possible in the language in question. Negative evidence, that is, information about what is ungrammatical in a particular language, is not available and when it is (e.g. through correction by caretakers), it is not used (Marcus, 1993; Brown & Hanlon, 1970). Nevertheless, children still come to know which strings in their native language(s) are ungrammatical.

This innate knowledge or language faculty constitutes the initial state for the language-learning child. It provides information about the range of possible languages, including a set of abstract universal principles common to all languages, and a set of parameters capturing the variation observed between different languages. Together, these are known as Universal Grammar (UG). UG is often likened to a pre-wired box with a set of switches (Chomsky, 1988, analogy attributed to J. Higginbotham). The wiring specifies the possible options and the switch settings determine the available choices. Language acquisition thus involves—metaphorically speaking—flicking the appropriate switches on the basis of the input which the child hears.

The abstract universal principles of UG take the form of constraints. Constraints restrict the grammar in such a way that certain combinations, for example, of words or of sounds and meanings, are prohibited. These constraints hold for all languages and they may not be violated. The sentences in (1) and (2) illustrate a well-known constraint on form which prevents certain phonological processes from applying after wh-movement has taken place. More specifically, these examples concern the optional contraction between want and to in English; as we shall see, this phonological process is only possible under certain syntactic conditions. Let us start by briefly considering the derivation of wh-questions. The question in (1)-a is formed by moving the wh-word, what, from its base position as the object of eat, to sentence-initial position, leaving behind what is called a trace in the process; this trace, considered to be psychologically real, marks the wh-word’s original position, as illustrated in (1)-b, where t signifies trace. In (1)-c, want and to are contracted to wanna, whereas in the wh-question in (2) this is ungrammatical, as indicated by the asterisk in (2)-b. This rather subtle difference results from the fact that the questioned constituent in the infinitival clause is different in each of the two sentences: in (1), the questioned constituent is the object of eat, and in (2) the subject of go. More specifically, the trace left by the wh-word in (2)-b intervenes between want and to, as shown in (2)-c, and consequently, wanna-contraction is not possible. Native speakers of English have implicit knowledge of this constraint, which manifests itself in other languages, too, albeit in different ways; knowledge of this constraint entails that (2)-b is ungrammatical (see Crain & Thornton, 1998, Chapter 23, for more details).

(1)	a.	What do you want to eat?
	b.	Whati do you want to eat ti?
	c.	What do you wanna eat?
(2)	a.	Who do you want to go home?
	b.	*Who do you wanna go home?
	c.	Whoi do you want ti to go home?

The sentence in (3) illustrates a constraint on meaning.

(3)

Roy loves Angelai, but Traceyj hates heri/*j

The pronoun her refers to a singular antecedent who is female. The sentence in (3) contains two such potential antecedents, namely Tracey and Angela. As the co-indexing indicates, however, there is only one antecedent which is grammatical, namely Angela. Tracey cannot function as the antecedent for her because—and this is the relevant constraint—anaphoric interpretations are prohibited in certain structural configurations. (To be specific, the relevant constraint here is Principle B of the Binding Theory—Chomsky, 1981.)

Variation is instantiated in parameters linked to universal principles. For example, there is a principle which states that all phrasal constructions have “heads.” A noun phrase (NP) thus has a noun as its head, and a verb phrase (VP) a verb, etc. The position of this head relative to its complement is, however, subject to variation, and this is captured in a so-called Headedness Parameter, which has two options, head-initial and head-final. Languages such as English and French are head-initial, for example, whereas Japanese and Turkish are head-final. In recent versions of generative theory, the locus of variation has shifted to the lexicon, and more specifically, to functional categories.² For example, the functional category Tense is said to be weak in English and strong in French. This difference in “feature strength” accounts for the word order variation with respect to negation illustrated in (4).

(4)	a.	I do not eat peas very often
	b.	Je ne mange pas de petits pois très souvent I eat not of peas very often “I do not eat peas very often”
	c.	*Je ne pas mange de petits pois très souvent

In (4)-a, the main verb, eat, appears to the right of the negator, not, whereas in (4)-b, the main verb has “raised” over the negator, pas, and appears to its left. This is because the tense feature in French is strong and therefore requires the verb to move to that projection; if it does not, ungrammaticality results, as illustrated in (4)-c.

