自從二十世紀後期，語言學就進入了百家爭鳴的時代， 各種新的理論、思維和潮流層出不窮， 雖然各學派相互影響、相互激盪，但是不少學者仍舊有很深的門戶之見， 語法學家Andrew Carnie在著作《Syntax: A Generative Introduction》， 第一版文字當中透露出許多自己個人的看法和感想， 在這裡摘錄部分跟大家分享.... (第二版以後就修去了比較個人情感方面的文字， 雖然看起來更像一本理性的教科書，但卻失去讀者一窺作者身為語言學家的感觸...) (p.371-372) CHOOSING AMONG THEORIES [...] We briefly turn now to the very thorny question of which theoretical approach is right. If you ask this question at any major syntax conference you are likely to get lynched. Most linguists hold to their theories the way religious fanatics follow their beliefs or the way nationalists feel about their countries. I admit that I personally am guilty of this at times. [....] Unfortunately, there is rarely rational dialog on the question of what theoretical approaches are the best. At the same time, the theories quite liberally borrow from one another. [....] Now it is true that to a greater or lesser degree the different theories make different empirical predictions. One might think that on empirical grounds alone, you should be able to choose the right theory. However, if you take this approach you are treading on dangerous ground, for two reasons. First, while one thoery provides a nice account of one part of syntax, another theory will do better at a different component, so you have to carefully balance what parts of syntax are the most important. Second, some theoretical approaches are better worked out than others. More people work in P&P/Minimalism than in the other approaches, so the empirical coverage of that theory is unsurprisingly greater. You might think instead that we can compare the theories on the ground of elegance or precision. [....] But this doesn't cut it either: Precision or elegance does not necessarily mean that the theory is an accurate representation of human Language. In fact, the only real grounds along which we could ever accurately gauge the correctness of a theory is on the basis of how well it models how Language works in the mind or brain. Despite some scholars' claim to the contray, we are long way from being able to test such modeling reliably. I suspect that when we do, we'll discover that all of our theories are wrong in many important respects. In the meantime, we're left with a number of theoretical approaches that do roughly the same range of work, for the same basic goals. Instead of trying to determine which one is "right" (probably a fruitless work), it is better to understand the advantages of each approach, and the insights they give us into the nature of human Language. --

※ 發信站: 批踢踢實業坊(ptt.cc) ◆ From: 115.81.205.97