Theories of first language acquisition

Everyone at some time has witnessed the remarkable ability of children to communicate. As small babies, children babble and coo and cry and vocally or nonvocally send an extraordinary number of messages and receive even more messages. As they reach the end of their first year, children make spe­cific attempts to imitate words and speech sounds they hear around them, and about this time they utter their first "words." By about 18 months of age, these words have multiplied considerably and are beginning to appear in two-word and three-word "sentences"—commonly referred to as "tele­graphic" utterances—such as "allgone milk," "bye-bye Daddy," "gimme toy," and so forth. The production tempo now begins to increase as more and more words are spoken every day and more and more combinations of two- and three-word sentences are uttered. By about age three, children can comprehend an incredible quantity of linguistic input; their speech capacity mushrooms as they become the generators of nonstop chattering and incessant conversation, language thereby becoming a mixed blessing for those around them! This fluency continues into school age as children internalize increasingly complex structures, expand their vocabulary, and sharpen communicative skills. At school age, children not only learn what to say but what not to say as they learn the social functions of their language.

How can we explain this fantastic journey from that first anguished cry at birth to adult competence in a language? From the first word to tens of thousands? From telegraphese at eighteen months to the compound complex, cognitively precise, socioculturally appropriate sentences just a few short years later? These are the sorts of questions that theories of language acquisition attempt to answer.

In principle, one could adopt one of two polarized positions in the study of first language acquisition. Using the schools of thought referred to in the previous chapter, an extreme behavioristic position would claim that children come into the world with a tabula rasa, a clean slate bearing no preconceived notions about the world or about language, and that these children are then shaped by their environment and slowly conditioned through various schedules of reinforcement. At the other constructivist extreme is the position that makes not only the rationalist/cognitivist claim that children come into this world with very specific innate knowledge, predispositions, and biological timetables, but that children learn to func­tion in a language chiefly through interaction and discourse.

These positions represent opposites on a continuum, with many pos­sible positions in between. Now we are going to analyse main positions in the study of first language acquisition.

The theory of imitation

The "oldest" one - is the theory of imitation. It has adherents even nowadays. The essence of this theory: the child hears speech samples around and imitates these designs.

This theory, in our opinion, is not convincing enough and "exhaustive". We give only a few objections. Even from a large mass of diverse mono-sentences, which adults use, child among the first sentences, almost naturally, "selects" statements like "Mom", "Daddy," " Grandmom " "Auntie", "Uncle," "Father," "Give», «Take» and some others. On this objection the adepts of the Imitation Theory give the following argument: first words, sentences reportedly consist of the most common in the articular pronunciation sounds and the articulation of these sounds, the child has the ability to perceive visually.

However, until now there is no clear definition of criteria of sounds’ articulatory complexity (simplicity) and their hierarchy according to this feature. There is no evidence to suggest that, for example, the sound [d] more difficult or easier to sound [b], although the latter usually comes before the sound [d]; just as there are no grounds to assert that the sound [l] easier or harder then the sound [r], [f] easier or harder then [h], etc.

Of course, it does not depend on sounds’ articulatory "simplicity" or "complexity", especially in their "observability" and "unobservability" (blind children without other anomalies, learn the sounds in the same sequence as others). The point is in functional significance for the formation of language phonetic(or rather - phonemic) sounds system. Sounds [a] [a], [i]; [m], [p], [b], [t>], [t], [d], [d>], [n] comes first, not because they are articulatory "easier" then others, but because they are mostly pronounced ([a] - [o] [p] - [a] [p] - [m] [p] - [t]; [t>] - [d>];, etc.) and provide the necessary basis for the formation of other sounds (or rather - phonemes). With these basic sounds (phonemes) child is able to build the first words-sentences codified language to communicate, seeking to satisfy their needs (biological or social).

