Part II: How do Homo sapiens sapiens children acquire speech?

From an evolutionary perspective, especially in a highly social species, communication of information seems to be intrinsic to the survival of the organism. Individuals within a social group need to be aware of individuals, relationships, objects and actions. In general, species which care for dependent young, the primary caregiver, usually the mother, must be able to identify the vocalizations of its offspring. Additionally, the offspring must be able to identify the sounds of its primary caregivers, members of its own species (allies and rivals), other species and noises produced by the natural world (i.e. water, wind, fire).

In this sense, it does not seem that modern human infants possess any extraordinary adaptive advantage over any other vocal species. However, there seems to be a consensus amongst researchers that human language is intrinsically different from other forms of nonhuman animal communication. For example, the de Boysson-Bardies (1999:40) states:

Acquiring a language requires associating sounds and meanings according to the phonological and syntactic rules of a particular (human) language... Human infants must accomplish the following: select the proper sounds which constitute the repertoire of sounds used in the language of their environment, and acquaint themselves with the combinations of these sounds... They must assimilate the prosodic cues (accent, rhyme and intonation) which link linguistic units into organized forms.

Due to the influence of Chomsky's Universal Grammar theory, the study of early language acquisition in children tends to be focused upon the identification of adult-like categories of language (i.e. vowels, syntax). When studying language development, in the context of experimentation, extraneous variables tend to be removed from consideration of the desired observation. In the case of infant language acquisition, factors such as touch, eye contact, positive reinforcement, and constant repetition (training) are largely ignored. Since it is known that infants are capable of learning any of the world languages, elements of adult language have been correlated with specific infant behaviours (prosody preferences, babbling and first words). To this end, researchers have attempted to isolate elements of adult grammar that infants seem capable of identifying.

While it is difficult to perform experimental tests with infants younger than 6 months of age, it appears these infants are able to form associations between repeated prosodic phrases, such as their names and identities of their parents (Trainor et. al. 2000:188; Fernald 1991:44). These prosodic cues, cross-culturally, consist of changes in vocal pitch and tone, which are characteristic of (adult) human languages (Aslin et. al. 1998). In general, even adults without children of their own, and children, all adopt the higher pitched prosody when dealing with young infants (ibid.). It is uncertain whether this behaviour is similar that provoked by the physical features of babies (large head, big eyes, small body and limbs) which humans transfer to other young animals (and inanimate objects such as dolls), or strictly cultural in nature. Additionally, infants appear to be aware of foreign phoneme differences, and show a preference for prosodic cues which resemble the (adult) language of their home environment (Werker and Desjardins 1995:77).

The ability to pick out, arbitrary, visual and auditory patterns seems to be a highly developed trait of modern humans. One such auditory pattern is prosodic pauses, which adults identify as indications of separate vowel and consonant sounds (Blake 2000; Nazzi et. al. 2000; de Boysson-Bardies 1999; Aslin et. al 1998). Since human speech is, generally, produced during exhalations and not inhalations of air, minute pauses for inhalation provide cues to word breaks (Aslin et. al. 1998). From the adult perspective, these vocalizations are content (and contextually) rich (Morton and Trehub 2001).

In the development of language, vocal volubility in infancy is correlated with talking in childhood (Locke 1998:193). While it appears that vocalizing leads to talking, it also appears that talking leads to grammar (ibid.). A working knowledge of a language entails the simultaneous functioning of memory and emotive cues, working in conjunction with the physiological apparatus for speech production. However, one can listen to a conversation or song in a foreign language, and pick out both words and word stops without having any clear concept of the meaning of the spoken words. Infants seem to exhibit this type of behaviour. It seems questionable, in this instance, that anything other than the ability to identify a particular auditory pattern is occurring. There is obviously much more going on during early language development than has been revealed by the current directions in research. Cognitive development and visual cues must play some factor (Trainor et. al. 2000:194). For example, in the case of deaf infants raised in a signing household, babbling alters from being a vocal exercise to manual babbling (de Boysson-Bardies 1999:68).

