Syntax random

[bradrn]

Oh, on the first point, here's an interesting list of significant mid-level results of generative linguistics that a group of linguists came up with a few years ago.

I found that pretty interesting! A question: they say that ‘The hierarchy of projections as reflected in free words is the same one that is reflected in morphological structure when morphemes express the same notions as the free words’. This sounds pretty interesting, but I can’t quite understand what they mean by this. Can anyone explain what they’re saying here?

And here’s a couple of other interesting generalisations I’d like to highlight:

Cinque’s version of Greenberg’s U20: Only one unmarked order is found prenominally for Dem, Num, and Adj, namely Dem > Num > Adj > N; ordering possibilities increase as N is further to the left in the sequence. The facts suggest (i) a universal hierarchy Dem > Num > Adj > N, where these categories exist, (ii) the possibility of leftward but not rightward movement of projections of N to derive some other orders, and (iii) the absence of such movement of adnominal modifiers alone (e.g. no information-neutral movement of Adj across Num and/or Dem unless it is in a projection containing N) (May generalize to other categories)

SOV scrambling: All SOV languages allow a degree of word order freedom (scrambling); VO languages may not.

Structural agreement (Very robust): There is a structural bias affecting agreement such that nominals higher in the clause are agreed with in preference to lower nominals, except where marked case on a higher nominal may disqualify it. (reflected in subject agreement over object agreement)

Ergative subjects (Robust): Asymmetries between arguments for purposes of unmarked word order, binding, and control work the same way in nominative and ergative languages. Clause structure in ergative and accusative languages is homomorphic

No Active Case: No language has an active system of case marking, whereas active systems of agreement marking are possible. (Baker and Bobaljik in press/in progress, but well documented)

[akam chinjir]

I found that pretty interesting! A question: they say that ‘The hierarchy of projections as reflected in free words is the same one that is reflected in morphological structure when morphemes express the same notions as the free words’. This sounds pretty interesting, but I can’t quite understand what they mean by this. Can anyone explain what they’re saying here?

Yeah, "projection" is one of those technical terms that's really hard to get used to, imo. Here it works pretty well if you read it just as "phrase."

Here's an example that I hope is straightforward. Suppose you have separate morphemes for past tense and future, and both can occur in a single clause (and we're not worried about anything else except the verb itself).

First, the future morpheme takes the verb phrase as its complement, resulting in a new phrase (a new projection).

Then, the past tense morpheme takes that new phrase as its complement, resulting in yet another new phrase (another new projection).

The idea is that this is how it works regardless of whether the past tense and future are represented by independent words.

One important source for this sort of view is Mark Baker's The mirror principle and morphosyntactic explanation. (Baker was one of the linguists who helped draw up the list, though I expect this one would've been on there regardless. And as stated it's also drawing a lot on Cinque's work on adverbs, which compares adverb order with affix order in many languages.)

Cinque’s version of Greenberg’s U20: Only one unmarked order is found prenominally for Dem, Num, and Adj, namely Dem > Num > Adj > N; ordering possibilities increase as N is further to the left in the sequence. The facts suggest (i) a universal hierarchy Dem > Num > Adj > N, where these categories exist, (ii) the possibility of leftward but not rightward movement of projections of N to derive some other orders, and (iii) the absence of such movement of adnominal modifiers alone (e.g. no information-neutral movement of Adj across Num and/or Dem unless it is in a projection containing N) (May generalize to other categories)

You actually get the same sort of thing with functional material like tense and modality morphemes: their order is more strict when they go before the verb, and the most common postverbal orders are (first) the reverse of the preverbal order, and (second) the same as it. (And this doesn't depend on whether they are represented by independent words or by affixes; though if they are after the verb they are overwhelmingly likely to be suffixes rather than independent particles.)

SOV scrambling: All SOV languages allow a degree of word order freedom (scrambling); VO languages may not.

I think this one may actually be an exceptionless universal; or at least it's sometimes talked about as if it is.

Ergative subjects (Robust): Asymmetries between arguments for purposes of unmarked word order, binding, and control work the same way in nominative and ergative languages. Clause structure in ergative and accusative languages is homomorphic

I know this one's been called into question for some syntactically ergative languages. (But like they say, it's fine to think that these generalisations don't work for all languages, just you'll probably need to supply an argument if you think some language is an exception.)

No Active Case: No language has an active system of case marking, whereas active systems of agreement marking are possible. (Baker and Bobaljik in press/in progress, but well documented)

I think we've had an exchange about this one (and I shared the Baker and Bobaljik paper with you). It's definitely an interesting pattern, even if it's got exceptions.

[bradrn]

I found that pretty interesting! A question: they say that ‘The hierarchy of projections as reflected in free words is the same one that is reflected in morphological structure when morphemes express the same notions as the free words’. This sounds pretty interesting, but I can’t quite understand what they mean by this. Can anyone explain what they’re saying here?

Yeah, "projection" is one of those technical terms that's really hard to get used to, imo. Here it works pretty well if you read it just as "phrase."

Here's an example that I hope is straightforward. Suppose you have separate morphemes for past tense and future, and both can occur in a single clause (and we're not worried about anything else except the verb itself).

First, the future morpheme takes the verb phrase as its complement, resulting in a new phrase (a new projection).

Then, the past tense morpheme takes that new phrase as its complement, resulting in yet another new phrase (another new projection).

The idea is that this is how it works regardless of whether the past tense and future are represented by independent words.

Maybe I’m not understanding this correctly, but are you implying here that the past tense morpheme occurs together with the future tense morpheme? That’s what it sounds like — you say that ‘the past tense morpheme takes that new phrase as its complement’, where I believe that ‘that new phrase’ refers to the phrase generated by the future morpheme.

