How to answer this question of phonology?

If you're new to these arts, this is the place to ask "stupid" questions and get directions!
Post Reply
wyl118
rupestrian
rupestrian
Posts: 6
Joined: 12 Jan 2020 14:15

How to answer this question of phonology?

Post by wyl118 »

null
Last edited by wyl118 on 19 Jun 2020 07:58, edited 1 time in total.
User avatar
Creyeditor
MVP
MVP
Posts: 5091
Joined: 14 Aug 2012 19:32

Re: How to answer this question of phonology?

Post by Creyeditor »

If you google "english plural suffix allomorphy" you should get at least three different possible solutions.
Creyeditor
"Thoughts are free."
Produce, Analyze, Manipulate
1 :deu: 2 :eng: 3 :idn: 4 :fra: 4 :esp:
:con: Ook & Omlűt & Nautli languages & Sperenjas
[<3] Papuan languages, Morphophonology, Lexical Semantics [<3]
User avatar
sangi39
moderator
moderator
Posts: 3024
Joined: 12 Aug 2010 01:53
Location: North Yorkshire, UK

Re: How to answer this question of phonology?

Post by sangi39 »

Yeah, there's several different analyses for this one (as there is for, say, the past tense ending in English).

One is that |z| is the basic morpheme, becoming /ɪz/ after a sibilant, and /s/ after a voiceless consonant, and then surfacing as /z/ elsewhere.

Another (which I think is bases on diachronics) suggests that the underlying morpheme is |ɪz|, which loses the vowel when the preceding sound isn't a sibilant, and then becomes voiceless after a voiceless consonant.

A third one takes |s| to be the morpheme, which gains /ɪ/ after a sibilant, and and then the /s/ becomes voiced after a voiced sound (either a voiced consonant or a vowel).

I think which one people choose is down to convention (which one were they taught), which one uses the fewest rules to go from the morphophoneme to the surface form, which set of rules seem to comply with other phonological rules more generally within the specific languages, etc., but from what I can remember a fair number of people use it more as an analytic tool to demonstrate that what seems to be different morphemes are probably the same morpheme, surfacing in different ways in different conditions (so it sort of simplifies paradigms), so as long as the rules they come up with for that specific morpheme makes sense, are consistent, and can describe observation, it probably doesn't matter too much which "underlying form" you pick.
You can tell the same lie a thousand times,
But it never gets any more true,
So close your eyes once more and once more believe
That they all still believe in you.
Just one time.
Salmoneus
MVP
MVP
Posts: 3033
Joined: 19 Sep 2011 19:37

Re: How to answer this question of phonology?

Post by Salmoneus »

I expect that if your teacher set that homework for you, it's because they expect that you can answer it yourself. And if you can't, leaning on getting the answer off of's the internet rather than actually getting yourself to the place, educationally, your teacher expects you to be is not going to help you much in the long-run.




-------

That said, if we're going to be a cheating on your homework facility....



Sangi: I have some nits to pick. Three of them, actually. Sorry for the brusque tone here, I'm just jotting the points down before I have to do something else, please don't read any aggression into it...


First, if we're really going to do all this "underlying form" malarkey, and taking your analysis at face value, then it's not just a matter of convention, but absolutely literally meaningless to say that it's either |s| or |z| underlyingly. Your larynx isn't "underlyingly" humming, or not-humming! Where the "surface form" is 100% predictable in some dimension from the surrounding elements, it is meaningless to ascribe that dimension to the underlying form. The fact that the IPA embeds a choice between fortis and lenis versions of many consonants and doesn't let you choose a genuinely neutral, unspecified form is a flaw in the notation system (so I'm not blaming you for it!), but the real answer is, or would be, that it's unspecified for voice and takes its voicing from the adjacent elements. Yes, you could say it's voiced and becomes voiceless or voiceless and becomes voiced, but these angel-on-pinhead counterfactuals have no meaning or use.


Second, if we're going to go down the "underlying form" route, and if we're going to ascribe voice (etc) to an imaginary abstract morpheme, then the only sensible analysis is surely |z|.

