[LAGRAM] Andrew Nevins

Lundi 4 Juillet 2016 - 12:00 à 13:30
Lieu détaillé: 

Salle 255


Andrew Nevins
Towards a Morphopragmatic Characterization of Duals and Paucals

In order to further develop the theory of features in natural language, a great
deal of recent effort has gone into the semantic characterization of duals and
paucals in number systems (for languages where these are bonafide inflection),
in which dual is taken to mean something like `exactly two' and paucals are as
a category approximative number with an upper-bound cutoff. I would like to
argue, however, that attempts at semantic characterizations are incomplete at
best until we have begun to understand the contribution made by pragmatic
reasoning, and therefore outline a program of morphopragmatic research into
these categories, based on collaborative efforts in progress with Paul Marty,
Lanko Marusic, Yasu Sudo, and Rok Zaucer. To illustrate the utility of such
research, I will present experimental results that look at the semantics vs
pragmatics of the English word "a couple", and find that "a couple" semantically
means "at least two" and that it is the pragmatics that often restricts it to mean
something like "exactly two" (or "two plus epsilon"). As such, we might revisit
the dual, and find that given appropriate controls for pragmatics, its
semantics is indeed "at least two". In experiments with downward entailing contexts
in Slovenian, we find that indeed the dual is compatible with more than two.
Extending work by Farkas & de Swart (2010), I will argue for a principle by
which the markedness of number morphology matches the markedness of a
referential hierarchy, and that dual, while semantically meaning simply "at least
two", is nonetheless the most morphologically most marked, and is thus aligned
with usage in situations of exactly two because of competition with the plural.
In fact, this speci c implementation makes the prediction that with inherently
paired objects (like shoes), where reference to two is no longer the most marked
reference (Tiersma 1982's `local markedness'), the less-marked plural is used -
exactly as con firmed in Slovenian. (If the two shoes are from di fferent pairs,
however, then one is back in the referential hierarchy where two is more marked
than more than two, and here the dual is used). Moreover, if indeed dual has a
semantics of "at least two", and it is alignment with morphological markedness
that yields exactly two, this could shed light on the fact that once the dual
morphology came to be seen no longer in direct competition with plural morphology
in Slavic, a "paucal" category (2, 3, or 4) arose as an outgrowth or extension of
the dual.