One important source for this sort of view is Mark Baker's The mirror principle and morphosyntactic explanation. (Baker was one of the linguists who helped draw up the list, though I expect this one would've been on there regardless. And as stated it's also drawing a lot on Cinque's work on adverbs, which compares adverb order with affix order in many languages.)

Thanks for the recommendation — I’ll have to read that then!

Cinque’s version of Greenberg’s U20: Only one unmarked order is found prenominally for Dem, Num, and Adj, namely Dem > Num > Adj > N; ordering possibilities increase as N is further to the left in the sequence. The facts suggest (i) a universal hierarchy Dem > Num > Adj > N, where these categories exist, (ii) the possibility of leftward but not rightward movement of projections of N to derive some other orders, and (iii) the absence of such movement of adnominal modifiers alone (e.g. no information-neutral movement of Adj across Num and/or Dem unless it is in a projection containing N) (May generalize to other categories)

You actually get the same sort of thing with functional material like tense and modality morphemes: their order is more strict when they go before the verb, and the most common postverbal orders are (first) the reverse of the preverbal order, and (second) the same as it. (And this doesn't depend on whether they are represented by independent words or by affixes; though if they are after the verb they are overwhelmingly likely to be suffixes rather than independent particles.)

I didn’t know this was a more general principle! Do you have any more resources about word order in verb phrases? I’d be quite interested to know more about that.

Oh, and while we’re talking about this, I should probably also mention Dryer’s paper On the order of demonstrative, numeral, adjective and noun. (I believe it’s paywalled, but sci-hub has it if you’re willing to use that.) It’s a very interesting summary of all the attested (and unattested) Dem/Num/Adj/N orders, and an attempt (a very good one, I thought) to explain of why this is in terms of some general principles.

Relatedly, I’ve been wondering about the order of classifiers and classified words. Aikenvald’s Classifiers says that numeral classifiers always form a constituent with the numeral, and thus always occur next to the numeral, but clearly other classifiers can act differently, as seen with e.g. demonstrative classifiers in Hmong:

lub

CL:OBJECT

tsev

house

no

this

This house

SOV scrambling: All SOV languages allow a degree of word order freedom (scrambling); VO languages may not.

I think this one may actually be an exceptionless universal; or at least it's sometimes talked about as if it is.

Well, I hadn’t heard of it, which was why I thought it was interesting.

Actually, now that I think about it, could this be confusing scrambling and topic-prominence? I know that many topic-prominent languages — which have a very strong tendency to be V-final — appear to have ‘scrambling’, but really just have a word order rule in terms of topic rather than subject. For instance, from Lisu (Hope 1974, cited in Li and Thompson 1976):

làthyu

people

nya

TOP

ánà

dog

khù-a

bite-DECL

People_topic bite dogs / Dogs bite people_topic

ánà

dog

nya

TOP

làthyu

people

khù-a

bite-DECL

People bite dogs_topic / Dogs_topic bite people

No Active Case: No language has an active system of case marking, whereas active systems of agreement marking are possible. (Baker and Bobaljik in press/in progress, but well documented)

I think we've had an exchange about this one (and I shared the Baker and Bobaljik paper with you). It's definitely an interesting pattern, even if it's got exceptions.

Does it have exceptions? Ever since I first learnt about this pattern (in Dixon’s Ergativity), I always thought that no exceptions were known.

[akam chinjir]

Maybe I’m not understanding this correctly, but are you implying here that the past tense morpheme occurs together with the future tense morpheme? That’s what it sounds like — you say that ‘the past tense morpheme takes that new phrase as its complement’, where I believe that ‘that new phrase’ refers to the phrase generated by the future morpheme.

Oops! Yes, I was imagining a case where both past and future morphemes are present.

I didn’t know this was a more general principle! Do you have any more resources about word order in verb phrases? I’d be quite interested to know more about that.

I think the largest collection of data on this stuff that I've seen is in an appendix to Julien, Syntactic heads and word formation. One very direct discussion of that pattern is Svenonius's paper "1... 3–2" (yeah, great title!) in The Oxford handbook of linguistic interfaces.

Oh, and while we’re talking about this, I should probably also mention Dryer’s paper On the order of demonstrative, numeral, adjective and noun. [...] It’s a very interesting summary of all the attested (and unattested) Dem/Num/Adj/N orders, and an attempt (a very good one, I thought) to explain of why this is in terms of some general principles.

Dryer actually puts a great many of his papers on his webpage, though I don't know if that one's there. Also, there's an exchage between him and Cinque about this stuff. (I think at first Cinque found some extra word orders that Dryer hadn't counted, but maybe after Dryer argued that some are attested that Cinque rules out. But it's been a while since I read this stuff.)

Relatedly, I’ve been wondering about the order of classifiers and classified words. Aikenvald’s Classifiers says that numeral classifiers always form a constituent with the numeral, and thus always occur next to the numeral,

The orders that are usually recogised are NUM CL N, N NUM CL, and N CL NUM, the last one least common. (Notice again that word order shows more variation when things follow the lexical head.) Though the stuff I've read mostly doesn't agree that the number and a classifier form a constituent, fwiw.

but clearly other classifiers can act differently, as seen with e.g. demonstrative classifiers in Hmong:

lub

CL:OBJECT

tsev

house

no

this

This house

Interesting! I wonder if it matters that it involves a demonstrative rather than a number. (But I don't know anything about Hmong.)

Well, I hadn’t heard of it, which was why I thought it was interesting.

Yeah, I agree that it's pretty stunning.

Actually, now that I think about it, could this be confusing scrambling and topic-prominence?

