APPENDIX A

Flipping Categories

A few minutes after his talk, de Troy is in the cafeteria, sipping an espresso. The RSPEL member comes in from the conference room and takes a seat next to him.

RSPEL MEMBER: Dr de Troy, a word, if I may … Look, it’s not that I think your presentation wasn’t persuasive. And I’m rather loath to admit it, but I see now that destruction is somehow behind the emergence of even very elaborate structures in language. Still, I have to say that I feel rather cheated, since you never really answered my question about what it is that transforms something from a noun to a preposition, or from a verb to an auxiliary. You talked very generally about ‘content words’ and ‘grammatical words’, and you argued that the change from one camp to another was a gradual process, because it was the outcome of gradual erosion in meaning. But you see, what I really wanted to know was not so much the change from ‘content’ to ‘grammar’, but the actual transformation between syntactic categories. Surely, the switch from verb to auxiliary or from noun to preposition can’t just be a matter of gradual changes in meaning. After all, it’s not as if a word can be a noun and a preposition at the same time, is it? So there must have been something that actually transformed ‘back’ from a noun to a preposition, and there must also have been something that changed ‘go’ from a verb to an auxiliary. And what I would really like to know is when exactly these metamorphoses from one category to another took place, and what exactly sparked them. So please don’t just fob me off with stories about gentle changes in meaning again …

DE TROY: Well, I’m afraid that these mysterious metamorphoses are much less momentous than you might imagine … But you know what, instead of just theorizing about it all, why don’t I throw the question back at you? If you expect me to explain how something flipped from one syntactic category to another, isn’t it fair enough to ask first what the difference between these two categories is supposed to be?

RSPEL MEMBER: But that’s completely obvious, isn’t it? Anyone who’s had an elementary education knows full well that ‘go’ is a verb, ‘will’ is an auxiliary, ‘back’ is a noun, ‘under’ is a preposition.

DE TROY: But I wasn’t really asking about your elementary education. What if I were to claim that the flip from one category to the other only seems so mysterious because your schooling has inculcated these categories into you as absolute God-given entities? I said earlier that in actual language, words don’t walk around with designer T-shirts labelled ‘content’ or ‘grammatical element’. But you see, the same applies to the labels ‘noun’ and ‘verb’. So why don’t we just forget for a moment that ‘go’ is a verb simply because that’s what we’ve always been told, and try to agree on why we give it that label?

RSPEL MEMBER: Well, surely you hardly need reminding that verbs are words that refer to actions, nouns are words that refer to things, prepositions refer to relations, and auxiliaries mark tense and the like …

DE TROY: Hmm … But if this is all so simple, then don’t you think your accusation that my theory fell short in some way was rather uncalled for? Look, first you ask me not to feed you any more ‘stories about gentle changes in meaning’, and demand to know what the magic ingredient was that transformed a verb into an auxiliary. But then, when I ask you what the difference between a verb and an auxiliary actually is, what do you tell me? A nice story about … meaning! According to your explanation, once ‘going to’ was no longer used to express the action of movement, and came to mark tense, it should automatically be relabelled an auxiliary. And once ‘back of’ came to express the relation ‘behind’, it should automatically be termed a preposition.

RSPEL MEMBER: But you know as well as I do that there is more to it than that. I didn’t want to imply that meaning is the only difference between categories, and I realize of course that the match between meaning and syntactic categories isn’t perfect: ‘movement’ is a noun, not a verb, although it refers to an action. ‘Future’ is also a noun, not an auxiliary, although it refers to time. So besides meaning, there are also structural properties that set the categories apart.

DE TROY: And what exactly are these ‘structural properties’?

RSPEL MEMBER: Well, for one thing, auxiliaries appear before infinitive verbs: ‘I will wash’, ‘they could see’ – whereas normal verbs don’t appear before infinitives: you don’t say ‘I wash see’, or ‘I bring see’. On the other hand, verbs can appear before an object, as in ‘I see a cow’, but auxiliaries can never take an object. You can’t say ‘I could a cow’.

