On Tue, 27 Oct 2015 18:55:53 -0700, Jeffrey Brown <[log in to unmask]> wrote:

>Couldn't the reason be that of minimization of effort?
>Let's suppose, for some language, that there is some part of speech for
>which there is no conjunction. Then, to communicate the conjoining, the
>speaker would be required to use two sentences or two clauses:
>-- John went to the market and Tim went to the market
>-- John went to the market and he went to the gym
>-- John went to the market with his brother and he also went with his sister
>It seems natural that conjoining would develop for every part of speech.

You speak of parts of speech, and then all three of your examples are conjunctions of nouns!  Well, noun phrases.  Different syntactic positions, but all noun phrases.  The ability to conjoin noun phrases can, I think, be accepted as a universal, or close enough: the WALS chapter I cited didn't name any language where there was no strategy for that.  

Moreover, I wasn't asking jùst about parts of speech.  There might be some phrases which have no "part of speech", in the sense that there's no single word with the same syntactic distribution as that word, and one can still ask the question of whether they can be conjoined.  In fact, by restricting to constituents I asked my question too narrowly, as And pointed out:

[And:]>Also: the to-me noteworthy aspect of "I heard and saw the dog" is that the
>conjuncts share a complement, and that's why I introduced right node
>raising in my earlier contribution.

"Probably heard" is not a constituent of [probably [heard [the dog]]], and one still might want to conjoin it.  

Although, if we're on parts of speech, how about, say, determiners?  Let's say I see you're finishing up writing something with a pen and I need one myself -- can I say "Hand me your or a pen"?  Not in my English!

On Wed, 28 Oct 2015 09:20:37 -0700, Jeffrey Brown <[log in to unmask]> wrote:

>Rethinking what Alex Fink had said, and given the number and diversity of
>natlangs in the world, it wouldn't surprise me if there was one which did
>not allow the conjoining of some part of speech. If that is the case,
>though, I would suspect that under normal diachronic evolution, some word,
>a comitative for example, would be pushed into service to function as a
>conjunction for that part of speech.

I'm not convinced.  There are a whole whack of other syntactic strategies that might be usable, and which a language might resort to in order to reduce repetitiousness, that are not conjunctions.  Ray's Latin example, _Thomas canem auditum vidit_, is just the sort of thing I mean: instead of conjoining verbs, leave one of them as the main verb and demote the other one to a participle that syntactically modifies the noun.  Or one could take advantage of non-finite verb forms in adverbial function (also AIUI common in good Latin style), or even just various sorts of anaphora like "do so" augmented by a rule like omitting a repeated pivot (as in Jeff's analysis).  Simple juxtaposition is common too -- but perhaps you'd call that a "zero conjunction"...