It's not confusion to be a part of a different descriptive tradition Anyway topicalisation isn't the only pattern here (I don't know if it's even an especially common one). There's also focus, and I'm pretty sure you can get scrambling where the different word orders don't seem to have any semantic or pragmatic difference.

Does it [AK: the generalisation that no languagess have active/stative case-marking] have exceptions? Ever since I first learnt about this pattern (in Dixon’s Ergativity), I always thought that no exceptions were known.

There are at least some cases that Baker and Bobaljik think they have to address as possible counterexamples, and it's easy to imagine their arguments won't convince everyone. (According to WALS, Basque, Drehu, Georgian, and Imonda are counterexamples, fwiw.)

[bradrn]

I didn’t know this was a more general principle! Do you have any more resources about word order in verb phrases? I’d be quite interested to know more about that.

I think the largest collection of data on this stuff that I've seen is in an appendix to Julien, Syntactic heads and word formation. One very direct discussion of that pattern is Svenonius's paper "1... 3–2" (yeah, great title!) in The Oxford handbook of linguistic interfaces.

I’ll have to see if I can find those then. (Although I suspect that searching for Svenonius’s paper could be tricky!)

Oh, and while we’re talking about this, I should probably also mention Dryer’s paper On the order of demonstrative, numeral, adjective and noun. [...] It’s a very interesting summary of all the attested (and unattested) Dem/Num/Adj/N orders, and an attempt (a very good one, I thought) to explain of why this is in terms of some general principles.

Dryer actually puts a great many of his papers on his webpage, though I don't know if that one's there.

I checked — this one isn’t.

Relatedly, I’ve been wondering about the order of classifiers and classified words. Aikenvald’s Classifiers says that numeral classifiers always form a constituent with the numeral, and thus always occur next to the numeral,

The orders that are usually recogised are NUM CL N, N NUM CL, and N CL NUM, the last one least common. (Notice again that word order shows more variation when things follow the lexical head.)

Aikhenvald claims that CL NUM N is found in Ibibio (source: Greenberg 1972). In general, the pattern here doesn’t seem to be one of before-head vs after-head: rather, it seems that NUM CL is very common and CL NUM is vanishingly rare.

but clearly other classifiers can act differently, as seen with e.g. demonstrative classifiers in Hmong:

lub

CL:OBJECT

tsev

house

no

this

This house

Interesting! I wonder if it matters that it involves a demonstrative rather than a number. (But I don't know anything about Hmong.)

I suspect that this does matter, but my problem is that I don’t really have any data — aside from Hmong, the only language I know of with this sort of order is Ejagham, which has numeral classifiers with the rather unusual order CL N NUM.

Actually, now that I think about it, could this be confusing scrambling and topic-prominence?

… I'm pretty sure you can get scrambling where the different word orders don't seem to have any semantic or pragmatic difference.

Any examples? I thought that variations in word order always has some sort of semantic or pragmatic difference.

Does it [AK: the generalisation that no languagess have active/stative case-marking] have exceptions? Ever since I first learnt about this pattern (in Dixon’s Ergativity), I always thought that no exceptions were known.

There are at least some cases that Baker and Bobaljik think they have to address as possible counterexamples, and it's easy to imagine their arguments won't convince everyone. (According to WALS, Basque, Drehu, Georgian, and Imonda are counterexamples, fwiw.)

How does Basque show active-stative case-marking? But, now that I think about it, they’re right about Georgian — although I imagine that the fact that Georgian also shows a simultaneous animacy-based split would also help a lot in explaining this.

[Kuchigakatai]

Oh, on the first point, here's an interesting list of significant mid-level results of generative linguistics that a group of linguists came up with a few years ago.

Actually, if you read that, you might notice that the structure-dependence of binding is on the list. Does analysing things in terms of structure count as a tool? C-command in particular? Cyclicity? Those are pretty fundamental to how most syntacticians go about things, at least. (Whereas linear order isn't part of the usual toolkit.)

Oh, by all means, those three things (structure-dependence of binding, c-command, cyclicity) are definitely discoveries of Generative Grammar research! I don't mean to say that it hasn't discovered many things, including important things. Binding before GG was only talked about as going from an individual word to an antecedent as if the antecedent existed as an idea (maybe because ungrammaticality wasn't taken too seriously; often the response to such a thing was 'it's not attested', 'people don't say that', than saying something was ungrammatical). Islands were also a discovery of GG research I think, even if that article doesn't list them as their own item.

What I am observing or maybe better complaining about is that some of these things GG syntacticians attribute as discoveries of GG were already talked about, decades before, in the grammars of particular languages. It seems to me they became notions in general linguistics with GG because linguistics had also just recently gotten started when the GG program started... I appreciate that "transformations" is not on the list of your link, but I've definitely heard that from some of said people. That old grammars (of which Allen & Greenough is just an example) don't get credit for them mostly just reflects that they were aimed at people in other areas (like Classics as a field of literature and archeology in that case), since linguistics didn't exist much as its own thing yet in 1903. I notice that authors of Latin/Greek grammars these days tend to come from a linguistics background now, e.g. the authors of the recent Cambridge Grammar of Classical Greek (2018).

As for the approaches that Ser mentioned that I guess don't count as frameworks, I don't know that not offering university courses about a thing counts as pretending a thing doesn't exist. Like, say, if you looked at Kayne's book on French syntax, are there relevant phenomena he misses or mistakes he makes because he doesn't pay attention to that work? (That's assuming he doesn't, I don't know either way.)

It's not so much about "relevant" phenomena being missed, but about the lack of acknowledgement of the massively common model used to talk about syntax in the world in syntax courses and books at English-speaking universities, since the lion's share of the field seems to be taken by people who follow old Chomsky or new Chomsky. Open about every reference grammar for language learners, and the model used is that other one.

