Postpositions?

[Richard W]

EDIT, on topic this time: regarding the argument about "baking," it might be useful to point out that these are homonyms. English conflated two separate endings a thousand years ago, hence the identical pronunciation of the gerund and participle. But we could, if we really wanted to, treat them as two coincidentally identical forms. We've all made peace with the fact that English subjunctive verbs are usually identical to nominal derivations of those verbs ("We ask that he leave" vs "He is on leave"), without scratching our heads about how we reconcile these disparate uses. If we even insist on deciding whether "baking" and "baking" are one word or two in the first place, that is. Personally I'm not sure that question will help given the number of assumptions we'd hjave to agree on before we could get there.

I understood that the progressive tenses actually use the gerund, whence the occasional variant we were a-baking, where a is a doublet of the preposition on.

[zompist]

So do we have two word classes, one Yidong and one Shidong, to use the emic terminology? Or do we just have nouns with asterisks all over them?

I like that phrase. I think Brad's methods applied to Old Chinese would indeed produce a grammar with asterisks all over it. It's a language made of asterisks.

EDIT, on topic this time: regarding the argument about "baking," it might be useful to point out that these are homonyms. English conflated two separate endings a thousand years ago, hence the identical pronunciation of the gerund and participle. But we could, if we really wanted to, treat them as two coincidentally identical forms. [...] Personally I'm not sure that question will help given the number of assumptions we'd hjave to agree on before we could get there.

Yeah, I dunno either. I tend not to like dividing up words based on historical morphology, but nominal vs verbal "baking" is probably defensible since the syntax reinforces it.

FWIW a few noun forms are lexicalized in their plural forms: makings, takings. I don't think these particular senses even have a singular form, even though making/taking exist.

[Moose-tache]

I think as conlangers, we feel deeply unsettled by a language that is just a dictionary, i.e. a language in which each word carries its own explanation of how it interacts with other words. But in a way that is how languages operate. We make generalizations, but not consistently.

In my dialect, people have started saying things like "I don't know where it is, even," because "even" feels similar to other words that can freely appear at the end of a sentence. Meanwhile there is a precedent for words that modify clauses that cannot move in this way, hence I haven't yet heard sentences like *"I want to go home, just." What is the distinction between "even" and "just" that people are making here?

Can we bear to live in a world where the only difference is that they have been given different asterisks by English speakers who don't care how salient their generalizations are?

[bradrn]

All I can say is that I’ve yet to see a language where the best analysis uses fuzzy word classes, along the lines of ‘this is a list of words which grade between such-and-such prototypical word classes, and their behaviour can’t be discretised sensibly’. Word classes with concrete boundaries seem like a pretty good description for most languages.

OK, take a look at Classical Chinese. There's no morphology[*[, so you're entirely reliant on syntax. And individual words are really resistant to neat labeling. It's not lawless, of course. But it would be a severe test of your claim that word classes can be easily defined and have no exceptions or prototype effects.

Generally speaking, I get suspicious when people talk about languages where words have no word classes because ‘they can’t be labelled’. Often what turns out to be happening is that people have a picture of how English nouns and verbs behave, look at a language where they behave differently, and then conclude that language has no nouns or verbs. Of course, what’s really happening is that it clearly has word classes, you just have to define them quite differently to the English word classes.

But to double-check (because I can’t just wildly speculate!), I had a look at Pulleyblank’s Outline of Chinese Grammar (1995). And he seems to have no problem whatsoever with postulating discrete word classes — on page 12, even! He distinguishes as separate word classes nouns, verbs and adjectives (as a subclass of verbs). There is no mention of the idea that word classes may be inapplicable to Classical Chinese.

[I wrote this before seeing Moose-tache’s post. See below for more reasoning around this topic.]

Anyway, we seem to agree that word classes in typology exhibit prototype effects. But you want to deny them when looking at a particular language. I find that a bit odd— it's like saying that evolution is true for mammals, but not for otters. The overall theory should be true even for smaller instances or time depths, and can even point us to specific insights. It just won't be as spectacular.

I disagree. There are numerous insights in all areas of grammar which are only apparent when you compare multiple languages, or investigate diachronics. Few of them could reasonably have been deduced from internal evidence from a specific language, or even a small set of languages.

For a non-word class example, morphosyntactic alignments obviously form a continuum from accusative to split ergative to ergative. But this only becomes obvious when comparing alignment systems across lots of languages. Within one language, we can only see one individual alignment system — something which is best described without prototypes or gradation.

So it is with word classes:

And we do use typology. We're using it when we call something in a language a verb, or an aspect, or a dative. There's nothing wrong with saying that Latin adjectives are very nouny. (Well, you'd dress it up in more formal language.)

In this case, we are talking about how Latin adjectives compare to those of other languages. Looking cross-linguistically, we can identify certain constructions where nouns are most prominent, and certain other constructions where verbs are most prominent. From this we can, indeed, say that Latin adjectives share many important properties with prototypical nouns.

However, this information is not available to the syntacticist analysing Latin on its own terms. The terms ‘nouny adjective’ and ‘verby adjective’ didn’t even come into heavy use until the 1990s, when this sort of data became available! (Or perhaps a bit earlier: I’m going by the publication dates of Stassen and Wetzer’s well-known books on the subject.) Without that perspective, all we can say is that Latin adjectives share such-and-such properties with nouns and such-and-such properties with verbs, without being able to say anything about certain words being ‘more nouny’ or ‘more verby’. At most, we can say that adjectives are more similar to nouns than they are to verbs — which I accept! But a theory involving gradation, or prototypes, is not necessary to do this.

There are some good reasons to avoid history and typology at some stages of analysis. But we can look at those reasons and address them directly. The main one is that both can lead us to make dubious assumptions— e.g. "possessive s used to be a genitive case marking on nouns, so it still is", or "Because there's a Latin infinitive amare, the translation 'to love' is the English infinitive". But the lesson to draw is not "never look at history or typology", but not to make those types of declarations.

Exactly! This is why I am emphasising the fact that word classes are identified early in the analysis. After that, typological and historical concepts can certainly be used.

And in some languages it's more helpful to talk about verbal vs. nominal pathways rather than verbs and nouns.

I don’t understand what you mean by this. Could you elaborate please?

We talked about this before, e.g. the discussion of Quechua kawsay. Basically: in some languages, at least some roots can be inflected as nouns, or verbs, or something else.

You suggested that they be marked {N, V}. And maybe that's OK! But another possibility, especially if this is really common in a language, is that roots aren't specified for word class in that language. We'd have to see what makes for a better grammar. But I would suggest if (say) 75% of your lexicon is marked {N, V}, you'd be missing a generalization and improperly applying methods only suited to other languages.

Why? Are you implying that languages have to have the majority of words be assigned a single word class? If anything, this in itself seems an improper generalisation from more ‘familiar’ European languages!

This is why I insisted earlier on strictly pre-theoretic tests, along the lines of ‘this word can appear after the’ or ‘this word can take -ing’. This allows us to define word classes without reference to any sort of syntactic analysis. Then, those word classes may be used as the primitives on which we base further analysis of syntax. [...] My reasoning with regards to ‘baking’ is exactly parallel: I assume it is its own word class, unless there is a better description. In this case, there isn’t any — its distribution can’t easily be described as a union of that of any other set of English word classes. So, for lack of evidence to the contrary, I assume that it makes up its own, separate word class.

On "baking" in particular, you haven't actually given a reason to analyze it as one word; often people divide it into a more nouny word (gerund) and a more verby one (participle).

Let’s flip this claim around. Is there any particular evidence, within English itself, to conclude that the ‘-ing’ form is actually two separate word types? Apart, that is, from ‘all languages are exactly like Latin’, which was the initial justification for this particular claim. Remember, the null hypothesis is that each word has its own, single, word class; evidence is required for any other conclusion.

Of course, as both you and Moose-tache comment, the diachronics can lead you to a very different conclusion. But, though it can certainly be revealing in later analysis, I am not working diachronically. I am working purely synchronically, emically and pre-theoretically. (Those three adjectives are important; my whole analysis is based on them.)

