Linguist Forum

Specializations => Morphosyntax => Topic started by: Anne on September 12, 2018, 09:02:06 PM

Title: How can you test whether a word is being used as a conjunction?
Post by: Anne on September 12, 2018, 09:02:06 PM
It's been a couple years since I've taken a syntax class, and I've forgotten - what tests can you use to check whether a speaker uses a word as a conjunction? I seem to remember something about testing where the speaker can attach different clauses in the sentence, but I don't remember the details. I'd appreciate any explanations, or links to published papers discussing this topic!
Title: Re: How can you test whether a word is being used as a conjunction?
Post by: Daniel on September 12, 2018, 09:44:06 PM
"Conjunction" is a fairly broad term that can be used for a variety of elements linking clauses. If you can identify clauses and there is an element between them, you can call it a conjunction. More specifically you can try to determine whether the relationship is coordinate or [/i]subordinate[/i], and then refer to it as a "coordinating conjunction" or "subordinating conjunction".

The Coordinate Structure Constraint (Ross 1967) is one test that can be used for coordination: that you cannot extract non-parallel elements from half of what is coordinated:
*What did you drink soda and eat?

However, that doesn't doesn't always work:
Who did you go to class and meet?

And coordinating conjunctions can also coordinate sub-clausal constituents ("mother and father" or "sing and dance"), and there is some debate about whether that is directly related to the same structure as clause coordination.

As for clausal subordination, that's also a complex topic, with various different types of subordination, and also some other approaches to labeling the "conjunctions" (which is a pre-theoretical term), for example calling those that mark complement clauses "complementizers" as in "that": "He said that he was..."

And there are also other more complicated distinctions, such as various hybrid ideas like cosubordination, pseudocoordination, pseudosubordination, para-hypotaxis, etc. (For more details, see:'and')/msg32780/#msg32780)

The point is that this topic is either relatively simple (from a traditional or causal descriptive perspective) or very complicated (trying to get the right precise description or theoretical analysis today).

I'm not sure about "tests" to check for conjunctions per se, but I can imagine some ways to investigate the topic, depending on the circumstances.

Could you give an example of the sort of situation you're thinking about?

I'll just conclude for now by saying that although there has been a lot of research (I'm working on related topics myself), there is no consensus yet on exactly how to deal with clause combining, so some of the "answers" will also depend on your perspective/theory.

I can probably recommend something more specific if you can explain your topic a bit more.

Ross, John Robert. 1967. Constraints on Variables in Syntax. MIT Ph.D. dissertation.
Title: Re: How can you test whether a word is being used as a conjunction?
Post by: Daniel on September 12, 2018, 10:43:12 PM
Take a look at this paper:
Van Valin* establishes the idea of 'cosubordination' as a non-embedded but morphosyntactically dependent relationship. However, footnote 7 on p.557 briefly discusses the idea of a non-dependent but embedded construction, which could include parentheticals.**

The Role and Reference Grammar methodology (to which that belongs) is popular for descriptive work because it allows a more fine-grained distinction about clause linkage.

So you might think about that kind of distinction, whether aka-clauses are embedded or not, and then where to go from there.

It isn't immediately apparent to me whether "aka" would be embedded or not, although I would default to a parenthetical explanation. But if you start to find instances that seem more integrated, you might be seeing the initial development of a conjunction.

Further, you would also want to show that "aka" itself is actually part of the clause, or a linking element between clauses (depending on your analysis), rather than a parenthetical element itself. For example, you might look at intonation, although using textual examples from Twitter won't give you data like that.

The other area that this borders on is 'discourse particles', words like 'like' or 'um' (traditionally grouped under 'interjections'), and that might also be helpful. But there are well known cases of discourse particles grammaticalizing as conjunctions, for example quotative 'like' in English-- "He was like, 'Hello!'"

[*You can find several [text]books that go into more detail about RRG if that seems helpful. That article is just easily accessible and concisely about this point in particular.]
[**I know that parentheticals were explicitly discussed as that possible fourth type somewhere in the RRG literature, although I'm not seeing it in that article at the moment. You might do a more thorough literature review to see if you can find that if it seems important to you.]

By the way, one other thing you could consider would be constituency tests in general, although in some ways this could get complicated if you have a (possibly) parenthetical element with probably fairly free word order. Tests like extraction or fronting might give you some information, but could be hard to interpret for that reason.
Title: Re: How can you test whether a word is being used as a conjunction?
Post by: Daniel on September 13, 2018, 06:58:27 AM
Generally it's best to leave up enough content so others can learn from the discussion. What matters more is doing a good, thorough job on the research and finding relevant results/analyses, rather than the original question. (It's almost certain someone has had the same question before. But have they done a good job answering it?)

By the way, more broadly, you might look at the history of English (relative, complementizer) "that" for example, because it originally comes from two separate clauses, something like "I know. That is interesting."