(Regarding old Chomsky, even though explicit talk about transformations from D-Structure to S-Structure (or LP to FP) seem to be gone to a large or complete extent, new research keeps coming out using X-bar theory, like this article from 2018 on the merits of DP vs. NP in Estonian, a language without articles, largely concerned about which one out of a (non-article) determiner or a main noun is worth selecting as the specifier or head. Note part of the author's argument involves changes in sentences in transformations...)

Dixon, who bradrn mentioned, does begin vol. 1 of Basic Linguistic Theory talking about it (§1.2, pages 1-4), in what I'd call an attack against formal theorists by misrepresenting them as failing at their analyses of parts of a language's grammar by not considering it whole or by using very different approaches from each other (for one, this can also be said of people of the "basic linguistic" model!, try reading two grammars of a sub-saharan African language by two different authors sometime... and he himself complains about this later on). He also says something about BLT descending from Greek and Sanskrit grammarians' thought, but that's true of anything... Strangely enough, in most of the rest of the three volumes he mostly talks about typology and practical language documentation rather than focusing on the model itself or the history of the model though.



Regarding "relevant" phenomena missed because of not paying attention to such other work, I've heard complaints about the lack of coverage of a given language's grammar in syntactic discussions even from Chomskian syntacticians themselves, so that's not something I was thinking of bringing up. I've heard of it in terms of not talking about enough "types of sentences".

I don't know about Kayne's book (I just saw it's a nice long one, 490 pp.), but e.g. a few times I've tried to find discussions by Chomskians on the Spanish sentence-initial que, specifically in the uses that don't involve 3rd-person imperatives, since I think they'd come up with interesting arguments about that, and so far I have failed. Que is the general subordinator, appearing in noun clauses (subject, object or else: que, de que, a que, en que), relative clauses (que, lo que, por el que, en el que) and adverbial clauses (a la vez que, mientras que, después de que), but sentence-initially it also marks 3rd-person imperatives, and also, of my interest:

- Complaints (with no imperative component at all): Que nos haya pasado esto, con todo lo que nos esforzamos... 'It's depressing that this happened to us, considering how much effort we put into it...'
- Conditionals (has to be sentence-initial unlike other constructions of conditions, and is also intonationally quite a bit separate but IMO not completely): ¿Que todavía no está listo?, pues apurémoslo. 'So he's not ready? Let's hurry him up. ~ If he's really not ready, let's hurry him up.' Que se lo crea, que no se lo crea, igual me da. 'Whether she believes it or not, anyway I don't care.'
- Summaries (often with an introductory adverbial): Es decir, que nos hacen falta ingredientes '(after a long rant) In other words, we're basically missing ingredients.'
- To emphasize that something is true: —Sí. —No. —¡Que sí! —¡Que no! "Yes." "No." "Yes, really!" "No, really!" —Los compré. —No, se te olvidaron. —¡Que (sí) los compré! "I bought them." "No, you forgot about them." "I really bought them!"
- As a que, to try to convince (including of insults): A que te termina gustando 'I'm telling you, you'll end up liking it', ¿A que sabe increíble? 'Do you see, it tastes amazing!' ¡A que eres idiota! 'I'm telling you, you're such an idiot!'

The RAE does talk about this in its grammar. I actually just checked one of their works to remember all these uses.


BTW, I found your comment and zompist's comment about the usefulness of a framework helpful. Also, the list you linked to said something about the potential great scope of quantifiers, which also reminded me of something nice about Classical Chinese 皆, 或 and 莫 I should've mentioned in this other thread...

[bradrn]

I don't know about Kayne's book (I just saw it's a nice long one, 490 pp.), but e.g. a few times I've tried to find discussions by Chomskians on the Spanish sentence-initial que, specifically in the uses that don't involve 3rd-person imperatives, since I think they'd come up with interesting arguments about that, and so far I have failed. Que is the general subordinator, appearing in noun clauses (subject, object or else: que, de que, a que, en que), relative clauses (que, lo que, por el que, en el que) and adverbial clauses (a la vez que, mientras que, después de que), but sentence-initially it also marks 3rd-person imperatives, and also, of my interest:

- Complaints (with no imperative component at all): Que nos haya pasado esto, con todo lo que nos esforzamos... 'It's depressing that this happened to us, considering how much effort we put into it...'
- Conditionals (has to be sentence-initial unlike other constructions of conditions, and is also intonationally quite a bit separate but IMO not completely): ¿Que todavía no está listo?, pues apurémoslo. 'So he's not ready? Let's hurry him up. ~ If he's really not ready, let's hurry him up.' Que se lo crea, que no se lo crea, igual me da. 'Whether she believes it or not, anyway I don't care.'
- Summaries (often with an introductory adverbial): Es decir, que nos hacen falta ingredientes '(after a long rant) In other words, we're basically missing ingredients.'
- To emphasize that something is true: —Sí. —No. —¡Que sí! —¡Que no! "Yes." "No." "Yes, really!" "No, really!" —Los compré. —No, se te olvidaron. —¡Que (sí) los compré! "I bought them." "No, you forgot about them." "I really bought them!"
- As a que, to try to convince (including of insults): A que te termina gustando 'I'm telling you, you'll end up liking it', ¿A que sabe increíble? 'Do you see, it tastes amazing!' ¡A que eres idiota! 'I'm telling you, you're such an idiot!'

The RAE does talk about this in its grammar. I actually just checked one of their works to remember all these uses.

That’s an interesting one! I can see how imperative que → emphasis, attempt to convince, summaries, but I really can’t see where any of those other meanings came from.

Now that I think about it, English must can be used in some similar ways:

Imperative: You must go now.
Emphasis: You must be right then!
But I don’t believe it has any of the other uses.

