自然語言的並列：preference semantics at its worst

NLU、NLP 多年來還有一個公認的難點，就是並列結構（conjoined structure）。並列在思維邏輯里沒有地位，它是語言表達的產物。並列是語言學中最不講道理的程咬金，它總是橫插一刀，而且任性，在任一層次。一切的 subcat arg structures 或 mod-head patterns 都必須為它讓道，否則就堵塞交通，讓 parsing 的路線斷鏈。然而，如果沒有並列，自然語言就會難以容忍地單調枯燥，盡失精簡。

舉個簡單例子：

這句話邏輯上展開以後怎麼樣呢？

頸椎間盤突出症的最常見和最典型表現是一側頸肩部及上肢的酸痛

==>

頸椎間盤突出症的最常見表現是一側頸肩部的酸痛

頸椎間盤突出症的最典型表現是一側頸肩部的酸痛

頸椎間盤突出症的最常見表現是上肢的酸痛

頸椎間盤突出症的最典型表現是上肢的酸痛這才牽涉兩個並列，一個句子出現五六個甚至上十個並列， 並不鮮見。

這才牽涉兩個並列，一個句子出現五六個甚至上十個並列， 並不鮮見。語言不是邏輯。沒有並列，語言面臨組合爆炸式啰嗦。很難想像，傳統的單層 parsing 系統，譬如教科書上經典的喬姆斯基式 CFG-based chart parsing，可以把各種並列處理妥帖。並列 can be so f* hierarchical, even for a very deep multilevel

Conjoinment can be so f* hierarchical, even for a very deep, multilevel parsing system: conjoin remains a challenge if not very carefully/skillfully handled by a very experienced linguist 』cause the boundaries are tough to identify and they just appear at any levels at will. The conjoined elements are semantically parallel but the parallelness, which ideally should be used as conditions to help identify the conjoined structure and its scope, is unfortunately in practice all relative and fuzzy, which can hardly be enforced. food can be conjoined with food, of course, but look at this:

我喜歡肥肉和哲學。

food and knowledge, totally different monsters of semantics, can also be conjoined, it is preference semantics at its worst.

OK, I am not going to elaborate solutions, which should be a long article by itself. This post serves as an introduction of this linguistic monster, to arouse the awareness of linguistic challenges in natural language parsing.

喜歡這篇文章嗎？立刻分享出去讓更多人知道吧！