The Poverty of the Stimulus Argument

The main argument for the position that we have some innate linguistic knowledge is the PoS line of reasoning. It’s a horse that was not only raced past the barn but also beaten to death. It basically says that the input linguistic data a child receives is by far not enough to account for the knowledge of its language that it acquires. The conclusion drawn from this is that the child has to have some prior knowledge, it comes pre-programmed (so) to speak (or in more biological terms, it has a ‘language instinct’). Some parts of language obviously have to be learned (e.g. your vocabulary: it’s arbitrary whether you call something ‘apple’, ‘pomme’ or ‘malum’) but for other parts it could be that this is not the case, e.g. for syntax where it’s conceivable that the way syntax works is already there and (depending on the program you work in) the child just has the task to set the parameters by learning e.g. whether its language is OV or VO.

To relate this to the previous post: I don’t know if that is true.

In all those years I’ve heard and read this argument uncountable times (the beating part mentioned above), it sounds pretty convincing and I usually tend to believe it. My problem again is that I’ve never seen an argument that is qualitatively different than the one you can find on Wikipedia ( – and by that I mean the kind of evidence that is offered for that. I’ve never seen a quantitative account of what kind of rules a child would have to learn if it were a blank slate, how hard it would be to extract these rules from the data that a child is confronted with, and how plausible it is that a human brain would be able to achieve this feat (regarding its memory and computational capabilities). And I don’t know anything about this. And neither do most linguists, I assume.

After linguists have provided a (work in progress) things a child has to learn to know a language, computer scientists should come in, tell you quantitatively how complex the child’s task it, what algorithms can be used etc… Then neuroscientists should come in, tell you which of these mechanisms are biologically plausible, how much information brains can store and so on. After that you should get some estimates on how implausible it is to learn a language from scratch – and I’m perfectly fine with order of magnitude estimates.

I’m not saying that this kind of paper doesn’t exist, I just don’t know of any (which doesn’t mean much), if anyone knows of something like that let me know. But this would elevate this innateness assumption from the status as lore to an empirically validated principle.

I find it pretty dubious that this argument is treated as fact by most linguists – after all, the opposite argument could sound similarly persuasive (the child has several years of learning a language, it is bombarded with input, etc…). And I doubt many linguists who believe that argument could easily link to articles that meet above criteria, let alone have read them.

Again, I’m not saying that this means that the PoS argument has to be false; and I’m also not putting up the strawman saying that you can only claim something if you have read most of the relevant papers and understand most details of all the scientific disciplines involved. What I’m saying is that if you believe this argument – especially if you find it ‘obvious’ and an apparent truth of your tribe – then at least know of studies/articles where this has been reliably shown by experts via the usual scientific tools.

Otherwise you have to present this way more carefully, e.g. as a conjecture, and not in a way that leads students and the public to believe that this is basically true and an accepted scientific fact like evolution.

At least for me it has been the experience that this didn’t happen. It was always presented as the prose argument you can find on Wikipedia, without reference to studies that really showed this. Which is sad if these wonderful studies exist.


One thing I forgot to mention is that there is huge variability in what is claimed to be innate. Sometimes it is just ‘the obvious fact that humans learn to speak just like birds learn to fly but not the other way round’, sometimes that it is a whole program with a lot of pre-programmed parameters whose values have to be set.

In my personal experience it is not often made clear for which version someone argues when they say ‘language is innate’. If one assumes no intention then this is unfortunate unclear writing; on the other side this can look like a motte-and-bailey argument where you argue for a strong version but when pressed defend yourself by saying that surely nobody would contest the claim that humans have a disposition to acquire language, and you never said anything else.


Here is a link to a nice blogpost by Martin Haspelmath over at Diversity Linguistics Comment about similar issues:


This entry was posted in Cognitive Science, Evolution, Linguistics and tagged . Bookmark the permalink.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s