I’ll also mention that terms like ‘gerund’ and ‘participle’ are tricky, because they’re really functional terms. A gerund is some verb form which is used as a formal noun, and a participle is some verb form which is used to formally modify a noun. That is, they describe the use of a particular verb form, not the form itself. (And in fact, saying ‘-ing is a gerund and a participle’ misses the fact that it’s also used as an infinitive and a converb; see Nedjalkov 1995 and Ross 2016.)

I understand what you're trying to do; my points all along have been a) it's not going to be as easy and ironclad as you think, and b) you're over-emphasizing the morphology.

Let's step back and pretend that we could analyze English from scratch, keeping the bins of morphology, syntax, and semantics hermetically separate. What do we get?

1. Morphology gives us a set of word classes— call them noun_m, verb_m, etc. It also leaves us a huge bin of uninflectable words.
2. Syntax gives us a much more complicated picture, with a much smaller trash bin (probably called "particles"). Part of this picture is the gradation I described; we can call the poles noun_syn and verb_syn.
3. Semantics gives us fuzzy clouds of things, but it does offer us a prototype agent— a body, probably human, which persists in time and does stuff— and a prototype action— a definite event in time which changes the objects around it, probably directed by an agent. We can call these noun_sem and verb_sem.

Now what is the relationship between these entities? Is noun_m = noun_syn = noun_sem? Or are they completely unrelated? Or related in a complex way?

…

What I was trying to say earlier is that noun_m / noun_syn / noun_sem are not the same thing, indeed can be frustratingly distinct— but that they are correlated in very interesting ways, which are best explained by prototype theory.

For a start, I am not insisting that morphology, syntax and semantics be kept ‘hermetically separate’. In fact, I have to admit that I very much doubt that a reasonable distinction can be drawn between morphology and syntax in general. Semantics is more distinct, but in numerous areas of morphosyntax it is absolutely required to do any sort of revealing analysis.

When analysing word classes, however, it is not so much that semantics is ignored, but that it simply is completely irrelevant. Indeed, an analysis of word classed must be done while deliberately ignoring semantics. And this is because different languages have a horrible habit of placing words with exactly the same meaning in completely different classes. (A good example: English hungry is a noun, Latin ēsurio is a verb, Dyirbal ŋamir is an adjective (Dixon 2010).) This is precisely what makes the semantics of word classes so hard to figure out! The best we can do is to see if there are any cross-linguistic tendencies around which meanings are most ‘nouny’ and which most ‘verby’ — your noun_sem and verb_sem. Since such definitions are only available after carefully analysing word classes for numerous of languages, and give only generalisations which may or may not apply to any one language, I prefer to ignore them for (pre-theoretic, emic) word class analyses. And, since they’re fuzzily semantic concepts, I’m happy to apply prototype theory to them.

As for morphological and syntactic word classes, I already talked about them earlier. Generally speaking, they coincide to such an extent that the distinction is not needed; it is only in rare cases (e.g. Kayardild) where the distinction becomes important.

E.g., the "most nouny" frames in my graded list of examples happen to be precisely the situations where nouns are most inflected. The most verby frames are those where the verbs are most inflected. I don't think that's coincidence. … English modals— which are not prototypical actions— lack number inflection. In many languages active verbs (closer to the prototypical action) have more aspect distinctions than stative verbs.

This is an interesting observation. I’m not quite sure what to make of it just right now; I’ll have to think more on this. In support of this, I’ll also note that subordinate forms have a marked tendency to be more deranked than others.

Also, the most semantically agentive situations correlate with noun inflection. E.g. in IE, m/f nouns but not neuters have a separate accusative. In Russian, there are subcases for animate nouns. In English, we still have 3s objective pronouns for animates but not inanimates. English mass nouns— less likely to be agents— lack a plural.

I note that all your examples here are European IE. I wouldn’t necessarily expect any of these to hold for ergative languages. For one thing, the lack of marking on neuters is just because accusative alignment does tend to be on animate nouns; for an ergative language it would be animate nouns which are unmarked. (I’d have to look through my references to give better examples, but off the top of my head, IIRC there’s one Australian language which has two ergative cases covering the most inanimate nouns. This is the opposite of the Russian situation, I think, though I don’t know Russian.)

I'm not quite sure what your answer would be, but it seems like you're expecting noun_m to be more basic (you called it a "primitive") and to drive the analysis of syntax. That's where I disagree.

…you do read my posts, right? Because I explicitly gave a counterargument to this particular claim:

I think you were misled by the fact that, for an example of a word class test, I happened to choose a morphological property. That will not always be the case. The sort of definitions I’m looking for are exemplified by Dixon’s definitions of English word classes (Basic Linguistic Theory vol. 1):

[verbs], takes suffix -ing
[nouns], may be immediately preceded by an article and need not be followed by another word
[adjectives], may be immediately preceded by an article and is then followed by another word (either one from class [nouns] or another word from class [adjectives])

One might quibble with the exact definitions, but that’s the style of thing I’m looking for — simple tests to separate words into different categories, each of which is homogeneous in its syntactic and morphological behaviour.

By ‘primitive’, what I actually meant was that the concept of ‘word class’ is primitive; as in, further analysis, both syntactic and morphological, tends to treat it as a given, inviolate, opaque abstraction. This is, yet again, why I insist on a purely emic, pre-theoretical analysis. The analysis might involve morphology; but in fact it is actually more probable that it will involve syntax. All that matters is that it is a simple test.

OK, take a look at Classical Chinese. There's no morphology[*[, so you're entirely reliant on syntax. And individual words are really resistant to neat labeling. It's not lawless, of course. But it would be a severe test of your claim that word classes can be easily defined and have no exceptions or prototype effects.

To give some illustration of what Zompist is saying, words in Old/Classical Chinese, much like English, had a lot of lexically determined notes about how they could be used and what connotations they acquired when used this way or that way. If you treated this as membership in a series of word classes, you'd need a lot of classes.

For example, 利 (profit) and 王 (king) were both nouns. As in English, they could be enverbled with impunity. But 利 as a verb meant "to consider something profitable," while 王 as a verb means "to be crowned." So do we have two word classes, one Yidong and one Shidong, to use the emic terminology? Or do we just have nouns with asterisks all over them?

I’d say that both your suggested analyses are unhelpful. Both 利 and 王 are just noun+verbs — that is, the same word class.

Let me clarify. ‘Word class’, as I use the term, is a purely morphosyntactic phenomenon. As I said above, semantics should not enter into the picture at all, and word class should be determined purely by morphosyntactic behaviour. Certainly, I don’t deny that the semantic relationship between the nominal and verbal forms is very interesting and deserving of study! It just isn’t relevant when defining word classes.

If you are inclined to argue, I ask: do you accept the traditional analysis that English ‘picture’ and ‘glow’ are the same word class, as they are both noun and verb? If so, then consider the fact that this is precisely parallel to the Classical Chinese example: as verbs, ‘picture’ means ‘to consider something as a picture’ (I pictured the house I grew up in), whereas ‘glow’ means ‘to have a glow form around’ (The mixture glowed as I added luciferase). Yet I don’t recall anyone separating these out as separate word classes. (Even Dixon, who is semantics-mad and as big a splitter as you would ever hope to see, hasn’t gone this far!)

[zompist]

But to double-check (because I can’t just wildly speculate!), I had a look at Pulleyblank’s Outline of Chinese Grammar (1995). And he seems to have no problem whatsoever with postulating discrete word classes — on page 12, even! He distinguishes as separate word classes nouns, verbs and adjectives (as a subclass of verbs). There is no mention of the idea that word classes may be inapplicable to Classical Chinese.

Oh come on. What do you expect on page 12 of a book? You have to start somewhere, and it's fine to say that "kill" is a verb and "horse" is a noun-- notice that he's using prototypes as examples. Notice also that he doesn't actually give a usable test, especially since he immediately mentions that verbs can be used as nouns and vice versa!

