Martin Haspelmath: Four stereotypes that have guided morphosyntactic thinking
Thinking about language structures is made difficult not only by their incredible complexity, but also by entrenched ways of thinking about grammatical and lexical patterns. Linguists do not investigate languages in fresh way, but against the background (and often on the basis) of a centuries-old tradition.
Could it be that these traditional and stereotypical ways of thinking sometimes get in the way of approaching our objects of study in a fair way? Few linguists would deny this possibility, so here I will list four ways in which this may have adversely affected morphosyntactic descriptions and general theories:
– the word stereotype (1)
– the grammar/dictionary stereotype (2)
– the building-block stereotype (3)
– the speaker directionality stereotype (4)
My really bad TLDR: words don't exist, grammar is like words and words are like grammar, language isn't done by putting things one after the other, and we study too much how we make language and not enough how we make sense of language. Bonus sub-point: we like to say A is made of B but we could also say B is made of A.
https://dlc.hypotheses.org/4343Open linkView original on lemmy.world
I disagree with a lot of what he says but it's an interesting take nonetheless.
I'll focus mostly on the "word" stereotype, as it's most of the basis for the rest of his text.
"Word" is a somewhat well-defined string of sounds that can be pronounced in isolation, and holds at least some meaning. It's specially visible in historical linguistics, since certain changes depend on word position.
For example. Latin /p t k/ become Western Romance /b d g/ between vowels. Now compare the following cases:
Note that accusative -m in #2 represented vowel nasalisation, and was lost rather early, probably already in Imperial times. As such, #2 does count as intervocalic, and yet that /p/ in ⟨illam paleam⟩ never became /b/ in Spanish or other W. Romance language, unlike the one in ⟨aperire⟩. Why?
The only way to explain why one /p/ becomes /b/ and the other is left alone is to bring up the concept of "word": the change only applies word internally.
Now. Are there sequences that sit right at the boundary between "word" and "not word"? Certainly. I just mentioned one, that ⟨la⟩. And in some languages it's probably sensible to make the concept of word take a backseat, and talk about morphemes or something else instead. But that doesn't mean words don't exist, it's just that's is a bloody mess.
I agree a fair bit with his (3), though.
Zero morphemes were always a hack, and while some language alternations are nice to explain as building blocks, a lot of them are not. Plural in English seems to be one of those cases, less because of "sheep" and "deer" and more because of "goose"→"geese" and "mouse"→"mice".
For an extreme case, consider Arabic ⟨سُكَّر⟩ sukkar "sugar":
(Arabic speakers: please do tell me if I got something wrong above. Also I'm not too sure on how to translate the later two, passive voice is tricky to represent in English.)
If you try to handle this sort of alternation simply by "building blocks", you'll spam zeroes and most of those blocks won't really mean anything. For example that /u/ in "sukkira" doesn't really mean "passive voice", a lot of passive forms don't use it (like ⟨يُسَكَّرُ⟩ yusakkaru, the passive non-past masculine). It's more like fitting two highly complex pieces (the root vs. the conjugation) than just individual little blocks, you know?