Lots of my current thoughts are emerging out of my studies in Gaelic this semester. Today I want to talk a little about how we conceptualise a ‘word’. And why it’s so problematic a concept but ineluctably so.

I think that a lot of us, when we are children in our teens, perhaps earlier, go through this little literacy revolution, where we begin to conceptualise ‘language’ in terms of the written word. Our idea of what is really the language becomes defined by orthography and the word on the page. You see this a bit in how pedantic people, children, become about how their name is spelled. It must be spelled exactly right, because that’s my name. You see it too when people start pronouncing words wrong, because that’s how it’s written, and the mis-alignment between written language and spoken language becomes really a source of annoyance. It’s mostly a phase, and most of us grow out of it, but the shift still occurs. Our platonic ideal of ‘the language’ becomes tightly bound to the written word.

One of the great things about English orthography, by the way, is that it’s not the IPA. Every now and again someone thinks we should have spelling reform in English, and it’s a terrible idea. See, English is pronounced all sorts of ways all over the world, and so the idea of a word, tied up in how it’s spelled, being pronounced differently, enables a beautiful commonality of communication despite the differences in the way we say things. This, too, is why keeping an orthography stable across time means that you and I can read things from 500 years ago and understand them, even though the way English was pronounced then would be fairly unrecognisable to us.

So, what we tend to end up with, for English, is a concept of the word that reifies its spelling, and accepts a certain range of pronunciation. That range has limits though. Let me give an example at the edge: Chinese logographic writing, where the pronunciation of a ‘word’ can be wildly different between different languages, say Cantonese, Mandarin, and Japanese, despite them using the same written ‘word’ to represent it. We would struggle to say that those are the same ‘word’ even though they have the same character.

Now, what happens when orthography gets a bit more loosey-goosey? Ah, welcome to Scottish Gaelic. Firstly, modern Scottish Gaelic as a language of writing doesn’t really go back that far, and really only develops strongly from the 18th century onwards. Early texts often have divergent ideas about how to represent sounds. Yes, there is a literary tradition stretching back into Old Irish, but that is not entirely determinative. Gaelic did undergo a spelling reform in 1981, revised again in 2005 and 2009, the “Gaelic Orthographic Convention” (GOC). Gaelic orthography has a number of interesting features, some that pull in different directions. On the one hand, one written form can represent a number of pronunciations, covering the dialect spectrum. That is similar to English that I spoke of above. In Gaelic it can be a little more extreme. I sometimes hear speakers pronounce words and have no idea what they’re saying until I already know what ‘word’ they’re pronouncing.

On the other hand, when you read pre-reform texts, you see that there is this wide range of how people realised what they were saying, which in a sense makes the spoken word the primary form, in a reversal of the idea of the written word above. That there’s a single spoken ideal of a ‘word’ – the word as said, and then a pluriform range of ways of writing that. That puts our ideas about what a word ‘is’ upside-down.

Which delivers us to the tension: if one written word can be pronounced in multiple ways and still be the same ‘word’, and one spoken word can be pronounced one way but written in multiple ways and still be the same ‘word’, and yet both can also be true, and one word can be pronounced multiple ways and then also written multiple ways, is it still one word? Or is it many words?

I’ll try to give a Greek example for a moment. Are γίγνεται, γίνεται and γείνεται all the same word? (Yes, I think they are). This is just spelling variation. But it’s spelling variation that also represents variation in pronunciation. If you’re an Atticist, you think γίγνεται is the archetypal form, and the others are aberrations. And if you conceive of the spoken word as primary, then what you have is not spelling variation per se, but variation in realisation of a single word. But that’s precisely where the boundaries get blurry.

Hopefully I’ve made your head hurt just a little. To make it hurt more, it helps to know that there is no real consensus among linguists about how to define a word anyway. It’s a tricky concept and across languages it gets complicated enough that no definition works for all languages.