DE TROY: Great. So what you are saying, really, is that a syntactic category is a group of words which appear in similar positions in the sentence. And if I may generalize from all of this, the implication is that the reliable method for defining a syntactic category is not by searching for a common meaning, but for a common distribution: the particular slots in which a group of words appears. It’s true that things like ‘tree’ and ‘bucket’ are always nouns, but as you rightly point out, there are also nouns like ‘movement’ or ‘future’, which are certainly not physical things. The reason why we call them nouns is that they appear in the same slots as other nouns. For example, nouns typically appear in the slot X in ‘a great X’ – you can replace X here with ‘bucket’ or ‘tree’, but also with ‘movement’ and ‘future’. Auxiliaries typically appear in different slots, for instance ‘he Y see’ – you can replace Y here with ‘will’, ‘should’, ‘must’, ‘can’, and so on. And verbs appear in slots like ‘he Z-s the bucket’ – you can replace Z with ‘see’ or ‘kick’. And in general, a syntactic category is a group of words which we perceive to be similar, because they have a similar distribution – they appear in the same slots in the sentence.

RSPEL MEMBER: There is nothing to disagree with here. But all this still doesn’t answer my original question about how something switched from one category to another.

DE TROY: No, it doesn’t tell us that quite yet, but it does allow us to formulate the question in a more sensible way – which is more than half-way towards answering it …

RSPEL MEMBER: Fine then, so I should rather ask how it was that a verb like ‘go’ suddenly started appearing not in slots that are typical of verbs, but in slots characteristic of auxiliaries.

DE TROY: But don’t you see, now the whole question appears in a very different light. The point is that the verb ‘go’ as such never started appearing in slots characteristic of auxiliaries. What changed into an auxiliary was not ‘go’, but only a particular phrase, which had appeared in auxiliary slots all along.

RSPEL MEMBER: What? Now you are saying that ‘go’ never changed into an auxiliary?

DE TROY: No, I’m simply formulating things more carefully. Look, you’d agree, wouldn’t you, that it’s not any old ‘go’ that turned into an auxiliary. In a phrase like ‘I’m going to the cinema’, for instance, ‘going’ is an entirely normal verb.

RSPEL MEMBER: Of course it is, since you can replace it with other verbs: ‘I’m driving to the cinema’, and so on. But what I meant was ‘go’ in the construction ‘be going to do something’.

DE TROY: But that’s exactly it. What turned into an auxiliary was not just any ‘go’. It was the phrase ‘is going to’ in one very specific construction – when it appeared before an infinitive verb. And just think about it this way: forget for a moment that ‘is going to’ has three different components, and take it as one unit X = ‘is-going-to’. If you look at it this way, you see that ‘is-going-to’ appeared in the same slots as auxiliaries all along. Just like will or must, it fits into slots like ‘he X do something’.

RSPEL MEMBER: So are you trying to say that nothing happened at all? We started off with something that appeared in auxiliary slots, and we ended up with something that appeared in auxiliary slots …

DE TROY: But what has changed is that the internal structure of the phrase has collapsed. ‘Going to’ began life as a combination of different elements, a shorthand for ‘going (somewhere, in order) to’ do something. This phrase contained two independent parts: the verb ‘go’ denoted movement, and the preposition ‘to’ marked the purpose of this movement. Speakers didn’t perceive the similarity of this phrase to auxiliaries, because they didn’t think of it as a unit. If you want to get a feel of how ‘going to’ must have sounded originally, then think of a phrase like ‘working to’ in the sentence ‘he is working to earn money’. In theory, one could say that ‘is working to’, as one unit, appears in the auxiliary slot X: ‘he X earn money’ – after all, you can replace this X with will, must, and so on.

RSPEL MEMBER: So why don’t we analyse ‘is working to’ as an auxiliary?

