‘Simplicity’ and Linguistics

I was happily re-reading this excellent introduction to Linguistics this evening, when I stumbled across a sentence that annoyed me.

‘Obviously, [example X] is consistent with general scientific guidelines which require us to always seek the simplest and most elegant theory which is consistent with the data we need to explain.’ (Radford et al., 1999:330)

I’ve found that this kind of sentiment is so common in Linguistics that people barely mention it. Indeed, you might have read the above quotation and thought, yes, obviously. The idea is clearly influenced by Occam’s razor, which Wikipedia sums up as ‘a principle that generally recommends that, from among competing hypotheses, selecting the one that makes the fewest new assumptions usually provides the correct one, and that the simplest explanation will be the most plausible until evidence is presented to prove it false.’ Apparently, the principle is attributed to William of Ockham, hence the name. I think that this principle is so deeply embedded in our understanding of the world that most of us have never really thought about it. I must confess, when I started to notice I’d accepted this principle without thinking about why, I started to find it deeply dissatisfying. Who says that the hypothesis which makes the fewest new assumptions usually provides the correct one? What evidence is there of this? What governs this principle? Is it not more of a methodological suggestion of how to test theories, rather than an existential fact about stuff? The only times I have ever seen Occam’s razor discussed in any discipline is when it is appealed to as a kind of universal law which is assumed – obviously – to be true. I’m not necessarily saying I think it’s not true, just that very few people (in syntax at least) seem to challenge this assumption, something which has pervaded linguistic thought for a long time and which has reached its peak in the Chomskyan tradition.

Indeed, the Minimalist Program, which is currently the most fashionable theory of syntax, is underpinned by this principle, and this is implicit in its name. The following extract comes from an introductory textbook to the theory.

‘As in any other domain of scientific inquiry, proposals in linguistics are evaluated along several dimensions: naturalness, parsimony, simplicity, elegance […] etc. […] Put another way, the task is to find a way of taking the platitude that simpler, more elegant, more natural theories are best and giving them some empirical bite.’ (Hornstein et al., 2005:5-6)

Later, the authors introduce a caveat:

‘It’s possible that the language faculty is just “ugly”, “inelegant”, “profligate”, “unnatural” and massively redundant. If so, the minimalist project will fail. However, one can’t know if this is so before one tries. And, of course, if the program proves successful, the next question is why the language faculty has properties such as elegance and parsimony.’ (ibid., 7)

Am I the only one who thinks basing a whole research program around nothing more than a nice idea is a bit strange?

Naturally, the whole issue is underpinned by the question of whether the language faculty is ‘natural’, ‘parsimonious’, ‘simple’ or ‘elegant’. The first problem we have is that it is not abundantly clear what these terms actually mean with regards to language. I’ve come across few people who have attempted to explain what they mean by such terms. Usually linguists just use the terms as a sort of theoretical dustbin into which they shove any possible counterargument to their particular theory, because stating that one’s theory is more ‘simple’/’elegant’ is akin to trumping anyone’s ace.

If the language faculty is ‘simple’/’elegant’/etc. (and here we are presupposing that there is a language faculty of some sort), the only way we can access it is through the linguistic performance of a speaker, clearly. If the linguistic performance of a speaker is ‘simple’, then I might be able to be persuaded that his/her linguistic competence is ‘simple’. I would find it highly dubious if someone made the claim that linguistic performance is messy, but linguistic competence is not, as one cannot directly access the latter. What do others think?

As it happens, I’ve yet to be convinced that language (in the sense of performance) is simple, elegant, or parsimonious. If it were, why would we have to bend over backwards accounting for things in a complex way theoretically? Take plural formation as a banal example. In English, it’s nice and simple. You have your lexeme (word) in your lexicon (mental dictionary), e.g. book. To make a plural, you just whack on the suffix -s. That seems like a fairly straightforward default rule. If I asked you what the plural of gallyhop was, you’d say gallyhops, even though I can guarantee you’ve never heard the term gallyhop before and you have no idea what one is. Yes, there are exceptions like children and sheep, but we can account for those by listing them separately in the lexicon. But look at German. There are eight ways of forming the plural. If there is a ‘default’ way, which one is it? And how do we account for the other seven ways? Surely employing eight methods of forming the plural is hardly ‘simple’, ‘elegant’ or ‘parsimonious’?

It was put to me a while ago by one of my professors that the idea of the language faculty needing to be maximally simple might have had more to do with what computers could do at the time the idea came about than with what the language faculty can do. As little as twenty years ago, computers could store very little. I suppose it’s not surprising that one might think that storing information requires an inordinately large amount of space, and thus theoretical concerns were formed accordingly. Does anyone have any views on this?

References:

Radford, Andrew, Martin Atkinson, David Britain, Harald Clahsen, and Andrew Spencer. 1999. Linguistics: An Introduction. Cambridge University Press, Cambridge.

Hornstein, Norbert, Jairo Nunes and Kleanthes K. Grohmann. 2005. Understanding Minimalism. Cambridge: Cambridge University Press

Advertisements
This entry was posted in Linguistics and tagged , , , . Bookmark the permalink.

2 Responses to ‘Simplicity’ and Linguistics

  1. Richard says:

    My field isn’t linguistics (I’m a physicist) but I’ve always understood Occam’s razor to be a tool, not an axiom. You are seeing these words because we understand Maxwell’s equations, fundamental to electromagnetism, upon which the whole of electronics and telecommuncations are built. Maxwell’s fourth equation makes use of displacement current, which even people who have worked in the field (‘scuse the pun) for years find pretty counter-intuitive. As you delve deeper into Quantum physics, there are many phenomena (tunnelling – used in the transistors on the chips inside the computer you are reading this on) which are also very counter-intuitive.

    In both these cases (and there are many others) Occam’s razor is actually holding us back. It takes a genius like Maxwell to realise that the more complex solution may be the right one. The rest of us plodders, using Edison’s 90% persiration, 10% inspiration find Occam a convenient crutch.

    Sometimes Occam’s razor just doesn’t cut the mustard.

  2. I came to your blog by googling “occam’s razor linguistics” because I keep hearing reference to it in linguistics talks and when discussing linguistics with people – they often invoke Occams’s Razor in an ironic tone of voice sort of indicating an ambiguity in linguistics about this principle.

    Actually, as it turns out, I am having the same discussion as you bring up here these days with colleagues at a summer school in Leiden, Netherlands. In our discussions the issue applies mostly to the subfield of phonology where there seems to be a widespread consensus, and students of linguistics are told, that even if sounds, that is consonants, vowels and tone, pattern asymmetrically in a given language, these are surface phenomena which must be explained with reference to symmetrical underlying structures in accordance with the most economical rules – and it is generally accepted that the simpler the explanation the more likely is it to be true.

    Like you, I ask: why? says who? This is not physics we are talking about here but human language – Jeff Mielke in his “The Emergence of Distinctive Features” quotes Bybee in the following way:

    “… [P]honological and morpho- syntactic regularities are emergent. This means that such patterns are not basic but a secondary result of aspects of speaking and thinking: they are not necessarily categorical, symmetrical or economical, but vary according to the nature of the substance involved, and the demands of communication.” – Bybee 1998: 215, ‘Usage-based Phonology’

    I believe that Bybee is right in claiming that the current state of affairs in any language is simply the result of historical developments due to factors such as pragmatics and cognition and the real task in describing languages, such as in grammar writing which is my occupation these days as a PhD student, is to arrive at those historical changes by internal or external reconstruction.

Leave a Reply

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

WordPress.com Logo

You are commenting using your WordPress.com 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 )

Google+ photo

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

Connecting to %s