One area of debate in L1 acquisition research has been whether functional categories and their respective projections, for example Tense Phrase (TP), Determiner Phrase (DP) and Complementizer Phrase (CP), are available in the earliest stages of linguistic development. Proponents of so-called weak continuity claim that functional projections emerge one by one: children initially start with a VP and they subsequently add a TP, followed by a CP (e.g. Clahsen, Eisenbeiss, & Vainikka, 1994; Vainikka, 1993/1994). When children produce non-nominative subjects as in (5), it is claimed that this is because the position where nominative case is assigned, TP, is not yet available. This is also why verbs in children’s early utterances are often non-finite, as in (6).

(5)	My make a house (Nina 2;0)
(Vainikka, 1993/1994, p. 273, ex. 276)
(6)	a.	die helemaal kapot maken (Dutch, Niek 3;2) that altogether broken make-INF “completely break that one”
(Wijnen, 1997, p. 1, ex. 1)
	b.	dormir là Michel (French, P 2;2) sleep-INF there Michel “Michael sleep there”
(Deprez & Pierce, 1994)

Proponents of strong continuity (or the Full Competence Hypothesis—Poeppel & Wexler, 1993) argue that children do have access to all functional categories from the start, however (Weissenborn, 1990; Boser, Lust, Santelmann, & Whitman, 1992; Hyams, 1992). On this view, although not identical, child and adult grammars include the same grammatical objects and mechanisms, namely those made available by UG (see Guasti, 2002 for introduction to relevant issues). Within this approach, it has been argued that the development of syntax follows a predetermined biological schedule (Borer & Wexler, 1992); thus, children’s use of non-finite verbs in matrix clauses (so-called Root Infinitives) is due to immature principles of UG rather than the lack of functional categories (Rizzi, 1993/1994; Wexler, 1998).

To sum up, according to the generative approach, the initial state of L1 acquisition consists of a set of universal constraints, Universal Grammar, which restricts the hypotheses children (subconsciously) formulate when constructing their grammars. The variation observed in natural language is captured in a predetermined set of options from which children make a selection on the basis of the linguistic input to which they are exposed, be that from one or more languages. By around age 5, all of the correct options will have been selected.³

2 UG in Adult L2 Acquisition

The central question driving generative non-native (L2) acquisition research is whether L2 acquisition is constrained in the same way as L1 acquisition, that is, whether L2 grammars obey the same set of universal constraints as native-speaker grammars. Following early work by Corder (1967) and Selinker (1972), the non-native acquirer’s (L2er) grammar, or interlanguage (IL), is viewed as a system in its own right, rather than in terms of the L1 or the target language (TL). The questions which generative L2 research seeks to address include: What are the properties of the L2er’s interlanguage system? Are they properties which are characteristic of natural language grammar? How exactly are these properties acquired? As in most studies, the focus in this section will be on adult L2ers. The study of child L2 acquisition from a generative perspective is dealt with in the following section.

2.1 UG and/or L1 Transfer

Adult L2ers and L1 children differ by definition because L2ers already know another language. The extent to which L2ers use this knowledge in the L2 acquisition process, and whether and how the role of the L1 can be teased apart from the role of UG, has been subject to considerable debate.

In early generative work, the question of whether L2 grammars are UG-constrained was framed in terms of access to UG. L2ers were claimed either to have no access (e.g. Clahsen & Muysken, 1986), direct or full access (e.g. duPlessis, Solin, Travis & White, 1987; Schwartz & Tomaselli, 1990; Epstein, Flynn & Martohardjono, 1996) or indirect (or partial) access to UG (e.g. Clahsen & Muysken, 1989). On the no access view, L2ers are claimed to make use of general learning mechanisms, such as “linear sequencing strategies which apply to surface strings” (Meisel, 1997, p. 258).

In the approaches proposing access to UG, the existence of L1 influence was generally left implicit or denied, often because attributing a role to the L1 was considered to weaken the case for UG in L2 acquisition (White, 2003b, p. 27). For those espousing the indirect access view, UG is only available via the L1. If L2ers demonstrate unconscious knowledge of UG principles which could not be acquired as a result of...

Table of contents

• Cover Page
• Title Page
• Copyright Page
• Introduction
• PART I Theoretical Approaches to Language Development
• PART II Second Language Development
• PART III Non-verbal Aspects of Language Development
• PART IV Methodological and Neurolinguistic Aspects of Development