Let's look at just one example:

道可道，非恆道；
名可名，非恆名。

The first halves are usually translated something like "The Way that can be explained" and "the Name that can be named". But as you can see, the OC is quite simple, using 道 dao and 名 ming as both noun and verb. By a syntactic test, then, they are both. A literal translation would be "Way can way... Name can name..." You've said we can't use semantic tests, which is good, because it's hardly clear that "way, name" are things like horses.

You suggested that they be marked {N, V}. And maybe that's OK! But another possibility, especially if this is really common in a language, is that roots aren't specified for word class in that language. We'd have to see what makes for a better grammar. But I would suggest if (say) 75% of your lexicon is marked {N, V}, you'd be missing a generalization and improperly applying methods only suited to other languages.

Why? Are you implying that languages have to have the majority of words be assigned a single word class?

I didn't say anything of the sort.

A lexicon should avoid supplying information that can be given by a general rule. To use a silly example, it's not useful to say, for every noun in the language, "Can be used as the subject of a sentence." Of course it can. Slightly less silly, my dictionary occasionally includes definitions like "2. something resembling a (main word)." That's just as unnecessary: the word for any physical object can be used for things resembling it, models of it, pictures of it.

If hypothetically "can be used as a noun or a verb" is true of most words in the lexicon, then it's not a fact about those words at all, it's a fact about the language. It'd be more useful and more interesting to label the exceptions.

Let’s flip this claim around. Is there any particular evidence, within English itself, to conclude that the ‘-ing’ form is actually two separate word types?

Sure, because there are two or more syntactic frames involved. You keep agreeing that syntactic tests are good, but keep ignoring what they say.

I am working purely synchronically, emically and pre-theoretically. (Those three adjectives are important; my whole analysis is based on them.)

"Pre-theoretical" is propaganda. Of course you have a theory-- that's what you're trying to enunciate. Things like "you can't use semantics to define word classes" are statements of theory. I don't mind you having a theory; but it's a bit annoying to pretend that you don't.

I have no idea what you mean by "emic" here. If you're making an analogy from phonology-- you can't do phonology without doing phonetics. If you're using the anthropological meaning, that'd mean that you want to use speakers' own terms and theories, which we already know can be quite erroneous.

I’ll also mention that terms like ‘gerund’ and ‘participle’ are tricky, because they’re really functional terms. A gerund is some verb form which is used as a formal noun, and a participle is some verb form which is used to formally modify a noun. That is, they describe the use of a particular verb form, not the form itself.

That is, they describe syntax, which you keep saying is acceptable for defining word classes, and indeed applaud in Pulleyblank. Which is it?

For a start, I am not insisting that morphology, syntax and semantics be kept ‘hermetically separate’. In fact, I have to admit that I very much doubt that a reasonable distinction can be drawn between morphology and syntax in general. Semantics is more distinct, but in numerous areas of morphosyntax it is absolutely required to do any sort of revealing analysis.

Since I've been rather critical today, let me note that I agree entirely with the above.

A good example: English hungry is a noun, Latin ēsurio is a verb, Dyirbal ŋamir is an adjective (Dixon 2010).)

Not really a good example: "hungry" is an adjective, and Latin also has an adjective, esuriens. But this is not a point of contention anyway; no one is saying we should find words for "hungry" and call them all adjectives.

One might quibble with the exact definitions, but that’s the style of thing I’m looking for — simple tests to separate words into different categories, each of which is homogeneous in its syntactic and morphological behaviour.

Yes, I know, and I've given a bunch of reasons why your tests will either be not so simple, or your categories not homogenous.

It's fine to hope that your tests will be simple. But so far, you seem to keep looking at English morphology (verb_m and noun_m), pointing out how simple they are, and just declaring that this will be the case when you've also taken syntax into account. When you actually provide syntactic tests, they've been simplistic.

Again, you have to start somewhere, and if you want to start with morphology, fine. But I've already shown you that verb_syn and noun_syn are far more complicated-- they are obviously related to the morphological classes, but not identical.

I think you're not realizing the distinction between an informal analysis, and a fuller treatment of syntax. Yes, you can create rough-and-ready categories or rules that will be useful and can even be used in an introductory textbook. This is not the same as researching and defining syntactic categories that will cover the wide range of syntactic behaviors people have found in the last 60 years. To bring us back to the original subject, it's pretty easy to define prepositions as things that appear before an NP, and create a PP that acts as a modifier to another NP or to a VP. But remember the 90%-90% rule. That definition doesn't always help when you're trying to pick apart the trickier (but still everyday) parts of English syntax-- the stuff that isn't just P + NP.

[Moose-tache]

Yes, I know, and I've given a bunch of reasons why your tests will either be not so simple, or your categories not homogenous.

It's fine to hope that your tests will be simple. But so far, you seem to keep looking at English morphology (verb_m and noun_m), pointing out how simple they are, and just declaring that this will be the case when you've also taken syntax into account. When you actually provide syntactic tests, they've been simplistic.

Again, you have to start somewhere, and if you want to start with morphology, fine. But I've already shown you that verb_syn and noun_syn are far more complicated-- they are obviously related to the morphological classes, but not identical.

I think you're not realizing the distinction between an informal analysis, and a fuller treatment of syntax. Yes, you can create rough-and-ready categories or rules that will be useful and can even be used in an introductory textbook. This is not the same as researching and defining syntactic categories that will cover the wide range of syntactic behaviors people have found in the last 60 years. To bring us back to the original subject, it's pretty easy to define prepositions as things that appear before an NP, and create a PP that acts as a modifier to another NP or to a VP. But remember the 90%-90% rule. That definition doesn't always help when you're trying to pick apart the trickier (but still everyday) parts of English syntax-- the stuff that isn't just P + NP.

I think I've realized why you're not getting through to our friend. Elsewhere in the same post, he met my example of two OC words with different usage, and then repeated the same example in English.

as verbs, ‘picture’ means ‘to consider something as a picture’ (I pictured the house I grew up in), whereas ‘glow’ means ‘to have a glow form around’ (The mixture glowed as I added luciferase).

At first I thought "how is it a counter-argument to just agree with what I said?" Then it hit me. Bradrn is presenting this as a reason in favor of treating picture and glow as the same word class. In other words, we're out here talking about how to complete the task of syntactic word classification, and this whole time he's been advocating in favor of leaving the job half done. According to Bradrn, it is appropriate to say that picture and glow are the same.

So none of us really disagrees, in a way. The two of us are trying to explain how to fly from New York to Chicago, and Bradrn is patiently trying to tell us that you don't need an airplane to get to New Jersey.

[bradrn]

I feel like we’re getting somewhere, finally! That is, I still don’t agree, but I feel we’re understanding each others’ arguments a bit better.

But to double-check (because I can’t just wildly speculate!), I had a look at Pulleyblank’s Outline of Chinese Grammar (1995). And he seems to have no problem whatsoever with postulating discrete word classes — on page 12, even! He distinguishes as separate word classes nouns, verbs and adjectives (as a subclass of verbs). There is no mention of the idea that word classes may be inapplicable to Classical Chinese.

Oh come on. What do you expect on page 12 of a book? You have to start somewhere, and it's fine to say that "kill" is a verb and "horse" is a noun-- notice that he's using prototypes as examples. Notice also that he doesn't actually give a usable test, especially since he immediately mentions that verbs can be used as nouns and vice versa!

Yeah, these are all good points. I accept that Pulleyblank probably wasn’t the best source to ‘prove’ my speculations!

Let's look at just one example:

道可道，非恆道；
名可名，非恆名。

The first halves are usually translated something like "The Way that can be explained" and "the Name that can be named". But as you can see, the OC is quite simple, using 道 dao and 名 ming as both noun and verb. By a syntactic test, then, they are both. A literal translation would be "Way can way... Name can name..." You've said we can't use semantic tests, which is good, because it's hardly clear that "way, name" are things like horses.

…

A lexicon should avoid supplying information that can be given by a general rule. To use a silly example, it's not useful to say, for every noun in the language, "Can be used as the subject of a sentence." Of course it can. Slightly less silly, my dictionary occasionally includes definitions like "2. something resembling a (main word)." That's just as unnecessary: the word for any physical object can be used for things resembling it, models of it, pictures of it.

