This post is a bit of an exception.; I don’t generally plan to post book reviews, but I really enjoyed Michael Tomasello’s Constructing A Language and I’m hoping that this will be an informative post rather than just a review.  It’s also a good opportunity for me to start writing about language, which is something I’d like to do even though it’s not my area of expertise. In addition to being a well-written book on its own merits, Constructing A Language really struck me because it successfully challenged what I thought I knew about language acquisition and made me reconsider my position, which is always a remarkable and rewarding experience.  This book gave me the wonderful gift of educating me out of my complacent acceptance of an idea and forcing me to really think about it.

I am not a linguist; I’m just a biologist who loves languages and enjoys reading, thinking and learning about them.  I was exposed to the idea of an “innate language faculty” while I was growing up and took to it quite readily; later, I read books like Words and Rules (by Steven Pinker) and The Unfolding of Language (by Guy Deutscher) which reinforced this idea.  The idea of an innate language faculty was first proposed by Noam Chomsky about half a century ago in response to behaviourist models of language learning which suggested that children learn language through conditioning, similar to the way a dog will learn that a bell means food (and so start to salivate). Chomsky convincingly argued that inductive reasoning was insufficient for language learning and that children weren’t exposed to enough linguistic material to be able to learn all of the complexities of language (see the “poverty of stimulus” article at Wikipedia for more about this).  His solution was to propose an innate language faculty along with a universal grammar; that is, humans are biologically endowed with a specialised mental apparatus for acquiring language which includes a universal grammar composed of abstract, formal rules.  In order to learn a particular language, a child only needs to map aspects of its grammar (such whether the verb comes before or after the object) onto this universal grammar; in this way, the poverty of stimulus problem is overcome because the aspects of language that Chomsky argued would be impossible to learn are provided by the general, abstract underpinning from the universal grammar.

I’ve been pretty convinced that we have an innate language faculty for some time, so I bought Tomasello’s book in order to expose myself to the “other” viewpoint — that humans don’t have an innate biological faculty for learning language. The book sat unread on my bookshelf for an unusually long time; to be honest, this was partly because I had vaguely dismissive attitude towards it.  I wanted to read it mainly out of a sense of fairness — to hear what the “other side” had to say — but I expected to readily poke holes in the argument and come away unimpressed and perhaps even more confident in my conviction about an innate language faculty.  Instead, the opposite happened; despite my quite critical approach, Michael Tomasello made a convincing case for the constructionist viewpoint, forcing me to reconsider my position.

Constructing A Language argues against an innate language faculty on two fronts.  Tomasello makes the case that there is no poverty of stimulus because children don’t learn language through simple associative condition, but in a socio-cognitive context where they can make use of additional information; furthermore, he presents evidence that seems to be at odds with the idea of children learning language with the help of an innate, universal grammar.

To resolve the poverty of stimulus problem, Tomasello points out that children are aware of other humans as intentional, communicative beings like themselves; they share attention with adults and try to understand their communicative intentions.  Both of these behaviours are rooted in an awareness of other humans as mental beings and begin to develop in children 9-12 months old, which is around when they start learning a language.  These mental faculties (coupled with a powerful ability to detect and analyze patterns) allow children to extract much more information from a situation than would be possible by simple association.  Shared attention with an adult (e.g., playing with something together) limits the scope of relevant entities, while reading an adult’s intentions to understand their communicative goals (i.e., why they said something) provides additional useful clues.  For example, in one experiment a two year old child played with some toys with her mother and another adult; her mother then left for a brief period during which the experimenter introduced a new toy.  When the mother returned, she looked at the toys and exclaimed “Oh look! A modi!”.  The child was able to learn that the new toy was a “modi” just based on this experience, since her mother would be unlikely to be excited by the old toys.  By taking into account her mother’s intentions and mental state, a child is able to eliminate many possible (but incorrect) meanings for a word, allowing her to learn much more quickly and effectively than she could by simple association.  This sort of learning also makes it much easier to grasp the difference between similar words; for example, by trying to understand why an adult would choose to use “share” instead of “have” in a particular context, a child can quickly learn the distinction between the two words.  One of the things I really enjoyed about this book was the plentiful use of examples like these; in addition to supporting the argument being made, they were often fascinating and interesting observations in their own right.   (As an aside, it’s interesting to consider the recent findings about ravens treating each other as entities with a mind in this context.)

The second thrust of Constructing A Language is to present data that conflicts with the idea of an innate language faculty.  If children really do learn language by setting switches (based on what they hear) in an innate universal grammar, then they would be expected to learn an entire grammatical rule at once and apply it equally well everywhere.  In other words, a child who is able to understand the sentence “Close the door” should have learned (by setting a switch in her universal grammar) that verbs always precede objects in English.  Tomasello argues that this isn’t the case: children younger than three seem to be relatively conservative learners and don’t readily generalize syntatic rules (such as word order or case marking) from one context to another.  While this makes sense if children are learning language through a general pattern matching mechanism (coupled with the intention reading discussed above), it’s not really in keeping with the idea of a universal grammar, which would predict that a general rule is learned once a switch is set.

Despite my initial reticence, Michael Tomasello’s Constructing A Language turned out to be a favourite amongst the books I’ve recently read.  It’s well-written, pleasant to read and full of interesting examples. I also found myself constantly jotting down notes for things to follow-up on from the bibliography.  The book is structured to parallel the process of language learning, starting from individual words and moving through fixed phrases to simple and more complex grammatical constructions.  There’s also more to the argument than the bare-bones summary I’ve presented here, but this post is already too long.  If you’re interested in the subject and want to learn more, I thoroughly enjoyed the book and recommend it enthusiastically.  I’m looking forward to going through it again and I would love to read more from both sides of the debate.

On a personal level, it was really invigorating to read such a convincing, coherent and cogent argument against something I thought I knew — there’s nothing quite as invigorating as being challenged in that way!  Realizing that you were wrong shouldn’t be seen as a flaw, but as an indication of an open, receptive mind engaged in fruitful dialogue.  It’s not something to be ashamed of, but rather a sign of continuing growth and potential.