fox | ah, academic snark.

Reading over my thesis again, as the viva is tomorrow afternoon, and I've just come to this bit:

Fudge's goal, though, is simply to identify and describe the various possibilities, rather than to present any analysis of why the different types of two-word sequences may have different stress patterns. He therefore outlines many different types of compounds based on their lexical categories—(N₁:N₂)_N, (A:N)_N,and so on—then subdivides these into semantic groups (N₁+N₂=proper name, for instance) and predicts where on each compound the primary stress will fall, offering examples of and exceptions to each rule. But although these divisions are semantically well-motivated, the stress predictions seem not to hold up to much scrutiny: for example, one of Fudge's categories is (N₁:N₂)_N compounds, which he predicts will normally be initially-stressed. An exception is a group in which N₁ is a material and N₂ is made of N₁; these compounds will generally be finally-stressed. He gives as examples iron railings, garden path (suggesting that it is apparently safe to consider a garden a 'material'), and paper napkin, but allows that there are initially-stressed exceptions such as butter mountain, dunghill, snowman, and waterfall, as well as anything 'made from' cake, juice, or milk (birthday cake, apple juice, coconut milk; a flexible interpretation of 'made from', as apple juice and coconut milk are not made from apples and coconuts the same way birthday cake is made from cake—or, as in this group N₁ is the material and N₂ the product, apple juice and coconut milk are not made from apples and coconuts the same way birthday cake is made from birthdays).

[happysigh] I do like the part of the literature review where one respectfully points out that people who have gone before are, just, like, wrong.

ah, academic snark.

no subject