Numerous targeted surveillance of a language ontogenesis, and experimental studies have shown invalidity of Imitation Theory (275, 278, 284, etc.). In particular, it was proved that children usually do not use those sentences (syntactic structures) which are heard from mother. If the "average" child of 18-20 months, is offered to repeat the word "doll", "sit", “on”, "table", he will do it (of course, with a particular pronu

nciation of most words). However, having the ability to repeat isolated words, child can not repeat the sentence "The doll sits on the table". He will say: "Doll" or "Doll sits", or "Dolly table" and not otherwise, because in this age of syntactic and semantic components of its linguistic mechanism "work" in that way, and any kinds of imitation can not change this mechanism (to special events children’s "repeating" phrases are include). In addition, words which child repeats only at the insistence of adults, as a rule, would not be included into child’s independent speech.

Behavioristic Approaches

Language is a fundamental part of total human behavior, and behaviorists examined it as such and sought to formulate consistent theories of first language acquisition. The behavioristic approach focused on the immedi­ately perceptible aspects of linguistic behavior—the publicly observable responses—and the relationships or associations between those responses and events in the world surrounding them. A behaviorist might consider effective language behavior to be the production of correct responses to stimuli. If a particular response is reinforced, it then becomes habitual, or conditioned. Thus children produce linguistic responses that are rein­forced. This is true of their comprehension as well as production responses, although to consider comprehension is to wander just a bit out of the publicly observable realm. One learns to comprehend an utterance by responding appropriately to it and by being reinforced for that response.

One of the best-known attempts to construct a behavioristic model of linguistic behavior was embodied in B.F. Skinner's classic, Verbal Behavior (1957). Skinner was commonly known for his experiments with animal behavior, but he also gained recognition for his contributions to education through teaching machines and programmed learning (Skinner 1968). Skinner's theory of verbal behavior was an extension of his general theory of learning by operant conditioning. Operant conditioning refers to con­ditioning in which the organism (in this case, a human being) emits a response, or operant (a sentence or utterance), without necessarily observable stimuli; that operant is maintained (learned) by reinforcement (for example, a positive verbal or nonverbal response from another person). If a child says "want milk" and a parent gives the child some milk, the operant is reinforced and, over repeated instances, is conditioned. According to Skinner, verbal behavior, like other behavior, is controlled by its consequences. When consequences are rewarding, behavior is main­tained and is increased in strength and perhaps frequency. When conse­quences are punishing, or when there is a total lack of reinforcement, the behavior is weakened and eventually extinguished.

Skinner's theories attracted a number of critics, not the least among them Noam Chomsky (1959), who penned a highly critical review of Verbal Behavior. Some years later, however, Kenneth MacCorquodale (1970) published a reply to Chomsky's review in which he eloquently defended Skinner's points of view. And so the battle raged on. Today vir­tually no one would agree that Skinner's model of verbal behavior ade­quately accounts for the capacity to acquire language, for language development itself, for the abstract nature of language, or for a theory of meaning. A theory based on conditioning and reinforcement is hard-pressed to explain the fact that every sentence you speak or write—with a few trivial exceptions—is novel, never before uttered either by you or by anyone else! These novel utterances are nevertheless created by the speaker and processed by the hearer.

In an attempt to broaden the base of behavioristic theory, some psy­chologists proposed modified theoretical positions. One of these positions was mediation theory, in which meaning was accounted for by the claim that the linguistic stimulus (a word or sentence) elicits a "mediating" response that is self-stimulating. Charles Osgood (1953, 1957) called this self-stimulation a "representational mediation process," a process that is really covert and invisible, acting within the learner. It is interesting that mediation theory thus attempted to account for abstraction by a notion that reeked of "mentalism"—a cardinal sin for dyed-in-the-wool behaviorists! In fact, in some ways mediation theory was really a rational/cognitive theory masquerading as behavioristic.

Mediation theories still left many questions about language unan­swered. The abstract nature of language and the relationship between meaning and utterance were unresolved. All sentences have deep struc­tures—the level of underlying meaning that is only manifested overtly by surface structures. These deep structures are intricately interwoven in a person's total cognitive and affective experience. Such depths of language were scarcely plumbed by mediational theory.

