Meaning is a red herring

[dr_whom]

Ling 1 pet peeve time!

Introductory linguistics textbooks typically define phonemes, or minimal pairs, or contrastive distribution, somehow in terms of "meaning". A typical example from Language Files (Bergmann et al. 2007): "We say that these two sounds are noncontrastive in English, since interchanging the two does not result in a change of meaning. In Hindi, on the other hand, [p] and [p^h] are contrastive, since replacing one sound with the other in a word can change the word's meaning."

This is extremely irritating!

What matters isn't whether changing a sound changes the word's meaning; what matters is whether it whether it changes the word's identity—i.e., whether it changes it into a different word. These are not equivalent! (E.g., sled and sledge are arguably different words with the same meaning that differ in one phoneme—the White Witch takes Edmund for a ride on her sledge. Changing /d/ to /dʒ/ therefore changes the identity of the word but not the meaning.)

In practice, for immediate purposes, this isn't a big deal, since in general it leads students to the right conclusions about which segments are different phonemes and which aren't, which is what the point of that definition is when it's introduced. But it leads students to sloppy habits of thinking that what you're doing when you change a phoneme is changing the meaning of a word, rather than changing it into a different word. This can lead, for instance, to students learning that tone is phonemic in Mandarin and then believing something like "The Mandarin word ma can have four different meanings depending on what tone you pronounce it with," rather than "Mandarin has four different words whose segmental content is ma which differ in their phonological tone." (For comparison: "The Mandarin word ma..." is about as incoherent as "The English word b_t can have twelve different meanings depending on what vowel you pronounce it with." There's no sense in which we're talking about one word with multiple meanings here.)

This habit of thinking that features which aren't used in English somehow don't count as creating "different words" is hard enough to train people out of already; we don't need the phonemes-change-meaning red herring contributing to it as well.

Comments

[midnight-sidhe.livejournal.com]

"Mandarin has four different words whose segmental content is ma which differ in their phonological tone." (For comparison: "The Mandarin word ma..." is about as incoherent as "The English word b_t can have twelve different meanings depending on what vowel you pronounce it with."

Delightfully put!

Is this one of the things you're fixing with your Very Detailed Handouts?

[dr_whom]

I'm trying to! But I can't stop the students from reading that chapter of Language Files and getting wrong ideas put in their heads.

[midnight-sidhe.livejournal.com]

Technically you could...

[dr_whom]

They've all already got the book!

[midnight-sidhe.livejournal.com]

Inarguable! You could probably fix this in later years, though? What if you gave them a bulkpack with readings from Language Files not including the egregious ones? I think Dave and Charles used to do that.

[dr_whom]

Hmm, maybe so. Though I think if I were going to go the bulkpack route, I'd give them readings from a better textbook than Language Files.

[midnight-sidhe.livejournal.com]

It is pretty terrible. I think if I were teaching Ling 1 I would go bulkpack. Possibly I would then go insane and decide to write half the readings myself.

[tiamat360.livejournal.com]

You could always not assign that chapter, thereby guaranteeing none of the students will read it!

[dr_whom]

I don't really assign any chapters—I just list in the syllabus which chapters in the textbook are relevant to each week's lectures.

[lord-codfish.livejournal.com]

I'm with tiamat360 - I think the idea that your students might, of their own volition, read a chapter that was not assigned to them is not, in general, something to worry about. Except for maybe the really motivated and/or intelligent ones, but hopefully they'll have enough sense to ignore the dumb bits anyway (or at least be more amenable to have them purged my more reasonable ideas later).

[q10.livejournal.com]

i think the issue is that cultivating nonsemantic judgments for word identity is really hard, and there's a desire not to put students in the frustrating position of having to guess when two things are or are not the same word, without having good tools to do this.

[dr_whom]

Yeah, that's true. But at least we could say something like "changes it into a word with a different meaning" rather than "changes the meaning of the word", since the latter leads to confused notions about word identity later on (such as that Mandarin má and mā are in some sense the same word with different tones).

[q10.livejournal.com]

that would work.

...and now i'm trying to figure out a way to show that sled and sledge are different words that doesn't resort to the fact that /d/ and /dʒ/ are different phonemes in English, or to a ‘they just are, dammit!’ sort of intuition.

[dr_whom]

Yeah, even resorting to the fact that /d/ and /dʒ/ are different phonemes wouldn't be probative by itself, since /n/ and /ŋ/ are different phonemes but walking and walkin' are the same word. (And a slightly different example: /i/ and /ai/ are different phonemes, but /iðɚ/ and /aiðɚ/ are the same word.) I'm not really sure what to resort to here other than "they just are, dammit!" intuitions.

[q10.livejournal.com]

for walking and walkin', it's part of a fairly general pattern that's associated with a particular register-type-thing, so i think we'd be within our rights to say that what's going on here is that there's a fairly general rule that's only active in a certain register (and lots of things are pronounced differently in different registers), or at worst that they're technically different words, one of which is derived from the other by some mostly semantically vacuous morphological process, or something like that.

for either, i've got nothing.

[dr_whom]

"Only active in a certain register" is a little strong, as a matter of fact. But yeah, I'm pretty sure there actually have to be two different (but parallel) operations going on here: one, the morpheme -ing has a (variable) allomorph -in'; and two, there's a (variable) phonological rule changing unstressed [ɪŋ] to [ɪn] at the end of a word. So roughly speaking both of the things you suggest are right on some level, although the question of whether different allomorphs of the same morpheme can constitute "different words" or not is above my pay grade. (The etymology here is interesting too; -ing and -in' actually go back to different Old English morphemes which merged into allomorphs of the same morpheme sometime in Middle or Early Modern English.)

[q10.livejournal.com]

yeah, English is lamentably not as clean as Samoan in this regard (in Samoan, if i remember right, /n/ and /ŋ/ contrast in the formal register you use in school and when talking with foreigners, but both are universally realized as [ŋ] in the colloquial register that actual Samoans use when conversing).

i still feel like we ought to have some kind of data we can bring to bear to justify these very theory-laden intuitions about word identity (i share the intuitions, but i don't know how much that should count for).

[boojum42.livejournal.com]

Well, if I'm ever teaching intro linguistics again, I'll remember that distinction.

[dr_whom]

Every little bit helps!

[brokenwndw.livejournal.com]

Reminds me of the misguided war on centrifugal force. (Don't say it doesn't exist. Just don't. My blood pressure can't take it.)