We can discard |Iz| quickly. Assuming the vowel means that we need to assume a rule that /I/ is lost after any non-sibilant. This clearly isn't a phonological rule in English, so it would have to be a morphosyntactic rule and we'd have to assume one deep underlying morpheme, TWO subsurface morpheme forms not phonologically determined (but phonologically predicted), and only then the actual surface forms (taking into account things like voicing of the onset of the following syllable). This is not parsimonious. Whereas, if we assume a bare sibilant morpheme, we can introduce a simple phonological rule that you get epenthetic /I/ between what would otherwise be two consecutive sibilants in the coda of a syllable. You then need only one pre-phonological form of the morpheme.

Now, you could say that you need an additional phonological rule for this, which is just as bad. But I'd retort: no, it's not just as bad, because you're minimising the number of 'levels' or 'stages' of explanation needed. But also, I think that this rule probably shows up sometimes in loanwords AND, crucially, it also applies to the possessive clitic -s. If you go the |Iz| route, you need to assume two forms of BOTH the plural morpheme AND the possessive (or, you have to assume the two morphemes are the same, which introduces it's own big bucket of worms). Whereas assuming one phonological rule breaking up double-sibilant codas deals with BOTH these problems at one stroke. AND, come to think of it, third person singular verbal -s, which follows exactly the same rule ('hisses', not 'hisss').


That leaves us |s| and |z|. Now, if the situation were symmetrical I'd agree either 'theory' is as meaningless as the other. But they aren't QUITE symmetrical. Because sometimes, as in the pair knife-knives, you get back-assimilation instead. And while 'forward-assimilation' is symmetrical - both voiced and voiceless coda consonants project their voicing onto the suffix - back-assimilation is asymmetrical, because although voiceless codas can voice through assimilation with the suffix (knife-knives), voiced codas never devoice through this assimilation.

This asymmetry can be explained if you assume an underlyingly voiced suffix, but it can't be explained plausibly if you assume a voiceless suffix (because "clusters of two voiceless coda consonants spontaneously voice" isn't a very likely rule).




But, third, all these theories are wrong. There isn't an underlying form, there's just multiple suffixes.

How can we tell? Because the surface form is not predictable from the supposed underlying forms. Therefore they aren't really underlying forms.

Let's imagine an English word: 'boof'. What's the plural - is it /bufs/ or /buvz/? Without lexical information, there's no way to tell. Some -f words take -s (gaffs, tiffs, coughs), and some take -z (knives, roofs) (and likewise with -th). A lot of these words can take EITHER -s OR -z, depending on a complex soup of dialects and registers (mouths, hoofs, etc), so that the same person may choose differently in different circumstances.

Now, when no phonological or morphophonological rule predicts the surface forms in the standard dialects from a single underlying form, when different dialects have different surface forms (often still not predictably), where the same individual chooses different surface forms based on sociocultural context, and where the 'dialect' boundaries are so complex that different individuals from almost identical contexts who otherwise consider themselves to speak the same dialect can default to different forms, and in a way that, again, is still not predictable by any simple rule....

.... we're not really talking about one morpheme modified in its realisation by phonological and morphophonological rules, we're talking about two different morphemes.

Need further proofs (with either /fs/ or /vz/, depending on taste)? Well, here's a clear citation:

Sneaky little hobbitses, wicked, tricksy, false

What's going on here? Well, if you have two different plural morphemes in a languages, one thing you expect to see happen is language errors involving double-marking, using both plurals on the same word. And this is exactly what you do see - with children, non-native learners, people joking about, drunk people, you get what Tolkien's emulating here, occasional -ses double-plurals (in the same way you also get the odd "oxens" and "fungis" (NB you always get the |z| form last, suggesting that it's the default plural morpheme, and that these double plural morphemes when people produce a plural and then hypercorrect because it doesn't match their knowledge of the prototypical plural morpheme)).

So the obvious answer is that -s and (back-assimilating) -(I)z are just two different morphemes, largely but NOT entirely predictably distributed, and both double-pluralising and the variations found in some words are demonstrating the uncertainties that can arise when speakers aren't sure which of the two morphemes is appropriate.


------


But I'm not a linguist or anything, so the OP shouldn't use my answer in their homework!
User avatar
Creyeditor
MVP
MVP
Posts: 5091
Joined: 14 Aug 2012 19:32

Re: How to answer this question of phonology?