Yet another attempt to account for first language acquisition within a behavioristic framework was made by Jenkins and Palermo (1964). While admitting that their conjectures were "speculative" and "prema­ture," the authors attempted to synthesize notions of generative linguistics and mediational approaches to child language.They claimed that the child may acquire frames of a linear pattern of sentence elements and learn the stimulus-response equivalences that can be substituted within each frame; imitation was an important, if not essential, aspect of establishing stimulus-response associations. But this theory, too, failed to account for the abstract nature of language, for the child's creativity, and for the inter­active nature of language acquisition.

It would appear that the rigor of behavioristic psychology, with its emphasis on empirical observation and the scientific method, only began to explain the miracle of language acquisition. It left untouched genetic and interactionist domains that could be explored only by approaches that probed more deeply.

The theory of innate language knowledge

The theory of innate linguistic knowledge, rather "young" and popular in the last three or four decades. Supporters of this theory (239, 275, etc.), believe that child is born with certain genetically determined knowledge "of language universals: universals of semantic, syntactic, lexical, phonetic and other ”. Society also plays a role of a " push " or " activato r" to " launch " of innate linguistic mechanism.

It seems that the idea of an innate capacity for various kinds of symbolization (landmark designation) in this theory is productive. Probably, also productive is a thought of innate universals of language, especially since some of them (at least some semantic and syntactic "rules") associated with mental universal (thinking, emotions, etc.).

At the same time, features of different languages and different cultures, "social environment" where child acquires language, show us the uniqueness of language acquisition as a whole system of assimilation and identity of its individual components (syntactic, lexical, phonetic, etc.), by children of different nationalities. Consequently, not only congenital factors determine the ontogenesis of language and speech activities in general. Considerable role in child’s speech development belongs to social factors, in particular, the specifics of the language which child adopts.

The Nativist Approach

Nativist approaches to the study of child language asked some of those deeper questions. The term nativist is derived from the fundamental asser­tion that language acquisition is innately determined, that we are born with a genetic capacity that predisposes us to a systematic perception of language around us, resulting in the construction of an internalized system of language.

Innateness hypotheses gained support from several sides. Eric Lenneberg (1967) proposed that language is a "species-specific" behavior and that certain modes of perception, categorizing abilities, and other language-related mechanisms are biologically determined. Chomsky (1965) similarly claimed the existence of innate properties of language to explain the child's mastery of a native language in such a short time despite the highly abstract nature of the rules of language.This innate knowledge, according to Chomsky, is embodied in a "little black box" of sorts, a language acquisition device (LAD). McNeill (1966) described LAD as consisting of four innate linguistic properties:

1) the ability to distinguish speech sounds from other sounds in the environment,

2) the ability to organize linguistic data into various classes that can later be refined,

3) knowledge that only a certain kind of linguistic system is possible and that other kinds are not, and

4) the ability to engage in constant evaluation of the developing lin­guistic system so as to construct the simplest possible system out of the available linguistic input.

McNeill and other Chomskyan disciples composed eloquent argu­ments for the appropriateness of the LAD proposition, especially in con­trast to behavioristic, stimulus-response (S-R) theory, which was so limited in accounting for the generativity of child language. Aspects of meaning, abstractness, and creativity were accounted for more adequately. Even though it was readily recognized that the LAD was not literally a cluster of brain cells that could be isolated and neurologically located, such inquiry on the rationalistic side of the linguistic-psychological continuum stimu­lated a great deal of fruitful research.

More recently, researchers in the nativist tradition have continued this line of inquiry through a genre of child language acquisition research that focuses on what has come to be known as Universal Grammar. Positing that all human beings are genetically equipped with abilities that enable them to acquire language, researchers expanded the LAD notion into a system of universal linguistic rules that went well beyond what was origi­nally proposed for the LAD. Universal Grammar (UG) research is attempting to discover what it is that all children, regardless of their environmental stimuli (the language [s] they hear around them) bring to the language acquisition process. Such studies have looked at question formation, nega­tion, word order, discontinuity of embedded clauses, subject deletion, and other grammatical phenomena.

One of the more practical contributions of nativist theories is evident if you look at the kinds of discoveries that have been made about how the system of child language works. Research has shown that the child's language, at any given point, is a legitimate system in its own right. The child's linguistic development is not a process of developing fewer and fewer "incorrect" structures, not a language in which earlier stages have more "mistakes" than later stages. Rather, the child's language at any stage is systematic in that the child is constantly forming hypotheses on the basis of the input received and then testing those hypotheses in speech (and comprehension). As the child's language develops, those hypotheses are continually revised, reshaped, or sometimes abandoned.