If hypothetically "can be used as a noun or a verb" is true of most words in the lexicon, then it's not a fact about those words at all, it's a fact about the language. It'd be more useful and more interesting to label the exceptions.

I agree with all these. In fact, this point was a key thesis of that François paper I linked. If all nouns, without exception, cannot be verbed, then you don’t need to mark them {noun, verb} at all; you can just say that a property of nouns in language X is that they can be used to head a VP, or whatever similar test you end up with.

One thing to note here is that in all known languages, nouns and verbs act as different classes! Even if nouns can be used to head VPs, and verbs can be used to head NPs, there are always minor differences in properties between them. For this reason, even languages like Classical Chinese should in theory be able to have {noun, verb}s, as distinct from both nouns and verbs.

I like how Dixon pictures this in his book Basic Linguistic Theory, vol. 2. (You can tell I like Dixon’s approach to word classes, right? ) He distinguishes four schemes:

dixon-nouns-verbs.png (8.73 KiB) Viewed 6051 times

That is, a word class of ‘nouns’ can still have the function of heading a VP, and vice versa for ‘verbs’. But they are still separate word classes, because they differ in their other morphosyntactic properties. For instance, in Nootka, only nouns may ‘be modified with expressions of property concepts, quantity or quantifiers’ (Nakayama 2001, quoted in Dixon); verbs are unacceptable in this syntactic frame.

Let’s flip this claim around. Is there any particular evidence, within English itself, to conclude that the ‘-ing’ form is actually two separate word types?

Sure, because there are two or more syntactic frames involved. You keep agreeing that syntactic tests are good, but keep ignoring what they say.

Not every syntactic test immediately defines a new word class! If, as you say, two syntactic frames delineate two separate word classes, it would mean that every single ‘-ing’ word is a member of both these word classes… and not only that, there are no words which would be a member of the one class but not the other. You yourself argued against such a configuration of word classes! Far simpler to say that there is just one word class, which may be used in several different frames. (Just like, for instance, English nouns may be used both as the head of an NP and as the modifier to another noun. Doesn’t mean ‘nouns’ are two word classes!)

Another way of looking at it is that a syntactic frame distinguishes two word classes only if they differ with respect to this test. For instance, ‘can be modified by quantity word’ is a frame which distinguishes nouns and verbs in Nootka, because they behave differently when subjected to this test: one is acceptable there, the other isn’t. The mere presence of two syntactic frames for one word isn’t nearly enough to define word classes — it just means that, well, one word class can be used in two different places, which is perfectly normal.

I am working purely synchronically, emically and pre-theoretically. (Those three adjectives are important; my whole analysis is based on them.)

"Pre-theoretical" is propaganda. Of course you have a theory-- that's what you're trying to enunciate. Things like "you can't use semantics to define word classes" are statements of theory. I don't mind you having a theory; but it's a bit annoying to pretend that you don't.

This is a fair criticism. I suppose what I meant is that it was pre-syntactic analysis — my definition of word classes requires only phonological analysis to be applied. Sorry for speaking incorrectly.

I have no idea what you mean by "emic" here. If you're making an analogy from phonology-- you can't do phonology without doing phonetics. If you're using the anthropological meaning, that'd mean that you want to use speakers' own terms and theories, which we already know can be quite erroneous.

I mean ‘emic’ to mean that my analysis works solely from the data of the language itself, as opposed to starting analysis based on theories from other languages (which would be ‘etic’). It is true that this could be confused with the anthropological meaning, but I don’t know of any better word to describe this.

I’ll also mention that terms like ‘gerund’ and ‘participle’ are tricky, because they’re really functional terms. A gerund is some verb form which is used as a formal noun, and a participle is some verb form which is used to formally modify a noun. That is, they describe the use of a particular verb form, not the form itself.

That is, they describe syntax, which you keep saying is acceptable for defining word classes, and indeed applaud in Pulleyblank. Which is it?

…yeah, there seems to have been a lot of bad wording in my previous post. I’ll try not to post subtle linguistic arguments again at 1AM, shall I?

Stating this more clearly: ‘gerund’ and ‘participle’ are basically syntactic tests, not word classes. One language might have one word class which can be used as a ‘gerund’, and a different one which can be used as a ‘participle’. Another language might have a single word class, used for both. English happens to have one single word class — the ‘-ing’ form — used for all of ‘gerund’, ‘participle’, ‘infinitive’ and ‘converb’. I won’t give more examples here, but you can find plenty in Nedjalkov’s chapter for Converbs in Cross-Linguistic Perspective (ed. Bossong & Comrie, 1995).

A good example: English hungry is a noun, Latin ēsurio is a verb, Dyirbal ŋamir is an adjective (Dixon 2010).)

Not really a good example: "hungry" is an adjective, and Latin also has an adjective, esuriens. But this is not a point of contention anyway; no one is saying we should find words for "hungry" and call them all adjectives.

That was a typo; I meant hunger. My point is that the root is a noun in English, a verb in Latin, and an adjective in Dyirbal; though both of the former have adjectives, they’re derived from the base word (hunger→hungry, ēsurio→ēsuriens).

One might quibble with the exact definitions, but that’s the style of thing I’m looking for — simple tests to separate words into different categories, each of which is homogeneous in its syntactic and morphological behaviour.

Yes, I know, and I've given a bunch of reasons why your tests will either be not so simple, or your categories not homogenous.

I disagree with this. If you have an inhomogeneous category, then it stands to reason that the inhomogeneity can be detected via a simple test… and if it can’t, then how much inhomogeneity can the category really have?

It's fine to hope that your tests will be simple. But so far, you seem to keep looking at English morphology (verb_m and noun_m), pointing out how simple they are, and just declaring that this will be the case when you've also taken syntax into account. When you actually provide syntactic tests, they've been simplistic.

So, you’re refuting my theory that you can determine word class via simple tests by… saying that I’ve provided simple tests to determine word class? I think there’s some part of this reasoning that I’ve missed.

Again, you have to start somewhere, and if you want to start with morphology, fine. But I've already shown you that verb_syn and noun_syn are far more complicated-- they are obviously related to the morphological classes, but not identical.

Hmm, where did you say this? The only concrete example related to this I recall is your graded list of syntactic frames, and I’ve already given my reasoning (several times!) for why I don’t think this says anything about word classes.

I think you're not realizing the distinction between an informal analysis, and a fuller treatment of syntax. Yes, you can create rough-and-ready categories or rules that will be useful and can even be used in an introductory textbook. This is not the same as researching and defining syntactic categories that will cover the wide range of syntactic behaviors people have found in the last 60 years. To bring us back to the original subject, it's pretty easy to define prepositions as things that appear before an NP, and create a PP that acts as a modifier to another NP or to a VP. But remember the 90%-90% rule. That definition doesn't always help when you're trying to pick apart the trickier (but still everyday) parts of English syntax-- the stuff that isn't just P + NP.

I totally agree with all this! All I’m trying to do here is define word classes. I’m not trying to do any more complex syntactic analysis. Just because we might postulate that prepositions are ‘things that appear before an NP, and create a PP that acts as a modifier to another NP or to a VP’, that doesn’t mean we’re saying that those are the only places where prepositions may appear — just that that’s their defining trait, that they can occur in this position. There’s no shortage of further syntactic analysis to do!

Also, remember that though I don’t focus on fully analysing syntax, having a solid definition of word classes is important for that purpose. Though it is the case that word class theory doesn’t have anything to say about the detailed syntax of prepositions, it would be pretty tricky to talk about that syntax without the notion of ‘prepositions’!

I think I've realized why you're not getting through to our friend. Elsewhere in the same post, he met my example of two OC words with different usage, and then repeated the same example in English.

as verbs, ‘picture’ means ‘to consider something as a picture’ (I pictured the house I grew up in), whereas ‘glow’ means ‘to have a glow form around’ (The mixture glowed as I added luciferase).

