Beck, Oda & Sugisaki 2004: Parametric Variation in the Semantics of Comparison #
[beck-oda-sugisaki-2004] argue that Japanese yori-constructions are not
English-style degree comparatives: the yori-constituent is a free relative
denoting an individual ((34): ⟦John-ga kaita⟧ = what John wrote), which sets
the context for a positive or contextual comparative in the main clause,
rather than a than-clause denoting a set of degrees. Japanese, they
conjecture, lacks binding of degree variables in the syntax — their Degree
Abstraction Parameter — while English has it. Three empirical differences
follow, recorded in Data/Examples/BeckOdaSugisaki2004.json: variable
acceptability of degree (vs amount) yori-comparatives ((3)–(4)), the absence
of subcomparatives ((5-a) vs (5-b) — the English side is
Degree.subcomparative), and the absence of English-like negative-island
effects ((6)).
The standard clause of this construction only has a well-formed LF if the language abstracts over degree variables in the syntax.
Equations
- BeckOdaSugisaki2004.standardRequiresAbstraction e = (e.feature? "standard_requires_degree_abstraction" = some "true")
Instances For
The Japanese rows of (3)–(6).
Equations
Instances For
The paper's parameter at work in Japanese: a construction whose standard clause requires syntactic degree abstraction is ungrammatical, and every other recorded construction is acceptable. The degraded-but-not-star (4-a) is excluded — its variability is the paper's separate explanandum.
The English controls: both degree-abstraction constructions are well-formed LFs, and (6-b)'s ungrammaticality is instead Rullmann-style maximality failure over the degree set — undefined, not unbuildable.
Equations
Instances For
English has degree abstraction: the subcomparative (5-b) is acceptable even though its standard requires it.