DE TROY: Because we don’t perceive these words as one unit. They still feel like a combination of different components: a verb ‘working’, and a preposition ‘to’, denoting the purpose of the work. So while it’s true that ‘is working to’ would fit into auxiliary slots if it were a unit, this fact remains pretty irrelevant as long as the phrase does not really feel like one unit. Now, I suppose that for speakers in the fifteenth century, the structure of the phrase ‘is going to’ felt just like that of ‘is working to’ today. But with time, ‘is going to’ lost its independent meaning and came to denote the future, so in the perception of speakers, the individual components lost their distinctive roles. The whole ‘is going to’ thus came to be perceived as one unit, and was reduced to just ‘s gonna. And in this new role as a unit, the similarity between ‘s gonna and auxiliaries became apparent: here was a phrase that appeared in auxiliary slots and had a similar meaning to other auxiliaries: marking tense.
    Actually, in some varieties of English, especially in the States, the ’s of ’s gonna has already been ditched, and many speakers simply say things like ‘he gonna come’. I know you might view this change as the worst of vulgarities, but it’s really very natural and even logical, because the ’s no longer has any obvious function. Originally, the ‘is’ was there to introduce the -ing form on the verb ‘going’. But since not much is left of that original verb, the ’s nowadays just feels like excess luggage, and so speakers simply ditch it. And in the varieties of English where this change has already happened, gonna now appears on its own in exactly the same slot as auxiliaries like will: ‘he X come’. But you see, all this happened without any magic leap from one slot to another. What really made gonna similar to auxiliaries was nothing other than gradual erosion in meaning and sounds.

RSPEL MEMBER: And what about the changes between other syntactic categories, for instance when a noun like ‘back’ turns into a preposition meaning ‘behind’?

DE TROY: It’s the same story all over again. It wasn’t just the noun ‘back’ that turned into a preposition, but a particular phrase, ‘(at the) back of’. And when you think about this phrase as one unit, you see that it had been in preposition slots all along. ‘At-the-back-of’ had always fitted into the slot X in ‘the pool X the house’, which never accommodated nouns like ‘table’ or ‘knife’, but rather prepositions like ‘behind’ or ‘around’. So what brought ‘back of’ into the preposition camp was no gymnastic leap between slots, but just the collapse of its internal structure. Once ‘back of’ started being used metaphorically to denote a spatial relation, the individual components lost their relevance, and the phrase came to be perceived as a unit. In fact, it’s quite likely that ‘back of’ has already began to fuse into something like backa. But just like gonna, backa never had to somersault into its present slot, it had been there all along. What turned ‘at the back of’ into a preposition was the erosion of the original meaning, and resulting collapse of its internal structure.

RSPEL MEMBER: But can’t one at least pinpoint a definite time when the change took place? Can’t one say, for instance, that the point when ‘going to’ turned into an auxiliary was when it became one word, gonna, which appears in auxiliary slots?

DE TROY: Well, it’s not quite as simple as all that, because things can be more gradual than we have so far allowed, even in terms of distribution. I said before that auxiliaries are words that typically appear in characteristic auxiliary slots. This definition rather took it for granted that if a word fits into one characteristic auxiliary slot, it would also fit into all the others.

RSPEL MEMBER: And isn’t that the case?

DE TROY: Well, you can easily check this out with ‘gonna’. I’ve just mentioned that ‘gonna’ fits exactly into the auxiliary slot ‘he X come’ – at least in some varieties of English. But let’s just check for a moment whether ‘gonna’ fits into all other auxiliary slots. Take questions, for instance. When there is an auxiliary in a sentence like ‘he will come’, then you form a question simply by moving this auxiliary to the front, as in ‘will he come?’ In other words, the auxiliaries ‘will’, ‘shall’, ‘must’, and so on fit into the slot X in ‘X he come?’ But just try saying ‘gonna he come?’ – it doesn’t really work, does it? Even speakers who drop the ‘is’ in ‘he gonna come’ would never dream of saying ‘gonna he come?’ Instead, they shove the ‘is’ back in again, and move that to the front: ‘is he gonna come?’ So in questions, ‘gonna’ still behaves more like ‘working to’: the question ‘is he gonna come?’ is formed in the same way as ‘is he working to earn money?’

RSPEL MEMBER: So what exactly are you getting at? Would you now define ‘gonna’ as an auxiliary, or do you think it’s still a verb?

DE TROY: Well, I would simply say that ‘gonna’ is a word which fits into some auxiliary slots, but not into others. In some constructions it behaves like the auxiliaries ‘will’ et al., but in others it still behaves more like a normal verb. Not that ‘gonna’ has anything to be ashamed of, mind you. ‘Gonna’ may not have an identical distribution to the more established auxiliaries, but this doesn’t mean that it performs its function as a future marker any less well than ‘will’ or ‘shall’. It simply means that ‘gonna’ has a somewhat different distribution from the established auxiliaries – for complex historical reasons I don’t want to get into here. ‘Gonna’ just doesn’t fall neatly into either the group of verbs or of established auxiliaries.

