How does a human baby, still not able to dress itself properly, learn the first language with remarkable speed? Is language something it is born with, simply activated when exposed to certain stimuli or something it gets from others? There has been a debate since Chomsky, in the 1950s, claimed that children came to the world equipped with a prototype of all languages.
Represented by Chomsky, the “generatists” or “innatists” believe that language is innate, programmed in the level of human genes (Pinker, 1994). According to this view, a series of principles known as Universal Grammar (UG) underlie all languages. When presented with data, namely what their parents and other caregivers say to them, children simply set parameters of the UG in their mind, and thus construct the specific grammar of their mother tongue. Exposure to English makes them set “preposition comes first in a preposition phrase” while exposure to another may lead to “postposition appears at the end of a postposition phrase”, but without exposure to any languages the principle will remain “adposition + place = adposition phrase” and whether the adposition is a preposition or a postposition is uncertain.
A grammar is unlearnable without negative feedbacks, according to Gold’s Theorem (1967), an argument for the innatists. Caregivers of babies will usually not correct their errors, nor can the environment offer flawless input (see Chomsky 1976, Evans 2014), yet babies master the spoken language before they go to school, without generating alien sentences, so in a deep level, they must have an innate guide. It has nothing to do with learning strategies, task design, or other factors (Cook, 1989). Mere presentation of the language suffices.
Though Chomsky himself was never a computer scientist, his theory has influenced ideas on how computers could process natural languages like humans. The prospect of artificial intelligence understanding thousands of languages all in depth – because they are the same, just distinguished by some parameter settings – is attractive.
The theory has, however, been frequently challenged. Any single field of linguistics will show so great fundamental differences among languages that it becomes hard to believe there exist some universal properties (Evans and Levinson, 2009). Construction of a generative grammar for only one language raises problems (Berwick & Chomsky, 2016). Its own advances have been slow as both access to and research on human brains are not easy (ibid.).
Besides, if language grows like an organ in a child’s body, decided by genes, then forgetting it should be like having an organ removed. Following this thread, I myself have already lost a part of me without even realizing it. My parents tell me that I spoke a fluent local dialect when I was very young, but forgot everything of it because of only using “the standard Mandarin” after leaving the province. The dialect of my hometown, unlike skills of swimming or riding a bike, did not brand itself into my implicit memory.
While Universal Grammar is like a map that you connect with surface properties of a certain language when you recognize its manifestation, usage-based learning explains that you draw the map yourself after observing the reality. Where Chomsky claimed that the data offered by the environment to a child is too “limited and imperfect” (1976) to abstract complex knowledge, social constructionists say nay. Imperfect, maybe, but enough to guide them. A language, like a dance, is cultural other than biological (Sampson, 2005), and must be intentionally taught.
We are born with legs, but not born to dance; similarly born with what we need to develop a language, but not born pre-equipped with it. It is noted that infants are able to and would like to look for patterns, not only in languages (Evans, 2014). Besides, human cultural intelligence allows intention-reading. Babies share attention with their caregivers and imitate the adults’ roles to maintain communication. As for why they do not generate many grammatically wrong sentences (like an OSV sentence in English) without correction from caregivers, possibly they “take absence of evidence for evidence of absence” (ibid.); there is too little input of alien expression forms for them to imitate.
Opponents of the innateness position have claimed more evidence to argue against its defenders. They have argued that Chomsky’s “poverty of stimulus” is based on his logic instead of empirical evidence (see Evans & Levinson 2009, Evans 2014, Sampson 2005), and that Gold’s theorem is unlikely (Scholz 2004). A single counterexample would be enough to invalidate a “universal” claim, and as Sampson concluded, the social constructionists found more than one thing that Chomsky said would never happen.
They label Universal Grammar a groundless myth, but the innate hypothesis holds a special appeal that linguistics can hopefully be like mathematics, a complex system based on rather simple general principles, where deductive reasoning is sufficient to establish more rules that are specific. I will not be surprised to see new innatists in the future, and as long as the brain remains mysterious, I cannot confidently say that there will never be biological proof for them.
SHUYI LAN, English Language undergraduate, University of Chester, UK
References
Gold, E.M. (1967). Language identification in the limit. Information and Control. Pp.447-474.
Pinker, S. (1994). The Language Instinct, the New Science of Language and Mind. London: Penguin.
Language Debates 