The Sinitic Thread

[dhok]

Some of you might know that I'm doing a master's at Vienna in historical linguistics for which I have to learn intensive German. What you might not know is that being stuck in intensive German classes makes it really difficult to socialize with locals or find language-oriented friends (because all my classmates are Eastern Euros who are busy working). Also, a lot of stuff in linguistics is published in Mandarin these days, and not very many Westerners can read it. Also, I've gotten much better at learning vocab recently with strategies like word-lists and Anki. Also, I kind of didn't work very hard to learn Mandarin when I was in China and I regret that.

So for all these reasons, I've decided to get back up on the Mandarin pony! It'll go slow and take a back seat to German until I start the actual master's program, but after that's done with and my German's at a high level I'll have more breathing room. The goal is ~B1-B2 over the next three years or so. (Of course, I could just drop it and leave this thread to die, but let's be optimistic).

I've started working through New Practical Chinese Reader--thankfully in adulthood I have the patience and energy to do dialogue repetition and exercises to an extent that wasn't really possible for me as a teenager or even an undergrad. This thread is intended as half-learning diary, half interesting-things-I-learn-along-the-way, half discussion of anything in Sinitic linguistics, so please feel free to derail.

There isn't really a whole lot to report so far, other than a reminder that Mandarin (all of Sinitic? can anybody who knows Canto chime in?) adjectives are really just stative verbs. There are, afaik, virtually no real tests to distinguish the two that aren't strongly contrived.

(Also, a note for language-learners...the human brain is really bad at remembering additional information about a word or syllable like gender or tone. Write das Wasser and you'll be trying to remember a week later whether it's der Wasser or das Wasser; and if you write down and drill 好 hǎo, at least initially you're going to have trouble remembering which tone it is. In my experience, the best way to mark this sort of arbitrary info is color; Wasser is a lot easier to remember the gender of--it's a green word, so it must be neuter--and similarly if you write down and drill 好 hǎo you'll have an easier time with the tone. (Following the Pleco app, I use red for tone 1, green for tone 2, blue for tone 3 and purple for tone 4, with black marking neutrals: 一二三四五.)

[Vijay]

Mandarin Chinese is one of my strongest languages, but I've never read anything more than short passages in it. I have Practical Chinese Reader volumes I through V (had to use I-II for Chinese classes in college and bought IV-V at Half Price Books and then III through Amazon). I've read the first three volumes and part of the fourth, but I feel I forgot too much of the relevant vocabulary from volumes III and IV. I have a small collection of books in Chinese, too, so I've been trying to use those as well as whatever I can use online.

My next-door neighbors speak Teochew (and Vietnamese), and one guy I know online comes from a family in southern Zhejiang and speaks Qingtianese and Wenzhounese, two southern varieties of Wu. He says that in countries that don't have large Chinese immigrant populations, most of the Chinese people who do live there are also from southern Zhejiang (but for all I know, he may have been thinking specifically of Europe).

[Richard W]

There isn't really a whole lot to report so far, other than a reminder that Mandarin (all of Sinitic? can anybody who knows Canto chime in?) adjectives are really just stative verbs. There are, afaik, virtually no real tests to distinguish the two that aren't strongly contrived.

I suspect its areal.

Thai is similar. One might think that the two different prefixes for forming abstract nouns indicated the difference, but some words have both forms, indicating that the difference is semantic. Thai dictionaries may mark the difference, but I have a suspicion that they're basing it on English translations of the words. Mind you, I'm not convinced that being stative is a lexical rather than semantic difference in English, so my view may be suspect.

(Also, a note for language-learners...the human brain is really bad at remembering additional information about a word or syllable like gender or tone. Write das Wasser and you'll be trying to remember a week later whether it's der Wasser or das Wasser; and if you write down and drill 好 hǎo, at least initially you're going to have trouble remembering which tone it is. In my experience, the best way to mark this sort of arbitrary info is color; Wasser is a lot easier to remember the gender of--it's a green word, so it must be neuter--and similarly if you write down and drill 好 hǎo you'll have an easier time with the tone. (Following the Pleco app, I use red for tone 1, green for tone 2, blue for tone 3 and purple for tone 4, with black marking neutrals: 一二三四五.)