[zompist]

I don't know about Kayne's book (I just saw it's a nice long one, 490 pp.), but e.g. a few times I've tried to find discussions by Chomskians on the Spanish sentence-initial que, specifically in the uses that don't involve 3rd-person imperatives, since I think they'd come up with interesting arguments about that, and so far I have failed. Que is the general subordinator, appearing in noun clauses (subject, object or else: que, de que, a que, en que), relative clauses (que, lo que, por el que, en el que) and adverbial clauses (a la vez que, mientras que, después de que), but sentence-initially it also marks 3rd-person imperatives, and also, of my interest:

- Complaints (with no imperative component at all): Que nos haya pasado esto, con todo lo que nos esforzamos... 'It's depressing that this happened to us, considering how much effort we put into it...'
- Conditionals (has to be sentence-initial unlike other constructions of conditions, and is also intonationally quite a bit separate but IMO not completely): ¿Que todavía no está listo?, pues apurémoslo. 'So he's not ready? Let's hurry him up. ~ If he's really not ready, let's hurry him up.' Que se lo crea, que no se lo crea, igual me da. 'Whether she believes it or not, anyway I don't care.'
- Summaries (often with an introductory adverbial): Es decir, que nos hacen falta ingredientes '(after a long rant) In other words, we're basically missing ingredients.'
- To emphasize that something is true: —Sí. —No. —¡Que sí! —¡Que no! "Yes." "No." "Yes, really!" "No, really!" —Los compré. —No, se te olvidaron. —¡Que (sí) los compré! "I bought them." "No, you forgot about them." "I really bought them!"
- As a que, to try to convince (including of insults): A que te termina gustando 'I'm telling you, you'll end up liking it', ¿A que sabe increíble? 'Do you see, it tastes amazing!' ¡A que eres idiota! 'I'm telling you, you're such an idiot!'

These are quite interesting, and certainly belong in a grammar, but I'm not sure a syntactician would have much to say about them. Syntactically, what I'd say is that they show is that you can start with a full COMP node in Spanish. COMP is where you usually put the subordinator, and in many languages it's required for any sort of subclause. Normally the top COMP in a sentence is blank. So in these cases at least, it contains que.

I expect you need a philologist or a pragmaticist to say more. I'm guessing that these all started as subordinate expressions ("La idea que nos haya pasado esto...", "¡te digo que sí!"), and eventually the head got omitted.

[bradrn]

"La idea que nos haya pasado esto...", "¡te digo que sí!"

Could we have an English translation of this please?

[zompist]

Islands were also a discovery of GG research I think, even if that article doesn't list them as their own item.

Yes, they were discovered by John Ross, originally presented in his PhD thesis "Conatraints on variables in syntax", 1967. This book is so influential that even though he's a different school from the (post-1970) Chomskyans, they still cite him.

I think what you're complaining of is understandable, but also completely normal in linguistics. A good theory, or merely a good descriptive device, gets adopted, yet obviously grammars got written before it was invented. People talked about sounds before IPA was invented, and wrote grammars of Basque or Hindi before morphosyntactic alignment was well studied, or about Biblical Hebrew before aspect was well catalogued.

You can write a grammar without knowing X-bar syntax, or Minimalism, or the non-Chomskyan alternatives. And often you should, because explaining them takes a long time that wouldn't be appropriate for most language learners or even scholars.

But islands are a good example of what no one discovered without the framework of GG. For those who don't know what they are, here are some examples. Let's try to apply some simple transformations to a conjunction.

You went to Yale and Oxford.
*It was Yale you went to and Oxford.
*Where did you go to and Oxford?
*Where did you go to Yale and?
*I've been to Yale, which you went to and Oxford.
*To Yale you went and Oxford.

The conjunction forms an island, and most transformations can't extract just one of the conjoints.

The really odd bit is that a constraint isn't a transformation, or even a condition on a transformation. It would be wrong-headed to fix up the four different transformations above to each deal with this constraint. Instead, we add constraints as a part of the framework.

Now, you don't have to draw trees to explain this-- though you do have to informally treat "Yale and Oxford" as a constituent.

But my point is, you don't discover islands and constraints by pre-GG methods. Ross found them because of the mental habit, formed by years of doing GG, of running any given construction through the known set of transformations, and of analyzing sentences in terms of nested constituents. I give many more examples in my book.

(Note, the term "transformation" is old-fashioned-- though Move is a transformation. But one of the faults of later Chomskyan theories, noted even by some of its adherents, is that it threw out quite a bit of what GG had discovered. Rather than explore the weirder ramifications of syntax, Chomsky instead re-interpreted and re-interpreted the same fairly small set of constructions.)

[zompist]

"La idea que nos haya pasado esto...", "¡te digo que sí!"

Could we have an English translation of this please?

Sure. "The idea that that happened to us..."; "I tell you yes!"

Translating the latter, we lose the subordinator, but it's normal in Spanish here. Also French: "Je pense que oui" (I think yes).

French has similar constructions to those Ser listed, though not identical. One that comes to mind comes from this Boulet comic, sadly untranslated. The comic refers to a poem (written by a child) that starts

"Qu'il est beau, qu'il est gentille, le Coco."
(literally) That he is beautiful, that he is nice, the Coco.

These probably fall under Ser's category of emphasis, though maybe our French natives can be more precise.

(Coco is babytalk for "egg", but Boulet interprets it as a bird.)

[priscianic]

(Hi everyone, I'm the friend akam alluded to a few posts up talking about weak crossover; just thought I'd pop in to the ZBB because this is something I know a little bit about!)