Post by Creyeditor »

Okay, just a quick comment on the discussion. As I suggested in my first comment this is allomorphy, i.e. it is not derivable by general regular phonological rules, as Sal correctly pointed out. This means that the rules that Sangi describes have to be morpheme specific, i.e. only apply to the plural suffix. This becomes especially clear if you compare it to the genitive clictic -s and the third person -s. Another way to put this (and a slightly different analysis) is that there are three allomorphs that are inserted under different phonological conditions, as described by Sal and Sangi.
The forms like knife-knives were not part of the original data set and I think a lot of people tread them as exceptional stem allomorphy. That would mean that `boof' would always become `boofs', because only existing words can be exceptions to a rule. There is some research on this, and I think the reality is slightly more complex, because actual speakers derive plural forms of new words by analogy sometimes. So you could get `booves' in analogy to `hooves'. So, treating these as non-exceptional might mean that you have to formulate more complex morphophonological rules that have to take the phonological and lexical properties of the stem into account. In any case, there is just one abstrast plural morpheme and three allomorphs (contra Sal), with the conditions for insertion of the allomorphs being describable. Doubling of morphemes can happen, but it is not necessarily evidence of two morphemes, unless `hobbitses' would have a different meaning from `hobbits'. Another possibility is of course that some morphemes can apply more than once, as you can find it in voice alternations (passives of passives and applicatives of applicatives). Actually, if we follow the doubling argument, the default allomorph seems to be /Iz/. Depending on how you derive plural semantics, this might actually make sense (a set of sets of hobits or something similar).

I am a linguist, but still not infallible, especially around midnight [;)]
Creyeditor
"Thoughts are free."
Produce, Analyze, Manipulate
1 :deu: 2 :eng: 3 :idn: 4 :fra: 4 :esp:
:con: Ook & Omlűt & Nautli languages & Sperenjas
[<3] Papuan languages, Morphophonology, Lexical Semantics [<3]
Khemehekis
mongolian
mongolian
Posts: 3885
Joined: 14 Aug 2010 09:36
Location: California über alles

Re: How to answer this question of phonology?

Post by Khemehekis »

What about the allomorphs for the past tense/participle -ed? After another /d/ or /t/, you get /ɪd/; after a voiceless consonant that isn't /t/ you get /t/; elsewhere you get /d/. Sort of an analogous set of rules to /s/ vs . /z/ vs. /ɪz/.
♂♥♂♀

Squirrels chase koi . . . chase squirrels

My Kankonian-English dictionary: 87,413 words and counting

31,416: The number of the conlanging beast!
User avatar
sangi39
moderator
moderator
Posts: 3024
Joined: 12 Aug 2010 01:53
Location: North Yorkshire, UK

Re: How to answer this question of phonology?

Post by sangi39 »

And this is why answering this sort of question is fun as hell [:D] (not sarcasm)

I'm going to go with the same assumption Sal has, in that this is probably "homework", in that the wording of the question seems very "set by teacher" or drawn from a textbook (it's fairly close to the sort of "think of examples" questions that a book on linguistics in my college library had in it, where it would describe a feature, and then give you some other examples to work with, and set you off on figuring it out, based on what you'd just been told).

That being the case, the "correct answer" would probably depend on the information they've been given as a foundation for their homework (assuming, of course, that it has been set as homework, since so far we haven't actually been told that it has been set as homework at all, and not just, for example, curiosity based on casual reading of a linguistics book in their college library (the book I mentioned, for example, wasn't used once by the A-Level English language teachers, and very little of what was in it actually came up, and I'm not actually sure any course on offer at my college even required it, so I honestly have no idea why it was there, at least as a direct aid to studying the actual syllabus*)).

If it hasn't been set as homework, then Sal definitely covers the "answer" in a level of detail that should lead to a ton of reading (the idea of "underlying forms" is a topic in and of itself that's generated a mass of research and papers, IIRC, and things like allomorphy and morphophonology are equally as vast in terms of available reading). As Sal rightly brings up, examples like "knives" and "leaves" throw a bit of a spanner into the whole idea because they now involve two processes going on to indicate plurality (oh if only "mouse" became /maizɪn/ instead of "mice", that would be utterly brilliant for messing around with this sort of thing).



*From what I can remember of that one book, it was basically "the plural morpheme is -|z|, with three allomorphs, -/z/, -/s/, and -/ɪz/ applied in these environments" (rather than ascribing specific sound changes to explain how those forms arise, more "if a noun is of form A then suffix X is attached).
You can tell the same lie a thousand times,
But it never gets any more true,
So close your eyes once more and once more believe
That they all still believe in you.
Just one time.
Post Reply