On Mon, Nov 19, 2012 at 7:51 PM, Alex Fink <[log in to unmask]> wrote: > On Mon, 19 Nov 2012 11:33:59 -0800, Garth Wallace <[log in to unmask]> wrote: > >>On Mon, Nov 19, 2012 at 10:59 AM, Wm Annis <[log in to unmask]> wrote: >>> On Mon, Nov 19, 2012 at 12:49 PM, Padraic Brown <[log in to unmask]> wrote: >>>> I'd still hold that while indefinite >>>> subject / topic sentences may not be *as common* as defininte subject / >>>> topic sentences in English, they are not "unusual" or "surprisingly rare" >>>> in any way >>> >>> According to the following paper about English, the avoidance of agent >>> NPs is widespread, >>> >>> http://www.ccsenet.org/journal/index.php/ijel/article/view/17582 >>> >>> From section 2.1, "Preferred Argument Structure," >>> >>> "One of the factors that very likely interacts with Halliday and >>> Hasan's (1976) description of cohesive devices in English is known as >>> preferred argument structure in discourse. This has been shown to >>> exist in the discourse patterns of a range of the world's languages, >>> and may quite possibly be a universal of discourse organization in >>> human language (DuBois, 2003). It is preference in discourse >>> organization that consists of two related aspects: 1. Reduced forms, >>> such as pronouns, are generally preferred in the agent role >>> (e.g. transitive subjects); 2. This syntactic role also favours old >>> information rather than the introduction of new information." >> >>I'm not sure that isn't just an accident of statistics rather than an >>observation of how language works. After all, you generally only >>introduce a topic once, and as soon as it's introduced it's definite >>and available for anaphora. >> >>Indefinite subjects definitely don't seem at all awkward to my native >>speaker intuition, or even stylistically notable (like headline text, >>which drops articles and other "small words" whenever possible for >>brevity). > > I think this touches on a key point, which is that accidents (and 'accidents') of statistics can bec˛me how language works. Today's "statistically unlikely, though you wouldn't notice anything odd on hearing one" can easily become tomorrow's "so statistically unlikely that it _is_ surprising" and the next day's "ungrammatical". The conversation on this thread has established that even in English speech the proportion of indefinite transitive subjects _is_ less than the (proportion of indefinites) * (proportion of transitive subjects) that it would be if everything were due to chance, but it's not noteworthy to hear, so English is at stage Today, and perhaps not even moving. That doesn't mean the cline isn't there. Hmm, true, but my point wasn't that it couldn't become a prohibition (obviously, some languages DO prohibit it!), but that its origin isn't necessarily something inherent to language. It may be less "People tend to avoid indefinite subjects" and more "Indefinite subjects happen less often because things cease to be indefinite once they're introduced".