Linguist Forum

Specializations => Phonetics and Phonology => Topic started by: ncl62 on November 08, 2017, 05:19:04 PM

Title: Difference between phonological and morphophological rules
Post by: ncl62 on November 08, 2017, 05:19:04 PM
Can anybody help clarify for me the theoretical distinction between phonological rules and morphophonological rules?

For example, let's say there's a language with the distinct phonemes /s/ and /t͡s/. In this language, affricates never occur after a nasal. We have a morpheme -t͡sunu. When the morpheme n- is added as a prefix to t͡sunu, the result is nsunu. There is also a morpheme -sunu which, when combined with n- produces nsunu as well.

What should the phonemic representation of //n+t͡sunu// be: /nt͡sunu/ or /nsunu/? In other words, is postnasal deaffrication a phonological rule causing /nt͡sunu/ (the product of //n+t͡sunu//) to be realized phonetically as [nsunu] or is it a morphophonological rule causing //n+t͡sunu// to be realized phonemically as /nsunu/, which is then realized phonetically as [nsunu]?

I do not know how to tell whether this postnasal deaffrication is a phonological or morphophonological process.

I suspect it might be phonological on the grounds that no [nt͡s] occurs in the language, so the distinction between /t͡s/ and /s/ is neutralized in the postnasal context. However, the only way to demonstrate this deaffrication unambiguously is by combining morphemes, which would suggest that the phenomenon can only occur across morpheme boundaries.

Please share your thoughts and evidence. Thanks!
Title: Re: Difference between phonological and morphophological rules
Post by: Daniel on November 08, 2017, 07:07:28 PM
I would generally say that it depends on the theory and how you analyze the lexicon. Either analysis could work potentially.

Generally a morphological rule should only be posited if necessary. So if you are simply combining phonemes that change based on general phonological rules, there is no need to add morphology to that.

So assume the underlying representation is the one with the affricate in your examples, and then simplify that cluster based on the general phonological rule. No need to add allomorphs or different morphological tiers of phonological analysis.

Of course in some cases you might need to refer to morphology in your analysis if there is a difference between general phonology and special cases in morphology. Or there might be a theoretically motivated reason to do that.
Title: Re: Difference between phonological and morphophological rules
Post by: ncl62 on November 09, 2017, 05:37:04 AM
Thank you, Daniel. So the goal is to describe the system in the fewest rules, with a preference for the more general or automatic rules (the phonological ones).

Would it make sense then to posit an archiphoneme for either /s/ or /t͡s/ in a postnatal environment within a morpheme (as opposed to one created by the joining of two morphemes)? For instance, morpheme [ansa] could theoretically be either underlying /ansa/ or /ant͡sa/ and there is generally no way to tell which, unless perhaps another dialect does not engage in postnasal deaffrication and reveals the actual underlying composition. Or would that be unnecessarily complicating the system?
Title: Re: Difference between phonological and morphophological rules
Post by: Daniel on November 09, 2017, 08:44:18 AM
Well, what evidence is there that /ts/ exists at all? If you have some evidence, and the contrast is neutralized in that environment, then yes, that would make sense.
Title: Re: Difference between phonological and morphophological rules
Post by: panini on November 09, 2017, 09:19:22 AM
When the morpheme n- is added as a prefix to t͡sunu, the result is nsunu. There is also a morpheme -sunu which, when combined with n- produces nsunu as well.
The most important question that the analyst would have for you is, what is your evidence that the output is nsunu and not nt͡sunu? How do you know that affricates don't occur after nasals? The distribution you describe is common, and typically the fact is that fricatives and affricates don't contrast after a nasal, but you don't know exactly which thing it is that comes after a nasal. Perhaps only affricates (not fricatives) appear nasals. Perhaps an acoustic study of NC sequences supports the analysis: anyway, there has to be some argument in support of the conclusion you've offered.

The distinction between is a phonological and  morphophonological process is a rather old-fashioned one, valid only in certain views of linguistic analysis. Even then, there was no requirement that there also be morpheme-internal applications of a given phonological rule. It is nearly a logical necessity that such a rule could only be demonstrated unambiguously by combining morphemes, because non-alternating morpheme-internal sequences are always ambiguous: X derives by rule, or X is present underlyingly. You can sometimes satisfy the 'purely morpheme internal' desideratum and still get unequivocal evidence for the representation if there is syncope or epenthesis, e.g /ans-a/ → [ant͡sa], /ans/ → [anis].

