W.Schulze at LRZ.UNI-MUENCHEN.DE
Thu Feb 12 08:55:12 UTC 2004
Eitan's observations seem to put the discussion on a very interesting and
promising track. I do not know whether the assumption of a "discountinous
element" can be justified from an adequate theory of morphology (others can
perhaps help): Nevertheless, from a 'syntactic' point of view, this comes close
to what I think is crucial, namely that the type Pro:Pers + V:Pers (to put it
into a very simplified formula) in fact represents a 'construction' in a (albeit
modified) sense of Croft's Radical Construction Grammer. Here, it is not
important whether 'Pers' plays the role of a 'subject' (whatever this may be) or
not. The crucial point would be that every verbal relation is embedded in a
construction that works 'indexically' (be it SAP-oriented or Referent-oriented).
Basically, we can assume that language economy wil favor a 1-to-1 representation
of this indexal function, namely 'one element per constructional specification'.
>From this it follows that both 'Pro:Pers (~ Ref) + V' and 'V:Pers(~Anaph)'
represent the two options of the constructional prototype (another type would be
the amalgamation of V and Pers(~Anaph) producing a 'person-specific' lexeme
(compare (synchronic) German 'bin', English 'am', French 'suis' etc.; there
surely are much better examples)). Anything 'in between' these two patterns
would carry function (at least diachronically), but it would not alter the
constructional type as such. Accordingly, the redundant 'doubling' of the
indexal feature of constructions may function as some kind of structural
reduplication (with all the functional implicatures that are given for
reduplications). Just as say Udi me seri seri-ne (prox true-true-3sg) 'this is
true' does not open a new constituent slot (*'this is true (and) true'), the
'doubling' in say Persian man na-mi-da:n-am (I neg-ind-know:pres-1sg) 'I do not
know' retains the 1-to-1-pattern on the conceptual layer. The structure
'man....-am' in fact does just what Udi seri-seri (>serseri) does, namely
exploiting the iconic 'potential' of reduplication in terms of (for instance)
In other words: I strongly opt for the assumption that there is only *one*
conceptual constituent in structures like those Eitan has mentioned. The
question naturally is, whether we have good reasons to suggest that there are
*two* (ranked) constituents from a structural point of view. But this draws our
attention to the well-known question whether 'structural' is meant (again) as a
heuristic tool to describe natural language features (resulting in some kind of
'meta-structural' descriptional type), or whether 'structural' properties are
regarded as original properties of a language itself (hence object-oriented). In
some way, Croft's (and others' 'constructions' help to find a way out...
Naturally, there are good arguments in favor of a '2-constituent' reading of
Eitan's constructional type. For instance, conceptual accessibility is better
given for 'words' or 'lexemes' than for 'morphemes'. Words can be in contrastive
focus, morphemes usually not (except in highly marked usage), for instance
'I am old'
zu kala-zu amma un ki'c'i-nu
I old-1sg but you:sg young-2sg
'I am old, but you are young.'
Although Udi usually 'drops' its pronouns in an unmarked context, a phrase like
*kalazu amma k'ic'inu seems not to be processable by native speakers (by the
way, I always wonder whether there are languages that allow a contrastive focus
without echoing the 'verbal person' externally). Conceptual accessibility sure
is a criterion for 'autonomy'. However, 'autonomy' is not necessarily correlated
with contituency. Rather, it seems to make sense to start with some kind of
'mid-level' expliticness as a default, which can made 'more' explicit by adding
the lexical correlate or which can be lowered to even zero-marking. Admittedly,
this brings us to a Cognitive Linguistics approach, which is not always helpful
to answer purely 'structural' questions. But still I think that we can no longer
Eitan Grossman schrieb:
> Dear Typologists,
> Perhaps this would be an opportune moment to raise a question that has often
> bothered me. In many languages, the finite verb (understood as a personal
> marker + the verbal lexeme + the nexus between them) may indeed constitute
> an independent utterance:
> Modern Hebrew: halaxti "I went", elex "I'll go"
> Coptic: afsotm "he heard"
> However, actual linguistic usage is often different. In certain registers of
> spoken Modern Hebrew, for example, an independent pronoun is indispensable
> "ani 'alaxti" (I went), "ani (y)elex" (I'll go). The functional spectrum of
> such constructions aside, how would one best analyze their internal
> structure in terms of nucleus-satellite analysis? In other words, if the
> appositive pronoun is obligatory and in agreement with the explicit
> person-marker of the finite verb, should one analyze this as a discontinuous
> element, or is there any possibility of analyzing one or the other as
> nuclear, and the other as satellital? Is there any synchronic justification
> for analyzing one as prior, or "motivating", with regard to the other?
> As an aside, one might mention that "ani yelex" (I'll go) rather than the
> normative "ani elex" is not simply assimilation of the verbal prefix to the
> front vowel in the personal pronoun, as one finds it even when elements such
> as "gam" (also), "od" (still), "lo" (negation) come between the two
> elements. This, to my way of thinking, testifies to a close syntactic
> juncture between the two elements, rather than to the result of analogical
> levelling. This phenomenon could perhaps be compared to (modern) Celtic
> mutations, which signal syntactic relations rather than sandhi phenomena.
> I'd be glad to hear your opinions on this matter, which is to my mind a more
> troubling problem of linguistic analysis than any that terms such as
> "pro-drop" (and the way of thinking that goes with them) provide.
Prof. Dr. Wolfgang Schulze
Institut für Allgemeine und Typologische Sprachwissenschaft
Department 'Kommunikation und Sprachen' (Dep. II) - F 13/14
Tel.: ++49(0)89-2180-2486 (Sekr.) / -5343 (Büro)
Email: W.Schulze at lrz.uni-muenchen.de
More information about the Lingtyp