Motherese

Child directed speech and language

It is often noted, originally by Catherine Snow (1972) and Elizabeth Bates (1976), that there is a characteristic way of talking to small children, used particularly by mothers, with short, simple words, some modified to resemble children’s speech, with as few ’embedded’ clauses as possible, with sentences only in the ‘active voice’, with an exaggerated intonation.

Originally this was characterised as ‘motherese’. Later, reflecting the view that the term, motherese, was demeaning to women, it was generally replaced by ‘Child directed speech and language’, CDSL, or some variant.

Snow and Bates proposed that the simplified forms of CDSL would constitute templates around which the vastly more complex structures of fully-developed language could be built one by one in a natural way. Call this the ‘CDSL hypothesis’.

Since the 1970s, the CDSL has been adopted by many others since, including Cat Bohannon (2023).

Given that the rate of CDSL use obviously varies across families and individuals, it might seem that the CDSL hypothesis could be tested by taking a sample of families, measuring the CDSL and the rate of language development. The hypothesis would be confirmed by a positive corelation. Such a co-relation is commonly found. But it does not confirm the hypothesis. It is possible that CDSL is not so much a language acquisition device but more an expression of bonding with babies and small children. Such an idea is strengthened to the extent that something like CDSL seems to be sometimes used with small pets – who are surely not expected to learn to talk. It would be quite unsurprising if speech and language development was facilitated by strong bonding between parents and children.

There is a weaker version of the CDSL hypothesis by which children learn best from language within the range of their understanding. But this is difficult to test.

What if the CDSL varies to zero? What if some children never experience any CDSL from either of their parents? In a survey of CDSL by Clare Galloway and Brian Richards (1994) it is revealed that although CDSL is common across cultures, it is not universal. Yet language still develops in cultures where CDSL is not used.

And what if CDSL is taken to the limit? Children would have no experience of any complexity. It is hard to see how normal acquisition could proceed. But CDSL is not commonly taken to the limit. “Darling, do you want your milky?” is not commonly thought to be complex or in breach of a strict CDSL protocol. But the structure here is not simple at all. The fact that darling precedes the rest of the sentence marks it out as some special sort of entity, in this case, as a form particular to discourse. What does the child learn from the extra syllable in milky, with the K part of the second syllable, thus losing the complexity of LK on the right edge of the syllable, a sort of structure not allowed in many languages? What about the little word, do? How does it get into the sentence? And what is it doing when it gets there? Even if the child does not ask questions with do, there is an important lesson about the structure of English from learning to understand them.

Or consider “Do you want the jam you had yesterday?” It is not easy to see how this could be simplified without losing the sense. To ask the question, the relative clause in you had yesterday has to be included in the structure.

On such reasoning, it is the conclusion of most linguists who follow some version of the framework here that the CDSL hypothesis is misleading or mistaken. It seems that this conclusion may be sometimes misconstrued as an insensitivity to feminism. Bohannon, for instance, sets out the CDSL hypothesis with great confidence, but without mentioning that there is a strong and substantial alternative view.

But the CDSL hypothesis not withstanding, every expression of parental bonding is obviously invaluable for all aspects of child development – in ways going far beyond speech and language.