I don't know about Kayne's book (I just saw it's a nice long one, 490 pp.), but e.g. a few times I've tried to find discussions by Chomskians on the Spanish sentence-initial que, specifically in the uses that don't involve 3rd-person imperatives, since I think they'd come up with interesting arguments about that, and so far I have failed. Que is the general subordinator, appearing in noun clauses (subject, object or else: que, de que, a que, en que), relative clauses (que, lo que, por el que, en el que) and adverbial clauses (a la vez que, mientras que, después de que), but sentence-initially it also marks 3rd-person imperatives, and also, of my interest:

- Complaints (with no imperative component at all): Que nos haya pasado esto, con todo lo que nos esforzamos... 'It's depressing that this happened to us, considering how much effort we put into it...'
- Conditionals (has to be sentence-initial unlike other constructions of conditions, and is also intonationally quite a bit separate but IMO not completely): ¿Que todavía no está listo?, pues apurémoslo. 'So he's not ready? Let's hurry him up. ~ If he's really not ready, let's hurry him up.' Que se lo crea, que no se lo crea, igual me da. 'Whether she believes it or not, anyway I don't care.'
- Summaries (often with an introductory adverbial): Es decir, que nos hacen falta ingredientes '(after a long rant) In other words, we're basically missing ingredients.'
- To emphasize that something is true: —Sí. —No. —¡Que sí! —¡Que no! "Yes." "No." "Yes, really!" "No, really!" —Los compré. —No, se te olvidaron. —¡Que (sí) los compré! "I bought them." "No, you forgot about them." "I really bought them!"
- As a que, to try to convince (including of insults): A que te termina gustando 'I'm telling you, you'll end up liking it', ¿A que sabe increíble? 'Do you see, it tastes amazing!' ¡A que eres idiota! 'I'm telling you, you're such an idiot!'

These are quite interesting, and certainly belong in a grammar, but I'm not sure a syntactician would have much to say about them. Syntactically, what I'd say is that they show is that you can start with a full COMP node in Spanish. COMP is where you usually put the subordinator, and in many languages it's required for any sort of subclause. Normally the top COMP in a sentence is blank. So in these cases at least, it contains que.

I expect you need a philologist or a pragmaticist to say more. I'm guessing that these all started as subordinate expressions ("La idea que nos haya pasado esto...", "¡te digo que sí!"), and eventually the head got omitted.

There actually is somewhat of a literature within a broadly Chomskyan framework on this kind of phenomenon (you might see it called "matrix que", since it's a que appearing in a matrix/main clause); for instance, Etxepare (2007), Demonte and Fernández Soriano (2014), and Corr (2016), to name just a few. I suspect the Corr might be especially interesting to conlangers, since it looks at these kinds of "matrix complementizers" in a (micro)comparative perspective, looking at variation in different varieties of Ibero-Romance.

And there actually is quite a lot of stuff you can say about them, especially in the finer details of CP (i.e. COMP)-layer syntax, how the syntax relates to the pragmatics of speech acts (in particular whether we want to encode certain aspects of speech acts directly in the syntax), etc. Zompist is correct in that, in broad strokes, people place matrix que within the CP, but there are a number of persnickety details to work out there (of course!), especially given relatively recent proposals that there actually is or can be a lot more structure within the CP than one might think. And the general broad conclusion that this body of work argues for is that some of these uses of matrix que suggest that we do want to syntactically represent certain aspects of speech acts.

Now that I think about it, English must can be used in some similar ways:

Imperative: You must go now.
Emphasis: You must be right then!
But I don’t believe it has any of the other uses.

Though you're right that must can be thought of as bearing certain kinds of similarities to matrix que, it's fundamentally a different beast (both syntactically as well as semantically). It's a modal, and one of the characteristics of English modals (in contrast to, for instance, St’at’imcets modals) is that they're variable flavor. "Flavor" is just a cute term for "type" of modality—deontic (pertaining to orders, rules, and law), epistemic (pertaining to inference, belief, and knowledge), circumstantial (pertaining to actual circumstances and facts in the world), teleological (pertaining to goals), bouletic (pertaining to desires), etc.

Here you're noting that must has deontic readings and epistemic readings. You gloss the deontic reading as an "imperative", and indeed it does have the pragmatic whiff of an order. But I think the average semanticist would say that your example is just a declarative sentence, expressing a meaning roughly like "in light of certain orders, it is a necessity that you leave now", and the "certain orders" are contextually understood to be the speaker's orders, resulting in an imperative-like force. (It might also be worth noting that some people analyze imperatives as actually underlyingly containing a deontic modal, such that this is, very roughly, the meaning of "real" imperatives as well; e.g. Kaufmann 2012 (unfortunately not open-access, though it is on libgen).) I suspect your gloss of the second reading as "emphasis" is meant to be an epistemic reading, something like "given various premises, it's a necessity that you must be right" (generally, I think it 's wise to try to avoid the notoriously vague word "emphasis", or at least recognize that it's just a placeholder for a more precise analysis). For instance, I imagine a context where you and your interlocutor are engaged in a discussion, putting forth various kinds of arguments, and finally you accept your interlocutor's point, and want to convey that, based on their arguments, they must be right.

[bradrn]

1. You went to Yale and Oxford.
2. *It was Yale you went to and Oxford.
3. *Where did you go to and Oxford?
4. *Where did you go to Yale and?
5. *I've been to Yale, which you went to and Oxford.
6. *To Yale you went and Oxford.

Interesting, sentence 4 here isn’t entirely ungrammatical for me. (It’s not quite as grammatical as, say, ‘Where did you go to?’, but I wouldn’t describe it as completely unacceptable.) And sentence 5 becomes grammatical for me if I add a comma after to.

[zompist]

Thanks for coming by, priscianic! More syntax people are very welcome, and I hope you'll stick around!

And the general broad conclusion that this body of work argues for is that some of these uses of matrix que suggest that we do want to syntactically represent certain aspects of speech acts.