At first I thought "how is it a counter-argument to just agree with what I said?" Then it hit me. Bradrn is presenting this as a reason in favor of treating picture and glow as the same word class. In other words, we're out here talking about how to complete the task of syntactic word classification, and this whole time he's been advocating in favor of leaving the job half done. According to Bradrn, it is appropriate to say that picture and glow are the same.

Yes, exactly!

I might clarify this a bit, in terms of why I feel this is sufficient. I am talking solely about ‘syntactic word classification’. The relationship between nouns and their verbed correspondences, though fascinating, has nothing to do with syntax — the only morphosyntactic difference between ‘picture’ and ‘glow’ is that the former is transitive while the latter is intransitive. Otherwise, it is appropriate to treat them the same way, at least with regards to syntax.

[Moose-tache]

The relationship between nouns and their verbed correspondences, though fascinating, has nothing to do with syntax — the only morphosyntactic difference between ‘picture’ and ‘glow’ is that the former is transitive while the latter is intransitive. Otherwise, it is appropriate to treat them the same way, at least with regards to syntax.

So in my soon to be published "Madeline's Big Book of English Syntax," how am I to explain the fact that some words in the VP slot interact with surrounding arguments in one way, and some interact with surrounding arguments in a different way? If I say "just check the dictionary entry for each word," I've given up trying to explain English syntax in my book about English syntax. If I say "they're the same," then I'm lying to customers (this is unrealistic because it involves people buying a book I wrote, which I have recently learned is not a thing that is mathematically possible on this plane of existence, but let's just imagine). If there is a generalization, such as "this subclass always implies animate subjects" or "intransitives derived from nouns always imply [to have _]," then I should say that.

Speakers make lots of generalizations. Some of those generalizations are broad things like "nouny things can head NPs." Some of those generalizations are highly specific things like "epistemic modals in non-finite clauses can't be fronted," or something. It sounds like you want to take the five or ten most important generalizations about words and use that to make top-level categories, and then punch out! But what about all those other generalizations? Why do English speakers only allow certain adverbs to move freely? Why do they allow some time words to double as nouns, but not others? These are syntactic rules (or more accurately general syntactic guidelines) that English speakers use to create and parse sentences. Surely my Big Book of Syntax has to cover that stuff before I can purport to have told my readers (I know, I know, we're imagining here) that I've told them what kinds of words there are.

EDIT: I will repeat an example from earlier, with more information. For a lot of younger American English speakers, the word "even" can move to the end of the sentence, as in "I gave him my car, even. And he still wasn't happy!" Meanwhile, there is a word "just," as in "I just love your top!" that does not show this movement. There are some likely explanations for this difference. "Even" traditionally appears between the modal/auxiliary and main verb (or whatever you prefer to call these components of the verb phrase). It turns out, there is a huge open class of words that follow this behavior, and many of them can freely appear at the end of a sentence. It used to be, this did not include the word even. Now it does. Meanwhile the word "just" has a different pattern. Unlike "even," it appears just as frequently before and after the modal/auxiliary, and this is a pattern shared by very few other words. So while "just" is doing its own thing, "even" was behaving like an aberant member of a class and was brought in line. But it's even more complicated than that, because none of these "classes" are water-tight. They are just a series of overlapping generalizations. The word "tomorrow" can fill an NP slot. So can the word "now." Both can indicate time as adverbs or non-core argument phrases, or whatever you want to call them, but only "now" can split a VP when it does so. Which one is aberant? They're not going to flip the same switches on your Part-o-Speechanator 9000, nor will "even" or "just." And this is all without stepping one foot outside of the realm of syntax.

[bradrn]

So […] how am I to explain the fact that some words in the VP slot interact with surrounding arguments in one way, and some interact with surrounding arguments in a different way?

Let’s back up a bit here. What, exactly, are these different interactions you mention? I know I can’t think of any way picture and glow differ in their syntactic behaviour (apart from transitivity — which is actually pretty important, but not what you’re referencing, I think).

It sounds like you want to take the five or ten most important generalizations about words and use that to make top-level categories, and then punch out!

No, no, no, NO! That is not what I’m saying! If two different words have different syntactic behaviour, then they have different word classes. Thus, for instance:

Why do English speakers only allow certain adverbs to move freely?

Because the traditional term ‘adverb’, as is well-known, is but a cover term for a bunch of much smaller word classes which people don’t often disambiguate properly, but can be distinguished via tests such as this one.

Why do they allow some time words to double as nouns, but not others?

Because… um, I’m tempted to say that they’re different word classes again, but I actually can’t think of any examples of this, so I’ll just say I’m not sure exactly what you’re talking about here.

In general, we do end up saying ‘just check the dictionary entry for each word’ quite a lot. Some things just are lexically specified, and we can’t give a firm rule for which words have which behaviour. I liked how you yourself described this earlier:

I think as conlangers, we feel deeply unsettled by a language that is just a dictionary, i.e. a language in which each word carries its own explanation of how it interacts with other words. But in a way that is how languages operate. We make generalizations, but not consistently.

[Moose-tache]

I am a little confused. I think we might be talking past one another, because this post of yours seems like a complete 180.

Earlier you argued against both the proliferation of tiny word classes, and the idea that words cannot be lumped into word classes at all. But now you seem to accept these notions?

What, exactly, are these different interactions you mention?

While you were replying to me, I was adding an example to my previous post, which I think could be useful here. There is a lot of syntax going on beyond the question of transitivity and intransitivity. Whether a verb tells us that the subject is an experiencer or an agent or something else is part of syntax. Where words can move to in a sentence is syntax. There's loads of stuff to talk about. With "picture" and "glow," for example, the types of subjects they can take is not the same, and this is not something I could guess from the nouns "picture" and "glow." Or maybe I could? If there is a generalization that English speakers make when verbifying nouns that is relevant here, we should include it in our description. We might also find cases that mix traits from the two categories. The verb "to paint" does not have the same relationship with its subject, or with its own noun form, that "to picture" has. Maybe it's a matter of consulting a dictionary for asterisks; maybe it's a matter of generalizing from other verbs. Chinese grammarians do this, hence terms like Yidong and Shidong (and some words apparently can be Yidong or Shindong). It doesn't stop being syntax because it's lexically determined.

[bradrn]

Earlier you argued against both the proliferation of tiny word classes, and the idea that words cannot be lumped into word classes at all. But now you seem to accept these notions?

I don’t recall arguing against the former, and I did not and still do not accept the latter. My position is that word classes are discrete, and that gradations and prototypes are a poor way to analyse them, language-internally; none of this means that small word classes are unacceptable, and I am fully supportive of granularity in word classes.

What, exactly, are these different interactions you mention?

While you were replying to me, I was adding an example to my previous post, which I think could be useful here.

Thanks! See below for my response.

Whether a verb tells us that the subject is an experiencer or an agent or something else is part of syntax. … With "picture" and "glow," for example, the types of subjects they can take is not the same, and this is not something I could guess from the nouns "picture" and "glow."

Um, no, this is all part of semantics. In some languages (notably active-stative ones), this stuff is reflected in the syntax, but English isn’t one of them, as far as I can see.

Where words can move to in a sentence is syntax.

Agreed. But these two verbs don’t seem to differ in that regard.

Or maybe I could? If there is a generalization that English speakers make when verbifying nouns that is relevant here, we should include it in our description. We might also find cases that mix traits from the two categories. The verb "to paint" does not have the same relationship with its subject, or with its own noun form, that "to picture" has.

As I’ve said before, I fully agree that this stuff is all really interesting. It’s just… well, not relevant to this discussion, really.

I do have to admit that this is the weakest part of my argument so far, and it’s easy to come up with counterarguments. Dixon, for example, uses this as the basis for his reasoning that English nouns undergo zero-derivation when verbed, meaning that English is not a language where nouns can head a predicate. However here I prefer a simpler argument, requiring no semantics at all: English has some nouns which can’t be used as a predicate, and moreover those that can be used as a predicate have all the other properties characteristic of verbs, which leads to the same conclusion.