Pinker

Building upon the concept of a universal grammar and evolutionary adaptation, Steven Pinker (Pinker and Bloom 1990:726; Pinker 1994:365) proposed that the human brain is innately hardwired for language and that there is a gene for language. In other words, the evolution of language is equated with the evolution of the eye or hand. As opposed to Bickerton, Pinker and Bloom (1990:711) suggest that: 1) there was a gradual evolution from no language to modern human language, with random occurrences of mutation or recombinations; 2) language evolved as a function of earlier communication, not as a mechanism for mental representation; and 3) the development of grammar gave a selective advantage to early hominids.

However, language is not an anatomical feature, it is something which must be learned. It can be seen as having evolved from earlier forms of animal communication. Language is, essentially, dependent upon the actions of a speaker and a receiver(social interaction), unlike the functioning of an anatomic feature (i.e. eyes) (Beaken 1996:10). Secondly, unlike random genetic mutation, the development of a protolanguage or universal grammar cannot occur without social feedback (ibid.). While Pinker (1994, 1999) acknowledges that the preceding points are valid, he states that without an innate language acquisition device and universal grammar, human language would not exist. Additionally, like Chomsky and Bickerton, Pinker does not explain why a more complex form of communication would be more adaptive or effective, although they claim it provided a selective advantage (Beaken 1996:11; Aitchison 1998:17).

Pinker (1994, 1999, 2000) posits that children innately possess basic syntactic categories such as sentence subject and object, as well as innate linking rules. These linking rules enable children learning different languages to link the language they are learning to these innate categories (ibid.). However, in order to properly link the arguments of a verb with the relevant syntactic categories, the child must discover the syntactically relevant semantic features of the verb (Tomasello and Brooks 1999:179). In order to solve this problem, Pinker (1994, 1999, 2000) has suggested that the language acquisition device in the brain is a modular, computational system.

Theories which favour an innate mechanism for language lend themselves inevitably to a modular evolutionary scenario for language (Donald 1998:46). Since Chomsky (1968) had proposed that language is what separated humans from all other animals, these scenarios must account for the existence of a unique human grammar module that would be placed quite high in the architecture of the mind, and depend upon a number of lower level modules for its function (Donald 1998:46). Thus, language evolution involves an architectural issue: modularity. Modules are relatively autonomous brain systems, dedicated to one cognitive function, that gather specific classes of action (Donald 1998:45).

While these modular theories provide a clear analogy for behavioural functions, they are intellectual models which, as yet, cannot be mapped onto a physical brain. Pinker's (1994, 1999, 2000) computational models essentially subdivide the process by which human verbal communication is uttered. However, Pinker's theory has no empirical support (Tomasello and Brooks 1999). For example, until children have acquired the conventional semantic representation for each of the verbs in their native language, they will have the tendency to be use one verb for all situations. Pinker (1994, 2000) posits that children deal with this problem by gradually constructing a narrow range of verbs that do not participate in particular constructions. However, researchers have found that three and four year old children make more errors than younger children, which is contrary to Pinker's theory (Tomasello and Brooks 1999:180). Additionally, there is no evidence that children construct a narrow range verbs at the ages at which they begin to make these errors (ibid.).

It does not seem possible that there are innate rules which are invariably reliable in indicating to all children, learning all of the world languages, at all historical periods, how meanings they need to understand and convey are linked to some innate set of abstract syntactic structures. The human brain does not act like a preprogrammed capacity, since it does not automatically self install languages (Donald 1998:49). For example, people who are linguistically isolated from birth, such as the nonsigning deaf, do not invent languages, even though they are exposed to an environment rich with describable experiences (ibid.).

While Pinker, unlike Bickerton or Chomsky, attempts to fit his model within evolutionary theory by acknowledging the importance of environmental factors and gradual changes over time, his model is based upon the linguistic capacities of human (English speaking) adults. As demonstrated in the previous paragraphs, his computational theory is not valid for human children, therefore its usage as a model for the linguistic capacities of early hominids is questionable.

The theories proposed by Chomsky, Bickerton and Pinker are all essentially intellectual models which have questionable empirical support from child development studies. In order to develop a theory of language origins which would be relevant in terms of human evolution, it is important to examine how human children acquire language, the physiology of the human vocal tract, language specific areas of the brain, and the vocal communication of nonhuman primates.