Another suggestion for Chinese tone was that using Gwoyeu Romatzyh (GR) might help. The gymnastics of converting from pinyin to GR might also help force it into one's memory.

[Richard W]

There isn't really a whole lot to report so far, other than a reminder that Mandarin (all of Sinitic? can anybody who knows Canto chime in?) adjectives are really just stative verbs. There are, afaik, virtually no real tests to distinguish the two that aren't strongly contrived.

I suspect it's areal.

Thai is similar. One might think that the two different prefixes for forming abstract nouns indicated the difference, but some words have both forms, indicating that the difference is semantic. Thai dictionaries may mark the difference, but I have a suspicion that they're basing it on English translations of the words. Mind you, I'm not convinced that being stative is a lexical rather than semantic difference in English, so my view may be suspect.

(Also, a note for language-learners...the human brain is really bad at remembering additional information about a word or syllable like gender or tone. Write das Wasser and you'll be trying to remember a week later whether it's der Wasser or das Wasser; and if you write down and drill 好 hǎo, at least initially you're going to have trouble remembering which tone it is. In my experience, the best way to mark this sort of arbitrary info is color; Wasser is a lot easier to remember the gender of--it's a green word, so it must be neuter--and similarly if you write down and drill 好 hǎo you'll have an easier time with the tone. (Following the Pleco app, I use red for tone 1, green for tone 2, blue for tone 3 and purple for tone 4, with black marking neutrals: 一二三四五.)

Another suggestion for Chinese tone was that using Gwoyeu Romatzyh (GR) might help. The gymnastics of converting from pinyin to GR might also help force it into one's memory.

[Kuchigakatai]

There isn't really a whole lot to report so far, other than a reminder that Mandarin (all of Sinitic? can anybody who knows Canto chime in?) adjectives are really just stative verbs. There are, afaik, virtually no real tests to distinguish the two that aren't strongly contrived.

There remains an important distinction between stative verbs and action verbs nevertheless, and there are not many stative verbs that aren't adjective-ish.

I suspect its areal.

Thai is similar. One might think that the two different prefixes for forming abstract nouns indicated the difference, but some words have both forms, indicating that the difference is semantic. Thai dictionaries may mark the difference, but I have a suspicion that they're basing it on English translations of the words. Mind you, I'm not convinced that being stative is a lexical rather than semantic difference in English, so my view may be suspect.

I can't comment on Thai, but when it comes to stative verbs in English, I suppose it depends on what a linguist means by that. I've sometimes heard that those semantically stative-ish verbs that can't normally take the progressive construction except for some limited circumstances / meanings are the stative verbs of the language (e.g. "be", "want", "love", "know"). But I'm sure some linguists or grammar commentators talk about "English stative verbs" from a semantic point of view.

[akam chinjir]

There isn't really a whole lot to report so far, other than a reminder that Mandarin (all of Sinitic? can anybody who knows Canto chime in?) adjectives are really just stative verbs. There are, afaik, virtually no real tests to distinguish the two that aren't strongly contrived.

There remains an important distinction between stative verbs and action verbs nevertheless, and there are not many stative verbs that aren't adjective-ish.

I was reading a bit about hěn 很 yesterday. There are contexts in which a bare gradable adjective gets a comparative sense if it's not accompanied by a degree word, most often hěn (so Zhāngsān gāo 張三高 is ill-formed or, given the right context, means Zhangsan is taller; zhāngsān hěn gāo 張三 很高 is Zhangsan is tall). You don't get the same effect with stative verbs like xǐhuān 喜歡 like, love. Meanwhile nongradable adjectives mostly can't be used as predicates, as far as I know, so obviously they're not verbs.

(There are adjectives like bīnglěng 冰冷 ice cold that are ungradable but can be used as predicates; many such adjectives are compounds in which the first element---like bīng 冰 ice here---could be thought of as a degree word. Are there any predicable nongradable adjectives that aren't like this?)