EDIT: I will repeat an example from earlier, with more information. For a lot of younger American English speakers, the word "even" can move to the end of the sentence, as in "I gave him my car, even. And he still wasn't happy!" Meanwhile, there is a word "just," as in "I just love your top!" that does not show this movement. There are some likely explanations for this difference. "Even" traditionally appears between the modal/auxiliary and main verb (or whatever you prefer to call these components of the verb phrase). It turns out, there is a huge open class of words that follow this behavior, and many of them can freely appear at the end of a sentence. It used to be, this did not include the word even. Now it does. Meanwhile the word "just" has a different pattern. Unlike "even," it appears just as frequently before and after the modal/auxiliary, and this is a pattern shared by very few other words. So while "just" is doing its own thing, "even" was behaving like an aberant member of a class and was brought in line.

Well, I would use this as an argument for splitting adverbs into several discrete classes, one of which behaves like ‘even’, one like ‘just’, and so on. Except…

But it's even more complicated than that, because none of these "classes" are water-tight. They are just a series of overlapping generalizations. The word "tomorrow" can fill an NP slot. So can the word "now." Both can indicate time as adverbs or non-core argument phrases, or whatever you want to call them, but only "now" can split a VP when it does so. Which one is aberant? They're not going to flip the same switches on your Part-o-Speechanator 9000, nor will "even" or "just." And this is all without stepping one foot outside of the realm of syntax.

…this is probably the single strongest argument I’ve seen so far for the existence of gradation in word classes. I’ll have to think about this some more, consider some more examples in greater detail, and see for myself whether ‘adverbs’ can be discretised in any useful way, or whether they are a true gradation after all.

[cedh]

I have no idea what you mean by "emic" here. If you're making an analogy from phonology-- you can't do phonology without doing phonetics. If you're using the anthropological meaning, that'd mean that you want to use speakers' own terms and theories, which we already know can be quite erroneous.

I mean ‘emic’ to mean that my analysis works solely from the data of the language itself, as opposed to starting analysis based on theories from other languages (which would be ‘etic’). It is true that this could be confused with the anthropological meaning, but I don’t know of any better word to describe this.

Maybe "empirical" vs. "theoretical", or "data-based" vs. "theory-based"?

Whether a verb tells us that the subject is an experiencer or an agent or something else is part of syntax.

Um, no, this is all part of semantics. In some languages (notably active-stative ones), this stuff is reflected in the syntax, but English isn’t one of them, as far as I can see.

So part of the misunderstanding here seems to be due to different ideas of what counts as syntax and what doesn't.
(My own position would be that a verb telling us that the subject is an experiencer is semantics, but a verb telling us that its experiencer-like subject should be marked with a dative case, for instance, would be syntax. Transitivity is clearly part of syntax too.)

[bradrn]

I have no idea what you mean by "emic" here. If you're making an analogy from phonology-- you can't do phonology without doing phonetics. If you're using the anthropological meaning, that'd mean that you want to use speakers' own terms and theories, which we already know can be quite erroneous.

I mean ‘emic’ to mean that my analysis works solely from the data of the language itself, as opposed to starting analysis based on theories from other languages (which would be ‘etic’). It is true that this could be confused with the anthropological meaning, but I don’t know of any better word to describe this.

Maybe "empirical" vs. "theoretical", or "data-based" vs. "theory-based"?

I suppose this could work, though honestly I prefer ‘emic’, and I’m pretty sure I’ve seen it used that way before.

Now that I think of it, the other common terms are ‘language-internal’ and ‘language-external’.

Whether a verb tells us that the subject is an experiencer or an agent or something else is part of syntax.

Um, no, this is all part of semantics. In some languages (notably active-stative ones), this stuff is reflected in the syntax, but English isn’t one of them, as far as I can see.

So part of the misunderstanding here seems to be due to different ideas of what counts as syntax and what doesn't.

Yes, quite probably. I know I’ve had arguments (well, one argument) about this in the past, though I don’t believe my views on this are that different to mainstream now.

(My own position would be that a verb telling us that the subject is an experiencer is semantics, but a verb telling us that its experiencer-like subject should be marked with a dative case, for instance, would be syntax. Transitivity is clearly part of syntax too.)

I agree with all of this.

[zompist]

Stating this more clearly: ‘gerund’ and ‘participle’ are basically syntactic tests, not word classes.

I'm just going to focus on one issue, because it's where your position is most baffling to me. To me, you seem to accept syntactic tests (to define word classes!) when you must, as in Old Chinese. Yet in English you define your word classes with the morphology and then stop. You quote writers offering syntactic tests, but whenever I offer a bit of syntax you dismiss it.

However, I suspect some of this is because you think that the examples provided don't distinguish classes of words. E.g. you may look at my list of graded noun/verb behaviors and think that it applies to any verb.

First, we could look deeper at that list, using many verbs. I immediately find a lot of exceptions. First, of course, verbs vary in argument structure, thus the oddness of

?I love to want.
*I love to put.
*I love to dare.
*I am a looker of cookies.
*I am a looker at of cookies.
*I am a becomer of cookies.
*I am a putter on of clothes.
*I am a keeper of going around in circles.

Some verbs can't be both progressive, and given an interval: *I am considering/wanting/becoming/lasting from four to seven." Some don't seem to like the progressive at all: *I am having this property." (Note that this is non-auxiliary 'have', and "possess" doesn't improve it.)

The + gerund doesn't work for some verbs:
The baking of cookies is a fine art.
*The turning thirteen is a fine art.
*The playing dead is a fine art.
*The lasting forever is limited to deities
*The dying inside is mortifying.
Beats me! / *The beating of me is a fine art.

Removing "the" fixes most of these.

Impersonal verbs can't go very high up the nouniness hierarchy:
Its raining/happening all the time bothers me.
?It loves to rain/happen.
*Raining/Happening is a fine art.
*It is a rainer (for days, of cat and dogs).
*It is a happener (of) all the time.

If you carefully worked through all this, you'd have to divide up "verbs" into a bunch of categories, which won't all neatly correspond to ones grammarians already have. And note that all the above comes from about 15 minutes staring at a word list— it is not an exhaustive list nor does it summarize what syntacticians have discovered.

(Also— this is not snark, but it's not my job to point out subcategories for you. You could have found these things too— or looked at syntax books which talk about word classes and their complications. I know you do look at writers, but you have to resist the urge to just look for bits that confirm your pre-existing belief.)

Nouns have to be divided up too. We already have subdivisions like count vs mass, animate vs inanimate, common vs proper. But there are lots of subtle little ones, like picture nouns, discussed in the SCK. Compare:

John appreciates this picture of himself.
John appreciates this monograph on him/*himself.

Does this mean we can't talk about nouns and verbs at all? Of course not— we can and do. It's useful to talk loosely at times— given prototype theory, it's even rational. But all the extra detail is why your talk of morphological English categories as "primitives" and "inviolate" and "simple", is unconvincing to me.

One more thing: as we do define subcategories, we can of course give them abstract names— some Latinization of "verbs-which-resist-gerunds". Or we use existing names like "impersonal verbs" and carefully avoid asking what "impersonal" means. Or we can just use prototype theory, which gives us a reason why some of these categories exist.

[Moose-tache]

Whether a verb tells us that the subject is an experiencer or an agent or something else is part of syntax.

Um, no, this is all part of semantics. In some languages (notably active-stative ones), this stuff is reflected in the syntax, but English isn’t one of them, as far as I can see.

So part of the misunderstanding here seems to be due to different ideas of what counts as syntax and what doesn't.
(My own position would be that a verb telling us that the subject is an experiencer is semantics, but a verb telling us that its experiencer-like subject should be marked with a dative case, for instance, would be syntax. Transitivity is clearly part of syntax too.)