Very interesting... has this been run by Chomsky? 'Cos he really didn't like it when people started mixing in semantics and pragmatics into GG in the 1970s. That was basically why the GS crowd split off.

I'd have to see how it's proposed to do it... I used to read a lot of pragmatics, and I never saw a proposal involving syntactic trees that I liked. Kind of the weak point of syntactic frameworks is that as you add more and more stuff to them, they get opaque and less useful. The GS people were notorious for this, in fact. Cognitive linguists tend to create parallel mechanisms rather than adding to the syntactic tree.

[priscianic]

Thanks for coming by, priscianic! More syntax people are very welcome, and I hope you'll stick around!

Thanks for the warm welcome!

And the general broad conclusion that this body of work argues for is that some of these uses of matrix que suggest that we do want to syntactically represent certain aspects of speech acts.

Very interesting... has this been run by Chomsky? 'Cos he really didn't like it when people started mixing in semantics and pragmatics into GG in the 1970s. That was basically why the GS crowd split off.

Yeah; a lot of this kind of stuff is at its core a reimagining of a lot of generative semantics stuff, though (hopefully!) in a lot more sophistication. Chomsky probably wouldn't like it (but who cares what he thinks, really), even though most of this work is within a broadly Minimalist/post-GB framework, and thus could be coherently called "Chomskyan".

(On a more personal note, I'm someone who considers themselves a broadly Chomskyan syntactician who's interested in how the syntax and semantics interact, and how we get the meanings we do from the parts that they're built up from. So I really don't understand Chomsky's viewpoint, and I feel like he just doesn't understand what the point of (compositional) semantics is, to put it bluntly.)

I'd have to see how it's proposed to do it... I used to read a lot of pragmatics, and I never saw a proposal involving syntactic trees that I liked. Kind of the weak point of syntactic frameworks is that as you add more and more stuff to them, they get opaque and less useful. The GS people were notorious for this, in fact. Cognitive linguists tend to create parallel mechanisms rather than adding to the syntactic tree.

As you note, there are at least two kinds of reactions to assimilating new data into the model: complicate the model, or create new mechanisms. I'd say that the majority of the field is content with having the pragmatics stay firmly within pragmatics territory, using pragmatic tools and mechanisms, without reincarnating the generative semantics performative hypothesis and complicating the syntactic model.

The "speech act syntacticians" like to point to various kinds of syntactic phenomena that seem to be sensitive to speech-act-related material, like allocutive agreement, things like logophors and egophors, speech-act-specific word order, embedded speech acts, etc., in order to justify positing various kinds of "speech act projections" in the syntax. If you take a kind of modular approach to the architecture of the grammar, such that syntactic operations work strictly within the the syntax module, then the fact that syntax is sensitive to "speech act stuff" suggests that "speech act stuff" has to present and visible in the syntax. Of course, there's a lot of room to disagree here, and plenty of people do; for instance, you can dispute modularity, or you can dispute that the relevant phenomena really are syntactic in nature.

If you're interested in learning more about this side of things, Speas and Tenny (2003) is a classic foundational paper in this field, and Zu (2015, 2018) is a more modern reworking of Speas and Tenny's original proposals. (There are some pdf glitches in the Speas and Tenny paper I linked; if you want a cleaner version, it's in a book called Asymmetry in Grammar, which is on libgen iirc.)

If you're proficient in the formalism of basic truth-conditional semantics, there's also a literature on encoding speech acts/sentential force directly in the syntax via speech act/sentential force operators, and arguing for this kind of thing on the basis of semantic (and pragmatic) data. Sven Lauer has a presentation that presents the relevant issues in this domain pretty cogently, imho (though he is biased; he's a part of the "make pragmatics pragmatics again!" camp, and wants to argue against the syntacticization of sentential force).

[cedh]

I don't know about Kayne's book (I just saw it's a nice long one, 490 pp.), but e.g. a few times I've tried to find discussions by Chomskians on the Spanish sentence-initial que, specifically in the uses that don't involve 3rd-person imperatives, since I think they'd come up with interesting arguments about that, and so far I have failed. Que is the general subordinator, appearing in noun clauses (subject, object or else: que, de que, a que, en que), relative clauses (que, lo que, por el que, en el que) and adverbial clauses (a la vez que, mientras que, después de que), but sentence-initially it also marks 3rd-person imperatives, and also, of my interest:

- Complaints (with no imperative component at all): Que nos haya pasado esto, con todo lo que nos esforzamos... 'It's depressing that this happened to us, considering how much effort we put into it...'
- Conditionals (has to be sentence-initial unlike other constructions of conditions, and is also intonationally quite a bit separate but IMO not completely): ¿Que todavía no está listo?, pues apurémoslo. 'So he's not ready? Let's hurry him up. ~ If he's really not ready, let's hurry him up.' Que se lo crea, que no se lo crea, igual me da. 'Whether she believes it or not, anyway I don't care.'
- Summaries (often with an introductory adverbial): Es decir, que nos hacen falta ingredientes '(after a long rant) In other words, we're basically missing ingredients.'
- To emphasize that something is true: —Sí. —No. —¡Que sí! —¡Que no! "Yes." "No." "Yes, really!" "No, really!" —Los compré. —No, se te olvidaron. —¡Que (sí) los compré! "I bought them." "No, you forgot about them." "I really bought them!"
- As a que, to try to convince (including of insults): A que te termina gustando 'I'm telling you, you'll end up liking it', ¿A que sabe increíble? 'Do you see, it tastes amazing!' ¡A que eres idiota! 'I'm telling you, you're such an idiot!'