If there are no roots in the language like either [int͡se] or [inse] (a complete gap) that there would be no potential morpheme-internal applications of the neutralization rule. In no theory that I am aware of does that mean that such a neutralization is non-phonological.

Perhaps look at Trubetzkoy's book to understand the basic theory of neutralization, which is what is involved in the situation which you describe.

Title: Re: Difference between phonological and morphophological rules
Post by: Copernicus on January 02, 2018, 01:19:55 AM
There have been a lot of attempts to describe the difference between phonology and morphophonology, terms that really go back to Trubetzkoy's Principles of Phonology.  However, the phonological theory that Trubetzkoy was working with had actually been started in the 1870s by Baudouin de Courtenay, who first coined the word "fonema" (phoneme) and kicked off the synchronic study of languages.  Baudouin was the first linguist to develop the concept of sound alternations, which became the basis for Trubetzkoy's distinction between phonology and morphophonology.  Baudouin called Trubetzkoy's "phonology" "physiophonetics", and his "morphophonology" "psychophonetics".  However, Baudouin was not a structuralist in Trubetzkoy's or Saussure's sense of that term  His physiophonetic and psychophonetic alternations had very distinct psychological functions, whereas structuralists tended to downplay psychological function or to view linguistic systems as primarily social in nature--i.e. related more to Saussure's perspective on language.

For Baudouin, physiophonetic alternations were distinct phonetic pairs that represented a single underlying phoneme.  The phonetic alternants could be purely allophonic or involve complete phonetic neutralization.  So Baudouin consider the s/z alternation in, say, Polish mróz [mrus]/mroza [mroza] ("frost"/"of frost") to be simply /z/, since Polish devoices final obstruents.  That is, the nominative form was /mruz/ for him.  He believed that Poles (like himself) believed that they were pronouncing a final [z] as an [ s] sound.  OTOH, Baudouin held that the u/o alternation involved two phonemes /u/ and /o/, because Polish has no physical constraint against pronouncing either of those two sounds in that position in the stem of the Polish noun.

So, turning to Trubetzkoy, there are some significant lessons to draw from these historical facts.  One is that Trubetzkoy includes both allophonic variation and phonemic neutralization as part of his theory of "phonology".  Unlike Baudouin, he did not see final voiced obstruents in Polish as phonemes.  Rather, he saw them as archiphonemes that were realized phonemically as voiceless in final position and voiced elsewhere.  Trubetzkoy, a Russian linguist trained in Moscow (not Baudouin's St Petersburg), was simply modifying Baudouin de Courtenay's take on physiophonetic alternations.  Another lesson to take away is that Trubetzkoy explicitly excluded morphophonology from his theory of phonology, preferring to treat it as being a distinct component of grammatical description.  That is because Trubetzkoy took it for granted that Baudouin's physiophonetic/psychophonetic alternational dichotomy represented distinct linguistic phenomena.  He bought into Baudouin's dichotomy, but he modified it to make sense within his "structuralist" framework.  Archiphonemics and morphophonemics belonged to distinct areas of the grammar.  This concept of a fundamental dichotomy essentially evaporated in other structuralist approaches to phonemics and was completely obliterated by Chomsky and Halle in the 1960s.

Now, regarding the problem mentioned in the OP, here is likely the way Baudouin would have analyzed it.  The process in question is sometimes referred to as "assibilation", where an affricate like t͡s reduces phonologically to [ s].  Baudouin would likely have analyzed the underlying word as the phonemic sequence /nt͡sunu/, where the psychologically real  t͡s phoneme corresponded to the [ s] alternant of that phoneme in pronunciation.  Trubetzkoy would have used an archiphoneme to describe the same alternation.

Finally, it is worth noting that Baudouin's theory of phonetic alternations evolved into three distinct Russian schools of phonology:  the Leningrad School, the Moscow School, and the Prague School.  All three schools would have come up with different analyses of the t͡s problem, but the Moscow School would have retained Baudouin's analysis most faithfully.  The Leningrad School did not allow for phonemic neutralization, so they shifted the boundary of phonology/morphophonology, lacking the "compromise" of Trubetzkoy's archiphonemes.  Trubetzkoy retained Baudouin's original physiophonetic/psychophonetic dichotomy, but he renamed it phonology/morphophonology and added the concept of archiphonemes in cases of phonemic neutralization.