ABSTRACT
Section 3.2 discusses Ladusaw's early work. Ladusaw identifies the cruc ia l role of monotonic i ty in the phenomenon of po lar i ty sensit ivity. He makes a generalisation over the class of elements that license occurrences of N P I s , and analyses the phenomenon i n terms of scope relations between licensers and N P I s . Zwarts ' analysis, discussed in section 3.3, is a na tura l offspring of Ladusaw's approach and of the work on generalized quantifiers (see Barwise and Cooper , 1981). Zwarts ' m a i n concern is the variat ions in the d i s t r ibut ion of negative po lar i ty items. He notes that not a l l licensers can license a l l items, and refers to mathemat i ca l notions i n order to identify subclasses of licensers w i t h more restrictive properties. He projects onto the class of po lar i ty items the classification defined on licensers, and expresses it in terms of increasing requirements imposed by the items on the licensers. Section 3.4 discusses recent contr ibut ions by D o w t y (1993, 1994b). T h i s work analyses negative po lar i ty as one of the manifestations of negativity in natura l language. He considers sensit ivity together w i t h negative concord m a r k i n g and sets them in the
framework of monotonie reasoning patterns. T h i s goal makes Dowty ' s work relevant for the treatment of negative po lar i ty i n general. Another point of interest i n his contr ibut ion is that i n some sort he puts loca l i ty issues back on the agenda of semantic analyses. A s we saw i n section 2.3, Baker (1970) formulated triggering conditions i n order to account for the d i s t r ibut ion of po lar i ty items in double negative contexts. B a k - er's proposal was meant to improve the idea of evaluating the po lar i ty of each node of a tree-like representation by cancell ing out negations i n pairs s tart ing from the root ( K l i m a , 1964; Jackendoff, 1972). Such a strategy was shown to have poor predict ing power. T h e chapter closes w i t h a discussion of the points where this study ties i n w i t h preceding research and expands on them.
