Antonio Tarallo frisella at IOL.IT
Fri May 14 21:29:47 UTC 1999

In fact my idea was to suggest that Irish is not a pro-drop language, using Dahl's test, though it is mostly classified as pro-drop. Actually, I believe that  both  Irish  "pronouns"   and "verbal endings" are pronoun allomorphs, despite the traditional bipartition in the grammars. Nevertheless, I also believe there to be a certain (diachronic) bias for European languages towards considering such words as sé as pronouns, since they used to be such (i.e. NPs). I think that if Irish were a Papuan recently discovered/described and unwritten language, it would  probably be classified as having anaphoric agreement, of the sort described by Bresnan and Mchombo (Language 1987) for Chichewa: either  a  subject-person inflectional morpheme is present on the verb, or an independent NP (be it nominal or pronominal) is there, but the two  never cooccur. In fact overt NP and person index cooccurence does not seem to be a distinguishing feature of verbal affixes in such languages.
In the end the point seems to be: how much allomorphy is it allowed to be there to be able to group the (cliticized and syntactically constrained) pronouns under  single lexical morphemes rather than view them as verbal affixes? Note that, positionally, both subject NPs and verbal endings occur after the verb root in Irish.  A few "allomorphs" (e.g. -as, which is a 1st sg. "ending" used only for past forms) vary according to verb tense: wouldn't this be among the  criteria pointing towards verbal affix rather than word? But then the fact that *Canann sé Sea'n doesn't occur is not highly significant, since neither does * Chanas mé occur. Concerning these matters, any Irish dialect would actually need a different analysis, since the number of such allomorphs (the old "synthetic" endings) varies considerably in the dialects. Where even verbal affixes behave like NPs (be they old endings or pronouns), you have non pro-drop languages, but they look quite different from English, since the
affixes are neither similar to English -s (pure agreement) nor to italian -a in, say, (Giovanni/lui) parla 'John/he speaks' (ambiguous agreement).

Alan R. King ha scritto:

> In response to Elisa Roma, on the basis of her information I am left doubting whether modern Irish really qualifies as pro-drop. If we take the approach suggested by Östen Dahl in this discussion:
> >>>>
>      The question is
>      thus if there are languages ...
>      which pattern like English, with "John sings" and "He sings" containing the
>      same number of words.
> <<<<
> then it apparently does not, because _Canann Seán_ "John sings" and _Canann sé_ "He sings" have the same number of words. That is, you cannot omit "John" *except* if you insert "he" (or some other subject, of course). Now Roma seems to be suggesting (if I interpret correctly the gist of her message) that _sé_ may not need to be considered what I (in my earlier contribution) called a real pronoun (i.e. something that constitutes an NP) because it is a clitic form; the alternative analysis would be that it is a pronoun (even though cliticized) in allomorphic alternation with the non-clitic form _é_. How to decide which way to analyse it is the sort of problem with which this issue is typically plagued, but one fact pointing to the analysis of _sé_ as a pronoun is precisely that, as Roma points out, it cannot co-occur with a noun subject: if it were merely a subject-person index (not an NP), we would expect something like *_Canann sé Seán_ to occur, I think. So I believe the question
> remains open (on the basis of Roma's data, I mean).
> Alan R. King, Ph.D.
> alanking at
> - - - - - - - - - - - - - - - - - - - - - - - -
> SNAIL: Orkolaga plaza 3 1A, 20800 Zarautz, Basque Country, Spain.
> PHONE: +34-943-134125 / FAX: +34-943-130396
> Alternative email addresses: mccay at, a at, 70244.1674 at
> Internet: <>

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <>

More information about the Lingtyp mailing list