RSPEL MEMBER: But doesn’t it cause huge problems if a word like ‘gonna’ is half-way in between, and doesn’t know whether it’s this, that, or the other?

DE TROY: Problems for whom, exactly? It might cause a rumpus in school’s neatly packaged world, or for anyone who believes in perfect platonic forms. But well – most of us have never had the privilege of attending Plato’s academy. You see, we don’t actually speak in ‘verbs’ and ‘auxiliaries’ – we just speak in words. And as far as the word ‘gonna’ is concerned, we simply remember in which particular slots it appears. Why should that be problematic?

RSPEL MEMBER: For one thing, don’t problems arise from the fact that speakers need to memorize individually all the particular slots in which ‘gonna’ appears? I thought that the point about syntactic categories is that they group together words of a feather, so that speakers don’t have to memorize individually all the different slots in which every member of the group can appear. It’s enough, for example, to memorize in which slots one noun like ‘table’ appears, and then you already know that ‘chair’, ‘sausage’ and thousands of other words would appear in just the same slots.

DE TROY: I agree in general, but you see, ‘gonna’ is such a basic and common grammatical marker that remembering the few peculiarities in its distribution is really just a drop in the ocean of details that speakers have to contend with. Don’t you think speakers would react with amazement if you suggested to them that they might have problems with handling ‘gonna’, because it appears in some auxiliary slots but not in others?

RSPEL MEMBER: Well, maybe. But surely, this indecisive state of affairs can’t go on forever? At some stage ‘gonna’ is gonna have to make up its mind and turn into a real auxiliary … Or are you saying that it will just remain stuck where it is indefinitely?

DE TROY: Of course it’s always possible that ‘gonna”s circumstances might change at some stage – it would be rash to rule it out. So it’s quite possible, in theory, that a new generation of speakers might extend ‘gonna’ by analogy on ‘will’ to all auxiliary slots, and start saying things like ‘gonna he come?’ for instance. But while this is all imaginable, frankly, I don’t think it’s a terribly likely scenario. I don’t want to bore you with all the details, but I think that gonna is more likely to stay put where it is right now, ‘stuck’, so to speak, somewhere between verbs and established auxiliaries. It’s certainly not causing much grief to anyone …

RSPEL MEMBER: Gosh, I must say this rather flies in the face of everything I have always thought about syntactic categories. I had always assumed that a word must be one thing or the other, either a verb or an auxiliary, either a noun or a preposition. And now you are telling me that ‘gonna’ can be both, or perhaps neither. But if words don’t always fit neatly into one syntactic category or the other, then why bother with these syntactic categories in the first place? Aren’t these classifications pretty useless?

DE TROY: No, I wouldn’t go as far as that. I would only say that these syntactic categories are not watertight. As the linguist Edward Sapir once put it, ‘all grammars leak’. The main syntactic categories can be very helpful in capturing broad similarities between words. What’s more, a label like ‘verb’ must also reflect some psychological reality: the perception in people’s minds that words like ‘kick’, ‘bite’, and a great many others, behave in a very similar way and appear in similar slots. So syntactic categories can be very helpful, especially when you take a bird’s-eye view of language. But when you focus on the details, you often find that words don’t always fall conveniently under one of the main labels. A word can start acquiring the distribution of another category only gradually, and sometimes it can even remain stuck between categories. When one tries to describe a language, this should not pose serious problems, as long as one remembers that syntactic categories are only meant to be descriptive labels – they are supposed to serve us, not rule us. So when you ask a question such as ‘when did a word move from category A to category B?’, you should remember that the word never had to perform any complicated acrobatics. What you are really asking is: ‘when do I decide to stop using the label A for a word, and start using the label B?’ So if you discover that a word like ‘gonna’ won’t oblige, and won’t fit neatly under either of your labels, then you should remember that what’s problematic is not the word itself, but your labels. This doesn’t make your labels completely useless, it just means that they are not perfect in catching every aspect of language.

RSPEL MEMBER: Well, I will certainly have to ponder all of this. But … I’m terribly sorry, I have to dash off now – I really don’t want to miss the talk on ‘between you and I’…