No, all of this is part of syntax. Not in an exclusive way; it's semantics, too. But it is still relevant to syntax. Let me explain.
In English, some ditransitive verbs' behavior depends on what sort of arguments they take. If I say "I asked Tom" and "I asked a question," a listener knows that the animate argument is the indirect object equivalent in the sentence "I asked Tom a question." But not all verbs work this way. If I shorten "I gave a cat milk" to "I gave a cat," the implication is that "cat" has switched roles.
I think Brandrn wants to say that something is only syntax if it appears in an introductory paragraph about whether a language is SVO or SOV, and whether it's nominative-accusative or ergative-absolutive, but these considerations are part of syntax too. Imagine asking someone to draw a tree diagram of the above sentences. You could tell them "Oh, just look up how each verb behaves," but that doesn't mean your tree diagram stops being a syntactic diagram and becomes a semantic one. And even more importantly for syntax, these lexical differences will conform to patterns and generalizations. For example, our diagramer will come back and say "I looked up ask and gave, and found other verbs that behave the same way on my tree diagram. Then I looked up say and tell, and found even more patterns." Are we supposed ot throw up our hands and say "semantics be crazy?"

[zompist]

As I was re-reading my post above, I thought of some more interesting behavior. A + gerund seems to be more restricted than the + gerund, and varies by verb:

The baking of cookies is a fine art.
*A baking of cookies happened last night.
A mustering of troops happened last night.

In the grass, there was a rustling.
*In the kitchen, there was a baking.

The snoring bothered me all night.
*A snoring bothered me all night.
An itching bothered me all night.

Oh, and one more thing to think about. So far as I can think of, the frame "hang a N" allows only four nouns: left, right, Louie, Ralph. Does this simple test define a subclass of verbs? If yes, you are (as I've been saying all along) going to have a shit-ton of word classes. If not... how does this test differ from "__ NP defines a preposition"? Note that I can define a succession of larger classes of verb: "turn <direction>", which also allows cardinal directions; "go <destination>" which also allows up/down/home— and this may or may not be the same construction as "go N" as in "go communist, go crazy, go anti-vax, go native", but not "?go farmer, ?go international banker, *go sick, *go brother-in-law". Even wider: "become NP" allows a huge variety of things, but not, I think, "*He became the baking of cookies."

Now traditionally this gets swept under the rug as "idioms", but productive idioms are still syntax, and you can even make a syntactic theory (Construction Grammar) from this.

[bradrn]

So far as I can think of, the frame "hang a N" allows only four nouns: left, right, Louie, Ralph.

Wait, what? This doesn’t work for me at all. This is obviously a non-productive idiom, and we don’t expect those to obey normal restrictions on words.

(Still thinking about the rest of your and Moose-tache’s posts; not quite sure I agree with all of them, but very interesting food for thought nonetheless!)

[zompist]

So in my soon to be published "Madeline's Big Book of English Syntax," [...]

Hey, I'd buy it, based on the title alone.

McCawley wanted to call his syntax textbook "More about English syntax than you probably want to know", but the publishers nixed it.

[Richard W]

Stating this more clearly: ‘gerund’ and ‘participle’ are basically syntactic tests, not word classes.

I'm just going to focus on one issue, because it's where your position is most baffling to me. To me, you seem to accept syntactic tests (to define word classes!) when you must, as in Old Chinese. Yet in English you define your word classes with the morphology and then stop. You quote writers offering syntactic tests, but whenever I offer a bit of syntax you dismiss it.

However, I suspect some of this is because you think that the examples provided don't distinguish classes of words. E.g. you may look at my list of graded noun/verb behaviors and think that it applies to any verb.

First, we could look deeper at that list, using many verbs. I immediately find a lot of exceptions.

And if you look harder, you will find exceptions within exceptions.

First, of course, verbs vary in argument structure, thus the oddness of

?I love to want.
*I love to put.
*I love to dare.
*I am a looker of cookies.
*I am a looker at of cookies.
*I am a becomer of cookies.
*I am a putter on of clothes.
*I am a keeper of going around in circles.

But She is such a putter on of airs.

Some verbs can't be both progressive, and given an interval: *I am considering/wanting/becoming/lasting from four to seven." Some don't seem to like the progressive at all: *I am having this property." (Note that this is non-auxiliary 'have', and "possess" doesn't improve it.)

But when discussing timetables, I'm having the hall from three to five.. The problem with the initial examples appears to be incongruity. It is perfectly possible to say, 'I will be considering applications from four to seven'.

The + gerund doesn't work for some verbs:
The baking of cookies is a fine art.
*The turning thirteen is a fine art.
*The playing dead is a fine art.
*The lasting forever is limited to deities
*The dying inside is mortifying.
Beats me! / *The beating of me is a fine art.

Removing "the" fixes most of these.

I'm not sure you have the right generalisation here. You could also add:

*The baking cookies is a fine art.

I think we have the grammatical sentence 'The beating of the drum is a fine art'.

[bradrn]

Stating this more clearly: ‘gerund’ and ‘participle’ are basically syntactic tests, not word classes.

I'm just going to focus on one issue, because it's where your position is most baffling to me. To me, you seem to accept syntactic tests (to define word classes!) when you must, as in Old Chinese. Yet in English you define your word classes with the morphology and then stop.

Not so! I quoted Dixon’s definition of English word classes above, and said I agreed with it:

[verbs], takes suffix -ing
[nouns], may be immediately preceded by an article and need not be followed by another word
[adjectives], may be immediately preceded by an article and is then followed by another word (either one from class [nouns] or another word from class [adjectives])

Most of these are syntactic tests — perhaps not in the Chomskyan ‘tree diagrams analysed to the nth degree’ style, but certainly syntax, as opposed to morphology.

However, I suspect some of this is because you think that the examples provided don't distinguish classes of words. E.g. you may look at my list of graded noun/verb behaviors and think that it applies to any verb.

Yes, I definitely made that mistake: I thought it could apply to any verb. On reflection, this was of course false.

First, we could look deeper at that list, using many verbs. I immediately find a lot of exceptions. First, of course, verbs vary in argument structure, thus the oddness of

?I love to want.
*I love to put.
*I love to dare.
*I am a looker of cookies.
*I am a looker at of cookies.
*I am a becomer of cookies.
*I am a putter on of clothes.
*I am a keeper of going around in circles.

I suspect the differences here are almost entirely due to transitivity, and in obvious ways too. ‘Want’ and ‘put’ are both transitive; compare I love to wish and I love to throw, which are intransitive. I use ‘dare’ so rarely I’m honestly not sure about both its transitivity status and the grammaticality of ?I love to dare, but I think it’s transitive also: I dare you to [complement]. It takes a complement, but also a direct object. Meanwhile, ‘look’ is intransitive, and cannot be used with a direct object. I’m not sure why you asterisk I am a looker at of cookies and I am a putter on of clothes, since those are both 100% fine for me, if a bit stilted. (Compare Shakespeare’s Autolycus, a snapper-up of unconsidered trifles, though I do appreciate that not all of his language is still grammatical today.)

The only two here I’m unsure about are *I am a becomer of cookies and *I am a keeper of going around in circles: these could well be true instances of non-transitivity-related differences in argument structure, but even here I’m not convinced. For the first clause, some of it may simply be semantic incoherence: a becomer of trees doesn’t sound nearly so bad. If you put it in appropriate context, even the original clause can become positively grammatical, albeit quite literary: I am a lump of dough in the kitchen; I am a becomer of cookies. As for the latter, I’m not sure, and I suspect that this is indeed a true ‘in between’ word, especially since it seems to be in the middle of turning into an auxiliary.

Notwithstanding, this doesn’t seem like a particularly strong argument for a gradation in word classes! In fact, quite the reverse: most of the above words can be neatly separated into ‘transitive’ and ‘intransitive’ (and ‘ambitransitive’ too, which is really just {transitive,intransitive}), which explains most of their behaviour. And the distinction between those two categories is a word class distinction. This isn’t even controversial: guides to dictionary-making have been advocating this for a while. I myself have been distinguishing ‘vt.’ and ‘vi.’ word classes in my conlang dictionaries for quite a while too. The only reason we can talk about ‘verbs’ as a word class at all is because transitive and intransitive verbs are almost exactly parallel in their properties, almost the sole difference being that one requires an extra object. It’s only a small step from that to breaking out new word classes for smaller variations in argument structure; I strongly suspect there won’t be all that many, though I do need to properly confirm this first.

