Learning even the simplest language rules is not easy
A large interdisciplinary NWO research project attempted to discover the cognitive origin of the human ability to learn linguistic rules. This is not so simple, according to linguist Andreea Geambaşu and her colleagues. PhD defence on 11 December.
Geambaşu wanted to replicate an often-cited study by Marcus et al. from 1999. The experiments show that seven-month-old babies can learn simple patterns such as XYX, XXY or XYY in a short phase in series of syllables such as le-di-le, le-le-di or le-di-di. This research showed that babies are able to generalise a rule, that is, to adapt it to new information.
Attention is important
The researchers’ intention was to replicate the experiments of Marcus et al. and to continue to investigate learning of more complex rules. However, despite the attempts to reproduce and extend the experiments exactly, their results differed from the original results: they found consistent evidence that, regardless of the pattern offered in the learning phase, babies preferred patterns with directly repeated elements, such as XXY or XYY. The researchers discovered this by playing different sound combinations and observing which ones the babies chose to listen to longer.
Looking at photos
In a subsequent experiment, the researchers studied babies aged twelve to fourteen months. Infants of this age have just started talking and know the meaning of many frequently used words. Geambaşu and colleagues hypothesized that vocabulary knowledge may help babies learn linguistic rules. The babies were shown pictures of objects they knew and whose names they already knew, of objects they knew but whose names they did not know, and of nonsense objects. The pictures were arranged according to the simple XYY or XYX rules as in the previous studies. The ability to learn rules seemed to improve if the babies knew the words for the objects presented and if the rule they had to learn contained a repetition (XYY). Again, repetition seemed to be important for babies in their learning process.
Eight different languages
In yet another study, more than 40,000 oral expressions from children up to one year old from eight different languages were analysed. Another reason for doing this was to gain a better understanding of what types of patterns babies produce at the age at which the literature indicates they should be able to learn simple XYX-type rules. The study showed that, contrary to previous assumptions, infants do not start speaking by producing only repetitive syllables such as ba-ba-ba, and only later produce non-repetitive syllables such as ba-di-go; they produce both repetitive sequences and non-repetitive ones from the very beginning. In fact, non-repetitive sequences (XYZ) were consistently the most common type of sequences produced across languages. In addition, infants produced the XYX pattern they are expected to learn in the perception experiments very infrequently. These patterns may be difficult to learn because they are so uncommon in natural language.
Geambaşu’s experience shows that learning language is incredibly complex and at the same time that the learning capacity of small children is enormous.
About the researcher
Two months ago Geambaşu became the mother of a son who will grow up learning four languages. The family lives in the Netherlands and, given the fact that Geambaşu is now the LUCL lab manager, this will continue to be the case for the foreseeable future. The new world citizen will therefore come into contact with Dutch a lot. Geambaşu herself is Romanian, her husband is Turkish, and together they speak English. Her son will not be confused by this situation, according to linguist Geambaşu. ‘It is important to create an interesting, motivating language environment for each language,' she says. Then the child learns quickly, although we still don't know exactly how.
In any case, Geambaşu now has a practical example right under her nose and can easily observe how language acquisition proceeds in this specific case. This is not scientific, but it is undoubtedly very interesting for a linguist.
PhD defence Andreea Geambaşu
Simple rule learning is not simple: Studies on infant and adult pattern perception and production
11 December 2018
Text: Corine Hendriks
Mail the editors