Learning is unlearning

Apr 13, 2011 by

In yesterday’s posting I described some recent research that suggests that when it comes to language, babies are not born as a tabula rasa. Rather, all humans are born pre-wired (or pre-programmed) to learn a human language. One such tool that helps a baby to learn her “mother tongue” is the ability to descriminate her mother’s voice from those of other people around her.

But even after the baby figures out who to listen to, the task of acquiring the target language is monumental. The next item on the baby’s language acquisition agenda is figuring out what sounds her target language has. Here too, a baby is born pre-programmed with the list of possible human speech sounds, and the task of learning the sounds of her target language is reduced to the task of “unlearning” the sounds that are not necessary. Thus, research conducted by Dr. Janet Werker and others shows that smaller infants (6 to 8 months old) who are acquiring English as their native language are able to make distinctions between sounds that are not used in English. For instance, they can hear the difference between a dental and a retroflex pronunciation of d — a distinction that is made in languages like Hindi, but not in English. For non-trained adult English speakers it is notoriously difficult to hear the distinction. And so it is for an older infant, 10-12 months old. But younger babies are still able to hear that difference, just as clearly as they (and we, grown-ups) hear the difference between a /p/ and a /d/.

Another task on child language acquisition agenda is to figure out how to break up the speech stream into words, something that we, as adult speakers of the language do automatically. But to appreciate how difficult the task in front of a child is tune in to a radio station broadcasting in a foreign language: not only do you not know what the words mean, but you can’t even reliably say where one word ends and the next one begins. So how do we figure this out?

There is a universal tool-set for breaking up the speech stream into words, and languages vary as to which tools they use. One commonly used tool is prosody (intonation): words typically have one main stress. While this clue cannot reliably tell us where one word ends and the next one begins, it can at least tell us how many words we are dealing with.

Another way to draw boundaries between words is to pay attention to phonological processes that happen only within words or only at the word boundaries. And again, languages vary as to which phonological processes apply where in relation to a (phonological) word. One example of a rule that commonly applies at the word boundary (especially at the end of a word) is obstruent devoicing. This a rule that turns voiced stop, fricative and affricate consonants into their voiceless counterparts. Languages that have this rule include German and Russian. For example, in Russian the words /kot/ ‘cat’ and /kod/ ‘code’ are pronounced exactly the same: [kot]. Thus, in the case of ‘code’ the voiced /d/ is devoiced. How do we know that these words are not underlyingly the same? When the word appears in a different form, say, instrumental singular, the two stem-final consonants do not sound the same: [kotom] vs. [kodom].

While word-final devoicing is an example of a phonological rule that applies at the word boundary, other phonological rules may apply only within a word. One such rule is vowel harmony. Recall that the presence of vowel harmony in a language restricts the possible qualities of vowels in a word: if rules of vowel harmony are applicable in a given language, all vowels in every word must be either front or back. For example, a vowel harmony language would not allow an [u] and an [i] in the same word. In a language like that, words like [tubo] and [bedi] would be possible, but words like [tubi] and [bidu] would not be.

How would this help a child trying to break up the speech stream into words? Once the vowel backness changes, it must be a new word, not a continuation of the previous one. Research conducted by Toby Mintz of the USC shows that small infants are sensitive to the vowel harmony patterns even if the language they acquire does not have the rule of vowel harmony. For example, infants acquiring English (which does not have vowel harmony) are sensitive to this pattern: if presented with a made-up language that uses vowel harmony, they can pick out the pattern. However, older infants lose the ability to discriminate this pattern. In effect, they “unlearn” the vowel harmony rule as part of their acquisition of English.

In other words, learning one’s native tongue involves not just learning what sounds and what phonological rules the target language has but also “unlearning” those sounds and rules that are used in other human languages. Thus, language acquisition turns a child from a “universal listener” to a “native speaker”.

Previous Post
| Next Post

Related Posts

Subscribe For Updates

We would love to have you back on Languages Of The World in the future. If you would like to receive updates of our newest posts, feel free to do so using any of your favorite methods below: