Linguistic Miscellany Thread

[zompist]

That's my point: I wish they wouldn't. It's presumably for legal reasons: Coke would lose its trademark if "coke" was accepted to be a generic term, so they come down hard on the restaurant biz.

(I think I could tell the difference, but it's not big enough to care about.)

[Space60]

Before "game console" existed, people used "Nintendo" as a generic term for game consoles. Nintendo didn't like this and so they coined "game console" to encourage people not to use "Nintendo" generically.

[alynnidalar]

That's my point: I wish they wouldn't. It's presumably for legal reasons: Coke would lose its trademark if "coke" was accepted to be a generic term, so they come down hard on the restaurant biz.

(I think I could tell the difference, but it's not big enough to care about.)

I'm pretty sure they ask because people do actually taste the difference, not for legal reasons... or at least I'm pretty sure Random Diner #837 isn't all that concerned about being sued by Coca-Cola, yet they still ask.

[Pabappa]

Coca-Cola used to send people out to restaurants and order "coke" and see if they got something else. It was real. I'm not sure if they had the ability back then to know which restaurants Coca-Cola sold to ahead of time so they could move immediately to legal threats.

[Zju]

How do you parse Yangtze river dolphin? [Yangtze river] dolphin or Yangtze [river dolphin]?

[mèþru]

#2

[Travis B.]

#2

Same here.

[Pabappa]

How do you parse Yangtze river dolphin? [Yangtze river] dolphin or Yangtze [river dolphin]?

I dont think I'd bracket this at all, but given the choice of those two interpretations, I'd pick #1. Yes, some dolphins live in freshwater and can be called river dolphins. But the phrase "Yangtze River" has much more salience, .... its not like there's a Yangtze Mountain or Yangzte Sea.... so i prefer to group it the first way.

[Zju]

I dont think I'd bracket this at all

I thought that all groups of words, of length 3 or more, have an underlying binary tree structure, whose nodes can be a single word, a branching point, or the null set? At least according to some theories of syntax. Not very well versed in any syntax theory, though.

[Pabappa]

I dont think I'd bracket this at all

I thought that all groups of words, of length 3 or more, have an underlying binary tree structure, whose nodes can be a single word, a branching point, or the null set? At least according to some theories of syntax. Not very well versed in any syntax theory, though.

Ive never studied linguistics formally, so Im not well versed in it at all.

Perhaps what I call unbracketed is functionally equivalent to [Yangtze [river [dolphin]]], ..... or maybe [[[Yangtze] river] dolphin] .... but since I can't decide which of those two makes more sense I'm going to just stick with saying it's unbracketed.

[mèþru]

Similarly
I think of [New York (City)] [city bus], not [New York City] bus

[Travis B.]

Similarly
I think of [New York (City)] [city bus], not [New York City] bus

In the case of Yangtze river dolphin I stress river; likewise, in New York City bus I stress City.

[zyxw59]

If it's a type of (river) dolphin specific to the Yangtze, it's a [Yangtze] river dolphin.

If it's a dolphin that just happens to have been found in the Yangtze, it's a [Yangtze River] dolphin.

A [New York] [city bus] could be any city bus in the state of New York (but still probably in NYC); a [New York City] bus is specifically one in NYC, and is less marked for me (maybe helped by the fact that for me, "bus" defaults to a city bus, as opposed to an intercity or school bus)

[Vijay]

These are the lyrics of a Tamil movie song I posted not so long ago. I tried to write out the actual Tamil words in Tamil script (including onomatopeia), write out the English words in Roman script, and transcribe the nonsense words in IPA. Something like half of it is just English:

கண்ணே, I love you!
என் காதே, I love you!
எனக்காக எல்லாம் செய்யும்
என்தன் கையே, I love you!
மழையே, I love you!
பூ, இலையே, I love you!
என் உள்ளே என்றும் பொங்கும்
உட்சாகமே, I love you!
என்றும் என்றும் enjoy!
உள்ளம் எங்கும் enjoy!
எந்த நாளும் உட்சாகமாய் life enjoy!
Morning முதல் evening
Jolly-ஆன feeling,
இளமைக்கு எப்போதுமே புது meaning.

Chorus:
[t̪əkəd̪imit̪a]...
[t̪əkəd̪imit̪a], இந்த life என்றும் jolly தான்.
[t̪əkəd̪imit̪a], இந்த jolly-க்கு இல்லை வேலிதான்.

Enjoy the வாழ்க்கை! (2)
Boyfriend தான் Coca-Cola;
Girlfriend தான் Pepsi-Cola.
ரெண்டுக்கும் டிஷ்யூம் டிஷ்யூம் தான்;
அப்பப்ப kissing-kissing தான்.
[he] cool, cool, cool-ன்னு Fair & Lovely
Face-இல் பூசிக்கோ!
[he] ஜிகு ஜிகு ஜிகுன்னு Chennai Silks-இன்
சேலை காட்டிக்கோ!
Airtel, Aircel, RPG, BPL,
நீ சும்மா பேசு bill-ஐ உன்தன்
Boyfriend காட்டட்டும்!
(Chorus)

அம்மாவின் தோசை, office-இன் பாஷை,
ரெண்டுக்குள் சிக்கிக்கொண்டேன்.
அதிலே ஓர் இன்பம் கண்டேன்.
அப்பாவின் savings எனக்கு.
அவருக்கு pension இருக்கே!
நான் coffee போட்டா ஏனோ
அதுதான் tea போல் தோன்றிடும்!
நான் இருக்கிற வீடு வீடே அல்ல;
Football stadium!
மாப்பிளே, photo, அது பாத்தா பூச்சாண்டி!
[he] Playboy joke சொல்லும்
Grandpa வேண்டுமே!
(Chorus)

I'm almost tempted to ask how much of it you don't understand. (Not that it really makes any sense anyway...).

[Richard W]

I thought that all groups of words, of length 3 or more, have an underlying binary tree structure, whose nodes can be a single word, a branching point, or the null set? At least according to some theories of syntax. Not very well versed in any syntax theory, though.

That's wrong on at least two levels. Firstly, sometimes a binary structure doesn't work - 3-way splits may be necessary. Secondly, structures don't always respect word boundaries.

[zompist]

That's wrong on at least two levels. Firstly, sometimes a binary structure doesn't work - 3-way splits may be necessary. Secondly, structures don't always respect word boundaries.

Can you give examples that illustrate these?

(Minimalism insists on a binary tree structure... I'm not convinced myself, but I also am not aware of anything that really requires ternary trees.)

[akam chinjir]

I thought that all groups of words, of length 3 or more, have an underlying binary tree structure, whose nodes can be a single word, a branching point, or the null set? At least according to some theories of syntax. Not very well versed in any syntax theory, though.

That's wrong on at least two levels. Firstly, sometimes a binary structure doesn't work - 3-way splits may be necessary. Secondly, structures don't always respect word boundaries.

But it's not at all wrong to say, as Zju said, that "according to some theories of syntax" terminals are always words and all branching is binary.

(Minimalism insists on a binary tree structure... I'm not convinced myself, but I also am not aware of anything that really requires ternary trees.)

Offhand, challenges to exclusively binary branching have come from work on VSO languages and especially nonconfigurational languages, in both cases largely because they can seem to raise trouble for the idea that, underlyingly, OV (or VO, depending on other assumptions) is a constituent in all languages. I don't think this sort of argument is widely endorsed these days, though, at least among linguists who are broadly sympathetic to formal syntax.

As for not respecting word-boundaries, though---at least in things I read, it's very widely thought that words get put together in syntax, and that the terminal nodes in a syntactic tree might be correspond to affixes rather than full word. (Strictly speaking, on most of these treatments the terminal nodes are just feature bundles, up to the point when vocabulary gets inserted.)