These are quite interesting, and certainly belong in a grammar, but I'm not sure a syntactician would have much to say about them. Syntactically, what I'd say is that they show is that you can start with a full COMP node in Spanish. COMP is where you usually put the subordinator, and in many languages it's required for any sort of subclause. Normally the top COMP in a sentence is blank. So in these cases at least, it contains que.

I expect you need a philologist or a pragmaticist to say more. I'm guessing that these all started as subordinate expressions ("La idea que nos haya pasado esto...", "¡te digo que sí!"), and eventually the head got omitted.

Some linguists have used the term insubordination to describe a process of grammaticalizing subordinated clauses into markers of other things. Maybe this word is something to look out for if you go searching for phenomena similar to Spanish matrix que crosslinguistically.

[Ares Land]

French has similar constructions to those Ser listed, though not identical. One that comes to mind comes from this Boulet comic, sadly untranslated. The comic refers to a poem (written by a child) that starts

"Qu'il est beau, qu'il est gentille, le Coco."
(literally) That he is beautiful, that he is nice, the Coco.

These probably fall under Ser's category of emphasis, though maybe our French natives can be more precise.

(Coco is babytalk for "egg", but Boulet interprets it as a bird.)

Yes, it's used for emphasis. I'd translate as something like 'How beautiful! How nice' or maybe 'He's so beautiful, he's so nice' would be better, or 'What a beautiful bird! What a nice bird!'

An another similar construction is 'Oh que oui! Oh que non! Oh que si!' ~ 'Oh yes, oh no, oh yes it is/it does' and 'Que' introducing a clause with the main verb in the subjunctive: 'Que vos deux maisons soient maudites !' ('A plague on both your houses!').
All of these (and Sér's Spanish examples too) could be plausible evolutions of the frightening range of meaning of Latin ut, a conjunction that was replaced by quid (sic in the Balkans) at some point in Vulgar Latin.

Which leaves of course open the whole question of the usage of utin Latin. Does anyone have a good reference for Latin syntax?


('coco' is kind of fun. I'm aware of 'coco' meaning 'egg' because for unclear reasons it was used a lot in the French dub of a Clockwork Orange. I believe it's a little dated. Other meanings include 'commie', 'coke', 'pal', the stereotypical name for a parrot, and a general kind of affectionate nickname. Oh, and babies say that a lot, but I suppose that last one's cross-linguistic!)

[Ares Land]

Oh, while I'm here, is anyone aware of any resources on syntax of morphologically complex languages? My specific issue is that in researching Iroquoian languages, it became quite clear that they're most definitely not all morphology and no syntax... but reference grammars understandably focus on morphology. So ideally, it's Iroquoian syntax I'd be looking for, but I'm more than willing to settle on a good resource for any other comparably morphology-heavy language...

[priscianic]

Oh, while I'm here, is anyone aware of any resources on syntax of morphologically complex languages? My specific issue is that in researching Iroquoian languages, it became quite clear that they're most definitely not all morphology and no syntax... but reference grammars understandably focus on morphology. So ideally, it's Iroquoian syntax I'd be looking for, but I'm more than willing to settle on a good resource for any other comparably morphology-heavy language...

There's a pretty decent literature on the syntax of Algonquian and Iroquoian languages in the more theoretical/formal literature (more so for Algonquian, but I might be biased because I'm more familiar with Algonquian). Some relevant names include Amy Dahlstrom, Benjamin Bruening, Heather Bliss, and Will Oxford (among many others) for Algonquian, and Michael Barrie for Iroquoian. All these people have links to pdfs of their papers on their websites, so you can spend a good afternoon skimming through them. (The disclaimer is that I'm a syntactician and not an Algonquianist/Iroquoianist, so I probably am missing a bunch of relevant material that's more descriptive/documentary in nature.)

[Kuchigakatai]

(Coco is babytalk for "egg", but Boulet interprets it as a bird.)

I think it's supposed to be reminiscent of coucou 'cuckoo bird', while still using it as a babytalk word for egg for its/his round shape.

There actually is somewhat of a literature within a broadly Chomskyan framework on this kind of phenomenon (you might see it called "matrix que", since it's a que appearing in a matrix/main clause); for instance, Etxepare (2007), Demonte and Fernández Soriano (2014), and Corr (2016), to name just a few. I suspect the Corr might be especially interesting to conlangers, since it looks at these kinds of "matrix complementizers" in a (micro)comparative perspective, looking at variation in different varieties of Ibero-Romance.

Thank you for that!

By the way, there is a long tradition here... dating to 15 years ago or so... of welcoming people with images of some pickles and tea... I guess that tradition is getting well-worn and tired now, seeing that no one has yet done it. (!)

Another by the way, priscianic: have you read the final two books Priscian's grammar of Latin discussing syntax?

An another similar construction is 'Oh que oui! Oh que non! Oh que si!' ~ 'Oh yes, oh no, oh yes it is/it does' and 'Que' introducing a clause with the main verb in the subjunctive: 'Que vos deux maisons soient maudites !' ('A plague on both your houses!').
All of these (and Sér's Spanish examples too) could be plausible evolutions of the frightening range of meaning of Latin ut, a conjunction that was replaced by quid (sic in the Balkans) at some point in Vulgar Latin.

Although the use of que to express wishes in noun clauses (je veux que tu viennes) pretty much descends from the replacement of Classical ut with quod as Late Latin progressed and later still quid, Latin used bare subjunctives, not ut in the main clause to express 3rd-person imperatives. You may be familiar with this from the Lord's Prayer: sanctificētur (may it be blessed) nōmen tuum, adveniat (may it come) regnum tuum, fiat (may it be done) voluntās tua sīcut in caelō et in terrā. So this use at least seems to have involved adding a que at the beginning in analogy with other uses of the subjunctive...