October 27, 2004
Word Memory Shifts From Sound To Meaning As We Age
Somewhere between the age of 5 and the age of 11 human minds shift from associating words primarily by similarity of sounds to similarity of meaning.
The study found evidence of an age-related, developmental shift in language, suggesting that younger children process words primarily on the basis of phonology, or sound, while older children and adults process words primarily on the basis of semantics, or meaning. The findings are presented in the article "False Memories in Children: Evidence for a Shift from Phonological to Semantic Associations," by Steve Dewhurst and Claire Robinson of Lancaster University, United Kingdom. The article will be published in the November issue of Psychological Science, a journal of the American Psychological Society.
To test whether children would make similiar memory errors based on sound rather than semantics, the researchers used a version of this earlier experiment. They developed a list of words in which each word had at least one possible rhyme, then presented the list to children aged five, eight, or 11, who were asked to recall the words after hearing them. The results suggested a developmental correlation between age and language processes: The 11-year-olds performed in the same way as adults and falsely recalled words that were semantically related to the lists; the 8-year-olds were equally likely to falsely recall rhymes and semantic associates; and the 5-year-olds falsely recalled words that rhymed with those presented in the lists.
I think adults tend to forget and underestimate the intellectual difficulty of being a child because adults do not realize just how much contextual information they have built up and rely upon. For a child so much more of daily experience is novel and has no framework through which it can be sorted and organized.
For another demonstration of how minds develop better ways of classifying information as they age see my recent post Ferret Visual Cortex 80% Active Even In The Dark.
I would be interested to see how this works with second languages. The link doesn't give a whole lot of information, but it seems that the participants are all native English speakers. I wonder if a person who has recently learned a language would process it more like an adult or more like a child.
I have paid attention to this exact thing as I have learned languages, and I think it is a mixture, perhaps more like an eight year-old, but definitely more semantic. In general, the "passthrough" from sound to semantics is very developed in adults, and we often do not have the same capability to attend to and remember sounds. My three-year-old will remember sound/word strings in shocking detail over long periods of time, and most very young children need to hear a word only once or twice to have it added to their vocabulary. We seem to lose that level of attention as we age, which makes it harder to learn or memorize new words, but also reduces our chance of confusing things at the sound level. When learning a language, in cases where I have consciously attended to the sound, I find the propensity to confuse sounds is increased. But that is never a successful way to learn a language anyway; repetition is the way that adults learn language, and this embeds the semantics subconsciously. So even with minimal study, most of the mistakes become semantic. For example, just yesterday I found myself getting confused about whether to use "wo ming bai" or "wo ji dao" (I understand, vs. I know) in Chinese. I wanted to say "I know", but both phrases popped into the semantic slot in my mind, and I became confused that my mind was supplying two different-sounding phrases to fill the same semantic slot. I often have situations like this where the semantics are close.
Finally, I would note that there is another kind of confusion (at least one more). I am thinking of the confusion caused by applying the wrong grammar template when constructing an output. For example, my child talking about, "the monkeys are hugging them chother", my use of "amn't" as a child. Like in the meta-model/milton-model distinction. I notice this happens when studying languages too, and it seems to be a stage in-between sound and semantics for child development. Around 2 or 3, children get good at generalizing and applying grammatical templates to create new structures; and after a certain age, we become less adept at this -- people who learn the wrong grammar as children often fail to correct it as adults. So to the extent that we attend to and notice the application of grammar templates, we run the risk of mixing things up; but once it is internallized and no longer where we direct attention, we make less mistakes. Perhaps grammar comes after semantics, but I don't think so.
It begs the question of, if the confusion moves to the next level as we master each successive level (sound, grammar, semantics), is there another place we push confusion once we have mastered these three?
This is in regards to Xavier. That is an interesting question. My mother and I were learning a new languege together, and while I picked it up easily, my mother ended up having to turn to childrens guides to get a tiny foundation. She was never able to get a grasp on learning it and she eventually gav up. So I think it varies among people on how well they process things. I don't think it is hereditary either.