You can get two sorts of case. Sometimes, the resulting word gets its own node in the syntax tree---it's a constituent. So if you merge "play" as head of VP, and "-ed" as head of TP, with "play" then moving to adjoin to "-ed", you get "play+ed" as a constituent with internal syntactic structure.

played.png (5.91 KiB) Viewed 4902 times

Sometimes instead, at least on some views, you can get words that aren't syntactic constituents. I expect that "re-read" could be analysed this way, with "re" occupying the position in syntax of an iterative aspect marker (or whatever), depending on "read" phonologically but not forming a constituent with it.

reread.png (8.13 KiB) Viewed 4902 times

("Reread" will be a phonological word, and it'll have some of the syntactic properties of words, like relative uninterruptability, without being a constituent; on this sort of view, wordhood ends up being pretty epiphenomenal, and it's natural to think of it as coming in degrees.)

(Er, I hope it was okay to add attachments, I thought some people might find it helpful to see the trees and I'm not set up with my own image-hosting; I can remove them if it's a problem.)

Edit: obviously I looked back and immediately saw mistakes in the diagrams. Amateurs...

[Travis B.]

From everything I've heard about linguistic theories, I personally have come to dislike theory as a whole. All in all, most likely, every theory (properly speaking, hypothesis) is wrong, so it is probably best to assume from the outset that any given theory is wrong. Theories tend to make all so many assumptions about languages that there are almost certainly natural languages which do not obey what they claim as ironclad truth, and unless there is a theory that can account for all natural languages, it is probably not worth even considering such a theory.

[zompist]

As for not respecting word-boundaries, though---at least in things I read, it's very widely thought that words get put together in syntax, and that the terminal nodes in a syntactic tree might be correspond to affixes rather than full word. (Strictly speaking, on most of these treatments the terminal nodes are just feature bundles, up to the point when vocabulary gets inserted.)

Sometimes instead, at least on some views, you can get words that aren't syntactic constituents. I expect that "re-read" could be analysed this way, with "re" occupying the position in syntax of an iterative aspect marker (or whatever), depending on "read" phonologically but not forming a constituent with it.

reread.png

Thanks for the example. This has been a big bone of contention in the history of syntax. Chomsky started out with explicit nodes for basic morphology (tense, plural, EN...), all to be handled somehow by the phonological component. The generative semantics folks ran with this, making words like "destruction" and "runner" and "self-employed" derived within syntax from basic components. Chomsky hated this and came up with "lexicalism", meaning all these words already existed as such in the lexicon. But you just can't keep formal syntax from attempting to conquer morphology. It seems like it's all back in Minimalism.

I'm wary of this kind of analysis, mostly because unlike other constituents, the presumed basic elements are inaccessible by transformation. E.g. you can't say something like

A: I've re-read the book.
B: *Which book did you re-?

I've re-[read the book]_i in the time it took Harry to ø_i.

More than that, I just distrust really complicated syntactic theories. Things get really messy when you attempt to handle morphology, semantics, and pragmatics with syntactic trees, and I'm not convinced there's much conceptual gain.

[akam chinjir]

It seems like it's all back in Minimalism.

There are certainly people doing things very reminiscent of generative semantics, down to abstract CAUS morphemes (like kill = CAUS dead). In a way my favourite is the analysis of have as be with an incorporated preposition; I don't think that goes back to generative semantics, though.

I'm wary of this kind of analysis, mostly because unlike other constituents, the presumed basic elements are inaccessible by transformation.

Sure, but you're not generally free to move whatever constituents you want however you want. E.g., you can't move "book" from inside "the book" (*Whatᵢ did you read the ___ᵢ?), but presumably you think of "the" and "book" as being put together in syntax.

More than that, I just distrust really complicated syntactic theories. Things get really messy when you attempt to handle morphology, semantics, and pragmatics with syntactic trees, and I'm not convinced there's much conceptual gain.

I suppose the main conceptual gain is that you can say things like: the reason why analytic and synthetic causatives have the same semantics is that they have the same underlying syntactic structure. (Fancied up a bit, that's the main motivating argument in Mark Baker's Incorporation, which is very influential on this sort of approach to word-building.)

Of course you get arguments of other sorts too. It's a big simplicity gain if you can treat word-building in syntax with minimal additional stipulation. And if you've accepted (for example) that an incorporated noun got prefixed to the verb in syntax, it's easy to conclude that a tense marker that's prefixed to the noun+verb combination also gets added by syntax.