There's also an argument that adjectives and verbs behave differently in prenominal modification (Paul, Adjectival modification in Mandarin and related issues).

Are there any intransitive stative verbs that aren't adjectives?

I'm pretty sure exactly the same issues arise in Cantonese.

[dhok]

A quick note before I dash off to a busy day--one of the reasons I'm skeptical of inventing tests like that to distinguish verb and adjective is that there's something sort of Western European-centric about it in terms of grammar analysis. Modern linguistics grows mostly out of the Latinate grammatical tradition, and even though we've gotten pretty good about analyzing non-Latinate languages on their own terms, I think there's a sense in which we're "primed" to look for a verb-adjective distinction--because the distinction is so robust in Indo-European, Finno-Ugric and Semitic languages (in all three, adjective morphology looks like noun morphology). So even when both "verbs" and "adjectives" can stand alone as predicates, form relative clauses or become attributive with 的 de, and so forth, we really really want a distinction to exist.

"But syntactic/pragmatics tests!" Yes; but by the same token you can find other basic lexical classes elsewhere that you hadn't thought to look for. Plenty of languages distinguish transitive and intransitive verbs with different morphology, do not allow transitive verbs to be used without an object, require a causative to use an intransitive in an intransitive manner, and so on and so forth. It's quite easy to create syntactic tests to distinguish them. But we don't usually consider them different lexical classes, perhaps (but not necessarily) because the languages spoken in and near Europe are quite fuzzy on the distinction. It's not clear to me that the adjective-verb distinction is significantly sharper or stricter in the head of a Mandarin speaker than the transitivity distinction is in a language where transitives and intransitives are kept distinct.

[akam chinjir]

But when the tests work...

So even when both "verbs" and "adjectives" can stand alone as predicates, form relative clauses or become attributive with 的 [b]de[/b]

But, as I said, not all Mandarin adjectives can stand alone as predicates, and many get a comparative sense when they do, not at all verby behaviour. And it's really common to have attributive adjectives without 的, and adjective+的 phrases are often syntactically distinct from relative clauses (e.g., in the need for a degree word to avoid a comparative sense).

Tangential question: do verbs have anything analogous to the intersective/non-intersective distinction you get with adjectives? (E.g. the difference in interpretation of lǎo 老 old in lǎo péngyǒu 老朋友 old (=longtime) friend vs hěn lǎo de péngyǒu 很老的朋友 old (=aged) friend, friend who is old.)

[Vijay]

Zhāngsān gāo 張三高 is ill-formed or, given the right context, means Zhangsan is taller

I found this, but I haven't been able to tell yet whether it agrees with you or not.

[KathTheDragon]

But when the tests work...

Then you're simply left with the philosophical problem of distinguishing a pair of genuinely distinct categories from a pair of super-category and sub-category from a pair of sub-categories of some other super-category

[dhok]

So this is not really a thought about Chinese, but it is one inspired by Chinese.

The lack of inflectional morphology in Mandarin (or any variety of modern Chinese, really) means that L2 learners spend most of their "grammar" time learning syntactic transformations. IIRC this is basically the approach used by Thompson and Li. E.g., on day 1 of Mandarin class you learn the ur-structure

S V (O)

and then on day 2, you learn a structure for yes-no questions, a transformation of the basic structure:

S V (O) -> S V (O) ma

I've been dabbling with Haskell a bit recently, and this sort of reminds me of a Haskell function. The learner takes some basic ur-structure, such as that for basic subject-verb-object sentences. A transformation to the components is applied; the semantic meaning of the transformation is essentially arbitrary, inferred from context. Functions are recursive.

In other words, I sort of wonder how much "inner structure" of the sort that Chomskyans like to construct is really there. Children learn some basic structure like "I go" or "Mama eat". As time goes on, they learn to apply more transformations, adding objects, questions, subordinating clauses. The semantics of these transformations, which might be morphology as well as syntax, can essentially be a blank slate--what they do or mean is up for grabs, constrained only by human cognitive quirks, working memory load, and the existence of a diachronic pathway to some pattern. (E.g., there is no language--that I know of--where subject-verb inversion marks negatives; but there probably isn't any reason that one couldn't exist, except that there's no clear diachronic pathway to get there.)

