Extreme Adjectives in Comparative Structures and Even

This paper examines two related puzzles, observed in the literature about extreme adjectives (see, e.g., Paradis 2001, Rett 2008, Morzycki
2012), namely, (a) why such adjectives are questionable within comparative
structures and (b) why and how exactly the presence of even improves the
felicity of such constructions. After examining the solutions proposed in
Morzycki 2012 for these two puzzles, we propose an alternative solution
which integrates three components: (i) the fact that extreme-adjective comparatives necessarily presuppose the “positive form” of these adjectives
(building on Morzycki’s semantics for extreme adjectives); (ii) an updated,
gradability-based semantics for even (Greenberg 2015, 2018), which guarantees that comparatives with even presuppose the corresponding “positive
form,” with all kinds of adjectives (extreme and non-extreme alike); and
(iii) a local Maximize Presupposition!-type effect, such as that suggested
by Singh (2011), leading to the preference of the extreme-adjective comparative with even over a competing alternative without it. While the latter
component presents a number of challenges requiring further research, the
proposal is shown to be supported by cross-linguistic data and by comparing extreme and lower-closed adjectives in terms of scale structure and
behavior in comparatives.