Before generative linguistics came into vogue, Jean Berko (1958) demonstrated that children learn language not as a series of separate dis­crete items, but as an integrated system. Using a simple nonsense-word test, Berko discovered that English-speaking children as young as four years of age applied rules for the formation of plural, present progressive, past tense, third singular, and possessives. She found, for example, that if a child saw one "wug" he could easily talk about two "wugs," or if he were pre­sented with a person who knows how to "gling," the child could talk about a person who "glinged" yesterday, or sometimes who "glang."

Nativist studies of child language acquisition were free to construct hypothetical grammars (that is, descriptions of linguistic systems) of child language, although such grammars were still solidly based on empirical data. These grammars were largely formal representations of the deep structure—the abstract rules underlying surface output, the structure not overtly manifest in speech. Linguists began to examine child language from early one- and two-word forms of "telegraphese" to the complex language of five- to ten-year-olds. Borrowing one tenet of structural and behavioristic paradigms, they approached the data with few preconceived notions about what the child's language ought to be, and probed the data for internally consistent systems, in much the same way that a linguist describes a language in the "field." The use of a generative framework was, of course, a departure from structural methodology.

The generative model has enabled researchers to take some giant steps toward understanding the process of first language acquisition. The early grammars of child language were referred to as pivot grammars. It was commonly observed that the child's first two-word utterances seemed to manifest two separate word classes, and not simply two words thrown together at random. Consider the following utterances:

My cap All gone milk

That horsie Mommy sock

Linguists noted that the words on the left-hand side seemed to belong to a class that words on the right-hand side generally did not belong to.That is, my could co-occur with cap, horsie, milk, or sock, but not with that or all gone. Mommy is, in this case, a word that belongs in both classes. The first class of words was called "pivot," since they could pivot around a number of words in the second, "open" class. Thus the first rule of the generative grammar of the child was described as follows:

Sentence -> Pivot word + Open word

Research data gathered in the generative framework yielded a multitude of such rules. Some of these rules appear to be grounded in the UG of the child. As the child's language matures and finally becomes adult-like, the number and complexity of generative rules accounting for language com­petence of course boggles the mind.

In subsequent years the generative "rule-governed" model in the Chomskyan tradition has been challenged. The assumption underlying this tradition is that those generative rules, or "items" in a linguistic sense, are connected serially,with one connection between each pair of neurons in the brain. A new "messier but more fruitful picture" (Spolsky 1989: 149) was provided by what has come to be known as the parallel distributed processing (PDP) model (also called connectionism) in which neurons in the brain are said to form multiple connections: each of the 100 billion nerve cells in the brain may be linked to as many as 10,000 of its counter­parts. Thus, a child's (or adult's) linguistic performance may be the conse­quence of many levels of simultaneous neural interconnections rather than a serial process of one rule being applied, then another, then another, and so forth.

A simple analogy to music illustrates this complex notion. Think of an orchestra playing a symphony. The score for the symphony may have, let's say, twelve separate parts that are performed simultaneously. The "sym­phony" of the human brain enables us to process many segments and levels of language, cognition, affect, and perception all at once—in a parallel con­figuration. And so, according to the PDP model, a sentence—which has phonological, morphological, syntactic, lexical, semantic, discourse, soci-olinguistic, and strategic properties—is not "generated" by a series of rules (Ney & Pearson 1990; Sokolik 1990). Rather, sentences are the result of the simultaneous interconnection of a multitude of brain cells.

All of these approaches within the nativist framework have made at least three important contributions to our understanding of the first language acquisition process:

1) freedom from the restrictions of the so-called "scientific method" to explore the unseen, unobservable, underlying, abstract lin­guistic structures being developed in the child;

2) systematic description of the child's linguistic repertoire as either rule-governed or operating out of parallel distributed processing capacities; and

3) the construction of a number of potential properties of Universal Grammar.