(an example of a cognitive-quirk constraint: the way humans operate in the world seems to mandate or at least very strongly encourage physical objects in the world acting as arguments and physical actions as predicates. Sensory attributes like "green" or "tasty" seem to be up for grabs. There is no language where the equivalent of "there's a green bottle on the table" looks like "a bottle-ish green tables upon-ly". You could maybe do something like that as a literary experiment, but it doesn't line up with the way humans experience the world so no language operates that way.)

[bradrn]

So this is not really a thought about Chinese, but it is one inspired by Chinese.

The lack of inflectional morphology in Mandarin (or any variety of modern Chinese, really) means that L2 learners spend most of their "grammar" time learning syntactic transformations. IIRC this is basically the approach used by Thompson and Li. E.g., on day 1 of Mandarin class you learn the ur-structure

S V (O)

and then on day 2, you learn a structure for yes-no questions, a transformation of the basic structure:

S V (O) -> S V (O) ma

I've been dabbling with Haskell a bit recently, and this sort of reminds me of a Haskell function. The learner takes some basic ur-structure, such as that for basic subject-verb-object sentences. A transformation to the components is applied; the semantic meaning of the transformation is essentially arbitrary, inferred from context. Functions are recursive.

In other words, I sort of wonder how much "inner structure" of the sort that Chomskyans like to construct is really there. Children learn some basic structure like "I go" or "Mama eat". As time goes on, they learn to apply more transformations, adding objects, questions, subordinating clauses. The semantics of these transformations, which might be morphology as well as syntax, can essentially be a blank slate--what they do or mean is up for grabs, constrained only by human cognitive quirks, working memory load, and the existence of a diachronic pathway to some pattern. (E.g., there is no language--that I know of--where subject-verb inversion marks negatives; but there probably isn't any reason that one couldn't exist, except that there's no clear diachronic pathway to get there.)

I’m not a syntactician by any means, but how exactly does this proposal differ from the usual Chomskyan idea of a ‘deep structure’ (in this case, S V (O)) which then undergoes various transformations? Or am I misunderstanding what Chomskyan syntax is about? (The latter is most probable; I don’t know all that much about syntax.)

[dhok]

The Chomskyans propose all sorts of rather baroque tree structures for most syntax. What I'm saying is that I'm not sure most syntactic constructions have anything other than what they say on the tin. ma doesn't need to be assigned to any sort of question node; it's just there, derived mechanically from the ur-structure SV(O).

[bradrn]

The Chomskyans propose all sorts of rather baroque tree structures for most syntax. What I'm saying is that I'm not sure most syntactic constructions have anything other than what they say on the tin. ma doesn't need to be assigned to any sort of question node; it's just there, derived mechanically from the ur-structure SV(O).

I completely agree with you on this! As I mentioned already, I don’t know much about Chomskyan syntax, but every time I see one of those baroque structures, I end up wondering how anyone seriously thinks this is what language really is!

On the other hand, the Chomskyanist approach does explain many things quite neatly. For instance, the syntax of adjectives and other noun phrases can be explained by making them children of NP (or DetP), and then adding an NP node everywhere within a sentence where a noun can appear. But in your theory as you’ve presented it, you would have to have several different operations, one for each part of the sentence: for instance, adding an adjective to an S and adding an adjective to an (O) would be different operations. (But I can see at least two ways to resolve this, so it’s not that much of an issue.)

[zompist]

In other words, I sort of wonder how much "inner structure" of the sort that Chomskyans like to construct is really there. Children learn some basic structure like "I go" or "Mama eat". As time goes on, they learn to apply more transformations, adding objects, questions, subordinating clauses. The semantics of these transformations, which might be morphology as well as syntax, can essentially be a blank slate--what they do or mean is up for grabs, constrained only by human cognitive quirks, working memory load, and the existence of a diachronic pathway to some pattern. (E.g., there is no language--that I know of--where subject-verb inversion marks negatives; but there probably isn't any reason that one couldn't exist, except that there's no clear diachronic pathway to get there.)

