Data mining taught her to say “shut up”?

Hannah is an extraordinarily chatty girl. She talked before crawling, before walking, and while other toddlers gleefully climb the apparatuses at the playground Hannah’s content to go up and down a ramp, and tell you all about it. She’s a talker, not a fighter. It’s remarkably entertaining – she is constantly coming up with new commands and stories for us and sometimes I have no idea where she got them. The other day she marched into the living room and said “Ok, here’s the plan.” I didn’t catch the rest, but there was clearly a plan of some sort.

Unfortunately, she also learns words that I would rather she didn’t, and it’s drawn our attention to aspects of our relationship that might be hard to explain to outsiders. One morning, running around the kitchen, Hannah started saying Shazbat. That’s weird, I thought, because Mork and Mindy hasn’t been on in probably 30 years. A few laps of the kitchen later, I realized she was actually saying “shut up.” Once she got the pronunciation right it was unmistakable.

I’m embarrassed to admit that it wasn’t the question of where she learned to speak such a cruel thing that was hard for me to answer. Brian and I have a charming and quirky relationship – a typical evening starts with a “How was your day, dear?” followed by a “Shaddaaaap.” It’s sweet. The Shaddaaap is just as likely to be delivered by me as it is by Brian. And now, apparently, just as likely from Hannah. I’d like to say that that’s the only bad thing she’s ever said, but that’s not true. She’s been in hearing range when I’ve misplaced my keys. No one ever said I was super mom.

What’s interesting to me is not where she hears the bad words, but how she decides that they should be part of her lexicon. Easy words clearly come first – for Hannah it was “wassat?” It’s like the gateway drug of baby speak – with that one simple amalgamation of “what’s that” she was able to quickly move on to harder words. But why was “fence” in there before “kitty”?

There are people out there who are more badly afflicted with OCD than me, and they have compiled statistics on what words tend to come first when babies learn to speak. Called the Bates-McArthur Communicative Development Inventory, this database compiles the frequency of a word’s use as a child learns the language. You can access the database for both English and Spanish and take a look at the words that babies or toddlers tend to say for each month of life (http://www.sci.sdsu.edu/cdi/lexical_e.htm). For example, at one year, 35.2% of children say “uh oh,” while only 1.1% say “blanket.” “Shut up” wasn’t in the database.

You can see quickly that simple words and sounds, like “baa baa” or “uh oh” appear early in baby speak. These simple words are like the low-hanging fruit. Once these words are mastered and more complex words are attempted, a toddler enters the time called the “language explosion” which can vary quite widely but is generally around 18 months. Some research suggests that this explosion is the result of brain development that allows children to begin to process language. Conversely, the statistical distribution of simple versus complex words suggests that the language explosion corresponds to the time when a child has mastered simple words and moves on to those words that are both more difficult and more numerous. If word difficulty is distributed likeĀ  a bell curve, then the language explosion could simply correspond to the time when the toddler reaches the steeply rising portion of the curve (AAAS news 2007 )

However, this theory does not account for the highly social context in which children learn language. If all words were heard by a child at the same rate, then the simple metric of word difficulty might have more explanatory power. But children learn language from parents and caregivers who all have unique ways of communicating. For example, I’m sure that not all families lovingly say “shut up” as much as we do. However, teasing apart the context in which a non-verbal child starts to understand words is quite difficult.

Luckily, these metrics are much more simple to determine for adults and can be measured by a simple game of word association. For example, if I say the word “dog”, most people come back with the word “cat,” and if you test enough people with enough words, you can compile a very thorough database of word associations. Furthermore, you can look at these associations graphically and use graph theory statistics, such as the measurement of a word’s connectivity. If you consider words to be nodes in a network, the links between them are these simple associations, and the more links a word has, the more contexts in which it appears. For example, the word “bath” is often reached from the word “soap”, because they have a common context, but it can also be reached by “bird,” because those two words can form a compound word. In this simple example, the word “bath” has two links, and this is the measure of a word’s connectivity. So here’s a thought – does this connectivity of a word predict its time of acquisition by children? Turns out it does. In other words, if a child hears a word in a number of different contexts, that word is easier to learn (Hills et al. 2010. Journal of Memory and Language. 63:259).

This type of learning is much like data mining, wherein large amounts of data are sifted to find statistical relationships. In a test of this hypothesis, Linda Smith and Chen Yu of Indiana University taught toddlers new words by showing them two objects and reading their names at the same time, but without any information as to what word went with what object. Instead the objects and words were presented in new combinations a number of times. The children were surprisingly good at teasing it all apart and learning the new words, demonstrating how the word’s context was instrumental to the child’s ability to learn it (Science Daily 2008.)

So maybe the idea that context is everything can make me feel better about Hannah telling me to shut up. In the context of our family, I know that this is actually a loving and playful phrase to say. I just hope that the other kids at day home interpret it as such.

Comments are closed.