Some verbs can't be both progressive, and given an interval: *I am considering/wanting/becoming/lasting from four to seven."

…I’m confused about this. First, I thought you were wrong about this, and wrote a long message here about how these verbs are actually transitive, with counterexamples to your claim. Then I looked at my post some hours later, realised my ‘counterexamples’ weren’t all that grammatical themselves, and replaced it with another long message about how you’re entirely correct about this. Then, Richard W replied with a genuine counterexample (I am considering applications from four to seven). Now I don’t know what to think.

Some don't seem to like the progressive at all: *I am having this property." (Note that this is non-auxiliary 'have', and "possess" doesn't improve it.)

This is a well-known semantic restriction on stative verbs (found in e.g. Comrie). On the other hand, does this mean that stativity has an impact on word class? I don’t see why not; I hear ‘stative verbs’ are distinguishable as a separate word class in lots of other languages (e.g. split-S ones).

The + gerund doesn't work for some verbs:
The baking of cookies is a fine art.
*The turning thirteen is a fine art.
*The playing dead is a fine art.
*The lasting forever is limited to deities
*The dying inside is mortifying.
Beats me! / *The beating of me is a fine art.

Funnily enough, the same thing happened with this one as with the one above. I thought you were wrong, then I thought you were right, then I saw Richard W’s criticism and agreed with it. Now I don’t know what to think about this one either, though at least this time I know I agree with the criticism more than the argument.

Impersonal verbs can't go very high up the nouniness hierarchy:
Its raining/happening all the time bothers me.
?It loves to rain/happen.
*Raining/Happening is a fine art.
*It is a rainer (for days, of cat and dogs).
*It is a happener (of) all the time.

Well, I have no objection to calling ‘impersonal verbs’ their own word class. It’s only consistent with calling transitive and intransitive verbs separate word classes.

If you carefully worked through all this, you'd have to divide up "verbs" into a bunch of categories, which won't all neatly correspond to ones grammarians already have. And note that all the above comes from about 15 minutes staring at a word list— it is not an exhaustive list nor does it summarize what syntacticians have discovered.

I was about to admit that you were right here, but then Richard W’s post came along, I started to disagree with more of your arguments, and now — yep, you guessed it! — I don’t know what to think. But it certainly seems that all your examples were (a) not actually evidence for a gradation at all, or (b) incorrect.

(Also— this is not snark, but it's not my job to point out subcategories for you. You could have found these things too— or looked at syntax books which talk about word classes and their complications. I know you do look at writers, but you have to resist the urge to just look for bits that confirm your pre-existing belief.)

Yes, I know this. As you may be able to tell, I am pretty terrible at syntactic analysis: I make counterexamples that aren’t, and then miss ones that are really there, and get myself all confused in the process. At some point I need to properly sit down, with a wordlist and a blank page, and see if I can find any interesting behaviour.

…but not now! It’s the middle of the night, and my attempts at midnight syntax haven’t gone so well in the past, so I’ll wait till tomorrow, when I’m properly awake

Nouns have to be divided up too. We already have subdivisions like count vs mass, animate vs inanimate, common vs proper. But there are lots of subtle little ones, like picture nouns, discussed in the SCK. Compare:

John appreciates this picture of himself.
John appreciates this monograph on him/*himself.

Sure, I accept this as a valid example of a smaller word class.

Does this mean we can't talk about nouns and verbs at all? Of course not— we can and do. It's useful to talk loosely at times— given prototype theory, it's even rational. But all the extra detail is why your talk of morphological English categories as "primitives" and "inviolate" and "simple", is unconvincing to me.

I think there is some miscommunication here. When I talk about something being ‘primitive’, I mean it in the programming sense: like one might say that ‘in JavaScript strings are a primitive type’, and so on. In this case, I mean that word classes should be defined such that later stages of analysis can just assume they exist; that is, such that they can be treated as primitives for syntactic analysis. I meant ‘inviolate’ in the same sense: the later stages don’t need to question whether ‘verbs’, say, form a coherent category. And ‘simple’ applies to each individual test: combining all the tests together can make the defining features for a single word class quite involved, in fact!

One more thing: as we do define subcategories, we can of course give them abstract names— some Latinization of "verbs-which-resist-gerunds". Or we use existing names like "impersonal verbs" and carefully avoid asking what "impersonal" means. Or we can just use prototype theory, which gives us a reason why some of these categories exist.

I honestly don’t understand what you’re saying here. Of course we can ask questions like ‘what does “impersonal” mean’; it means that the verb can only take ‘it’ as its subject. And how, exactly, does prototype theory give a ‘reason’ for all this‽ It really gives no reasons, just post-hoc justifications; it just gives us a language to describe some phenomena we can see.

Also, though I am growing more and more convinced that categories like ‘nouns’ and ‘verbs’ are inhomogeneous as you say, it’s quite a leap from that to ‘prototypes describe anything’. The mental picture I have now is more hierarchical in nature. On one level, you have coarse word classes like ‘nouns’ and ‘transitive verbs’ and so on, which are the ‘primitives’ I mentioned above, defined by simple syntactic tests one can apply with minimal analysis. On another level of analysis, one realises that these apparently homogeneous word classes are in fact much more granular, and consist of innumerable smaller word classes defined by their ability to enter/not enter into specific constructions; these word classes tend to grade into each other, though mostly within the bounds coarser word classes (I have yet to see a good example of a gradation between ‘nouns’ and ‘verbs’ of any sort, for instance). But none of this requires prototype theory: all the more granular word classes are more or less equal in stature, except perhaps that some of them are a bit numerically larger.

…but as I said, I’m still not convinced that this level of granularity even exists, so it’s a bit of a moot point for now.

No, all of this is part of syntax. Not in an exclusive way; it's semantics, too. But it is still relevant to syntax. Let me explain.
In English, some ditransitive verbs' behavior depends on what sort of arguments they take. If I say "I asked Tom" and "I asked a question," a listener knows that the animate argument is the indirect object equivalent in the sentence "I asked Tom a question." But not all verbs work this way. If I shorten "I gave a cat milk" to "I gave a cat," the implication is that "cat" has switched roles.

I don’t accept that this is syntax. This is semantics, pure and simple.

Question: is there any syntactic test that could be used to distinguish ‘ask’ from ‘give’? For instance, unaccusative and unergative verbs can be distinguished by syntax: though the window broke and the man sung appear syntactically parallel in a similar way to your example, their radically different semantics are reflected by the fact that the past participle the broken window is grammatical, while *the sung man is not. If you can find a similar test to distinguish ‘ask’ from ‘give’, I will fully accept that this is syntax. (I note that the past participle test doesn’t work: the asked person and the given cat are equally grammatical.)

I think Brandrn wants to say that something is only syntax if it appears in an introductory paragraph about whether a language is SVO or SOV, and whether it's nominative-accusative or ergative-absolutive, but these considerations are part of syntax too.

At one point I did indeed think this (with a bit more subtlety of course), but I have now been persuaded by wiser minds that I was wrong. Lots of things are part of syntax… just not what you said.

Imagine asking someone to draw a tree diagram of the above sentences. You could tell them "Oh, just look up how each verb behaves," but that doesn't mean your tree diagram stops being a syntactic diagram and becomes a semantic one. And even more importantly for syntax, these lexical differences will conform to patterns and generalizations. For example, our diagramer will come back and say "I looked up ask and gave, and found other verbs that behave the same way on my tree diagram. Then I looked up say and tell, and found even more patterns." Are we supposed ot throw up our hands and say "semantics be crazy?"

But I’m not talking about tree diagrams! I’ve said this numerous times. I’m talking about word classes, at a level of analysis before that of tree diagrams. The diagrams themselves are an abstraction built upon other things including, I should point out, word classes — one of the ‘primitives’ required to build these diagrams in the first place. And if you’re doing this sort of pre-theoretic analysis (yes, I’m misusing the word, but you know what I mean), focusing only on the surface structure… then yes, ask and give have the same syntax.

(Also, I note that you’ve given exactly two patterns. How exactly is this meant to be ‘semantics be crazy’?)