You've mostly reinvented Construction Grammar here. Which is totally a thing; see Adele Goldberg or Michael Tomasello for more.

The analogy with computer languages is good too, not least because there is obviously not just one way to make a computer language, or write an algorithm. Writing my syntax book, I came to think that's one of the things Chomsky gets wrong. He thinks that because he has this huge formulation that (perhaps) explains syntax, it must be the right theory. It's exactly like thinking that only Pythagoras's method can come up with Pythagoras's theorem.

(I think there's a lot more to say for generative grammar... if everything was as simple as 吗, we wouldn't need it. But not even everything in Mandarin is that simple!)

[Moose-tache]

This isn't exactly a new complaint, but this is what I see every time I read something from the GG people:



I get that they've come up with a lot of wonderful insights into the nature of language and logic, but they can't possibly think that their analysis bears any resemblance to the way humans actually process language.

[Frislander]

One of the things that's really driving me up the wall with the syntax paper I'm taking this year (in addition to all the other stuff mentioned above) is that these people seem to have no actual clue what they're describing. Like for instance I've been subjected to several lectures about how "phrase structure is more important than grammtical relations", with the obvious straw-manning of what a theory of grammatical relations would actually involve, but also the following inconsistency.

So the theories as calibrated seem to focus about drawing up syntax trees which explain the surface structure of a language like English, and they try and argue that stuff like head movement constraints is somehow evidence that it's specifically phrase structure that's what's relevant here. And that sort of works because the tree is here effectively a representation of how the surface structure is derived. But then they come to a language that's non-configurational to some extent and then they claim that this is derived by "scrambling", and just state that it's still phrase structure that's important over grammatical relations, when as far as I'm concerned they're just the same thing once you abstracted away from the actual surface structure. It doesn't help that they couldn't be bothered to define what makes some aspect of syntax specifically a feature of phrase-structure as opposed to a kind of grammatical relation - the kind of straw-man theory of grammatical relations being used seemed to imply that for generativists a theory of "grammatical relations" is concerned purely with notions of subject vs. object (i.e. theta-roles), but I don't see, once you've abstracted away from the surface structure by scrambling, how notions such as c-command couldn't be considered a kind of grammatical relation as well.

Also because the relationship to actual morphology is tangential at best you end up with a situation where a morphological marker is not doing a job of marking a category in and of itself, but instead is a marker of the entire word having moved to a position in the tree that hosts this marker, and in the case of non-configurational languages it's then scrambled to move it out of the position it's in. One can only imagine the ridiculous amount of movement that must be required to properly derive a polysynthetic structure.

[vegfarandi]

There isn't really a whole lot to report so far, other than a reminder that Mandarin (all of Sinitic? can anybody who knows Canto chime in?) adjectives are really just stative verbs. There are, afaik, virtually no real tests to distinguish the two that aren't strongly contrived.

There remains an important distinction between stative verbs and action verbs nevertheless, and there are not many stative verbs that aren't adjective-ish.

I recommend this Haspelmath paper on "Escaping ethnocentricism in the study of word-class universals" https://www.academia.edu/2244681/Escapi ... universals

[Qwynegold]

(Also, a note for language-learners...the human brain is really bad at remembering additional information about a word or syllable like gender or tone. Write das Wasser and you'll be trying to remember a week later whether it's der Wasser or das Wasser; and if you write down and drill 好 hǎo, at least initially you're going to have trouble remembering which tone it is. In my experience, the best way to mark this sort of arbitrary info is color; Wasser is a lot easier to remember the gender of--it's a green word, so it must be neuter--and similarly if you write down and drill 好 hǎo you'll have an easier time with the tone. (Following the Pleco app, I use red for tone 1, green for tone 2, blue for tone 3 and purple for tone 4, with black marking neutrals: 一二三四五.)

Huh, I'll have to try that. Thanks!

[akam chinjir]

I recommend this Haspelmath paper on "Escaping ethnocentricism in the study of word-class universals" https://www.academia.edu/2244681/Escapi ... universals

An alternative link for the Haspelmath paper, for anyone who doesn't want to deal with academia.edu.