Language Acquisition/Printable version
| This is the print version of Language Acquisition
You won't see this message or any elements not part of the book's content when you print or preview this page.
Nature or Nurture/Animal Language
In our introductory book to linguistics, we looked at design features of human language that differentiate it from animal language:
- Reflexivity: We can use language to talk about language, e.g. Don't use a preposition at the end of a sentence!
- Displacement: We can talk about events spatially and temporally distant from us, e.g. I saw a strange guy in the pub yesterday.
- Arbitrariness: There is little connection between signifier and signified. For example, the word moon is completely random; the moon does not make a 'moo' sound like a cow or otherwise exhibit features that would justify the use of moon as a sign.
- Productivity (or creativity in Chomskyan linguistics): There is a limitless number of potential utterances in the language. The car ate the marshmallow-hating helicopter, for example, is a sentence that nobody has seen before, but any speaker of English can understand it.
- Cultural transmission: Culture is passed on from mother to child by language acquisition. This culture is not innate, unlike animal language. Cows are biologically designed to go moo, and pigs to go oink; there is no culture transmission there.
- Duality: Human language has two distinct levels, the sound and the meaning. Sounds are limited but meaning is not. We can play with the sounds in our language to produce different meanings. Consider the minimal pair made and paid. Made is broken down to the phonemes /m/, /eɪ/ and /d/, and we can swap /m/ for /p/ to produce a word with a different meaning altogether. However, animals do not have this two-level distinction. A pig may say oink, a word that means 'I'm starving', but it cannot reformulate the sounds to produce a different utterance, say, roink, replacing the glottal stop /ʔ/ at the beginning with /ɹ/. The signifier and the signified are processed on the same level, rather than two different levels.
Nature or Nurture/Linguistic Nativism
Linguistic nativism is the idea that language is 'hard-wired' into our brains at birth. The hard-wired knowledge is called Universal Grammar (UG), a grammar common to all languages in the world. This unit introduces you to this idea.
It is often said that UG is like a seed. Obviously, seeds of apple trees are not trees at birth, but given nourishment, such as sunlight and water, they will grow into apple trees. They will not grow into orange trees or pear trees.
UG is the knowledge of language innate to all humans. Everyone, save for those with certain language disorders, possess UG, according to Chomsky. UG morphs into competence, which we will later see. Its distinction from performance is an important part of Chomsky's work, and will be discussed in this chapter.
UG does not claim the existence of a set of rules common to every language, say, that the subject should be the first word of every sentence. This is clearly false, since languages with word orders other than SVO and SOV do exist, uncommon as they are among the major world languages. German, for example, has V2 word order, in which the verb must come second in a sentence, but any other element - subject, object, adjunct - can be placed first. Some linguists, most notably Joseph Greenberg, have attempted to generalise sets of such rules, such as this:
(1) In conditional statements, the conditional clause precedes the conclusion as the normal order in all languages.
This is a linguistic universal, but it is not part of UG. UG does not aim at stating rules. Instead, UG attempts to generalise a set of principles that govern language. We will look into the concept of principles in this unit.
Table of Contents[edit | edit source]
- Competence and Performance
- Principles and Parameters
- Poverty of the Stimulus
- Language Acquisition Device
Nature or Nurture/Competence and Performance
To understand the notion of UG, the distinction between competence and performance is essential. Competence is inside the language user's head, while performance is what the language user actually produces.
E-language and I-language[edit | edit source]
The notion of E-language ('externalised' language) is associated with Bloomfield's structuralist tradition of linguistics. Linguists who study E-languages collect utterances to create corpora. They then attempt to describe these data.
Chomsky believed the study of E-language was insufficient. E-language only looks at what people say. Any corpus, regardless of size, only contains a small percentage of what language users can produce. (This is because the number of sentences that can be produced is essentially limitless - we will look at this below.) Therefore, he proposed the notion of I-language, which describes the linguistic knowledge of language users. Language users know how to form sentences that conform to the grammatical rules of the language. They will not produces sentences like *Him is good very, and they can also tell that such sentences are wrong. This constitutes the linguistic intuition of language users.
Competence and Performance[edit | edit source]
Creativity[edit | edit source]
Nature or Nurture/Poverty of the Stimulus
The poverty of the stimulus argument (POS/POSA/APS) is of great importance on the nativist side. It argues that children do not receive sufficient input to generalise grammatical rules through linguistic input (or primary linguistic data) alone. Introduced and championed by Chomsky, it continues to have far-reaching influence in the nature/nurture debate. Opponents have criticised the argument as logically flawed or empirically baseless, whereas supporters have made attempts to refute these counter-arguments and provided empirical data of their own.
Chomsky did not clearly state the argument when it was first introduced. The lack of clarity has led to much confusion over the nature of the argument. Different scholars have introduced different premises to the argument, which were laid out in Pullum and Scholtz's critical review. Before we try to define the POS argument, let's look at what Chomsky calls Plato's Problem, a classic example of rationalism.
Plato's Problem[edit | edit source]
Chomsky's rationalist view towards language acquisition harks back to a Plato. Plato wrote that by drawing geometric figures in the ground, Socrates demonstrated that the slave was initially unaware of how to find twice the area of a square. Socrates then said that before he got hold of him ,the slave, who had been picked haphazardly from Meno's entourage, had spoken 'well and fluently' on the subject of a square double the size of a given square. Socrates then drew a second square figure on the diagonal so that the slave could see that by adding vertical and horizontal lines touching the corners of the square, the double of its area was created. He got the slave to agree that this was twice the size of the original square and said that he had 'spontaneously recovered' knowledge he knew from a past life without having been taught. Socrates was satisfied that new beliefs were 'newly aroused' in the slave.
Chomsky believes that a similar argument applies to language acquisition. Children's language ability are 'aroused' by primary linguistic data. The innate ability that was there had to come from some source. Chomsky calls this mystery 'Plato's Problem', and he believes this innate knowledge to be biological.
What is the POS argument?[edit | edit source]
Pullum and Scholz identified a number of separate claims made within typical descriptions of the POS argument, regarding the characteristics of child language acquisition and primary linguistic data. We will examine all of these claims one by one.
Characteristics of acquisition[edit | edit source]
- Speed: Children acquire language swiftly.
- Reliability: Few children fail at acquiring language.
- Productivity: Children can produce an infinite number of sentences.
- Selectivity: A number of grammars can be deduced from childrens' input, but they do not pick the wrong grammars.
- Undetermination: The grammars that children learn cannot be determined by the data.
- Convergence: Children arrive at similar mature competence.
- Universality: All children acquire human languages sharing similar traits that make up UG.
Characteristics of data[edit | edit source]
- Ingratitude: Children do not receive rewards for good performance,
- Finiteness: Generative grammar assumes that a finite set of grammar rules generates an infinite set of sentences. If empiricist claims were true, children would only be able to acquire what they have learnt from exposure to E-language data, which are finite in nature. This contradicts the fact that children are able to produce infinite sentences.
- Incompleteness: Children are not exposed to sufficient data to acquire language. They are not directly taught the language, either.
- Idiosyncrasy: Different children are exposed to vastly different linguistic data.
- Positivity: Children are not exposed to negative data. For example, children can learn that Bill is nice by listening to this sentence, but have no way of knowing that Bill are nice is not an acceptable alternative.
- Degeneracy of the data: The now-obsolete idea claims that parents produce erroneous input, which would, according to empiricist claims, result in abnormal language development in children. In light of empirical data, the claim has been refuted and is no longer held by nativists.