I'm really confused about how to determine underlying representation. Every thing I read seems to contradict the last.

Trying desperately to solve this problem and I just seem to be going in circles with both 1 and 2. Any suggestions?

**In Maga Rukai, an Austronesian language spoken in Taiwan, certain prefixes show an alternation between a mid vowel and a high vowel.

  1) For each alternation, state the underlying representation. Justify.
  2) What determines whether the prefix vowel is mid or high? State the rule that would describe this.**

o ~ u dynamic verb marker

o-kamɨ ‘back’ u-cŋulu ‘connect’

o-dranɨ ‘make a road’ u-ɖmiŋi ‘make a road’

o-lapɨ ‘hunt without dogs’ u-lupu ‘hunt with dogs’

i ~ e negative marker

i-tukruː ‘not hit with stones’ e-kapliː ‘not hold with hands’

i-ulpuː ‘not hunt’ (with dogs) e-lalpɨː ‘not hunt’ (without dogs)

si ~ se to wear

si-slivi ‘wear beads’ se-kceŋe ‘wear pants’

si-krɨkrɨ ‘wear a necklace’ se-kcabu ‘wear leggings’

te ~ ti to make

te-kceŋe ‘make pants’ ti-slivi ‘make beads’

te-tovnaː ‘build a hut’ ti-kunu ‘make a skirt’

te-sdamraː ‘cook a side dish’ ti-krɨkrɨ ‘make a necklace’

ke ~ ki harvest

ke-teθo ‘harvest turnip’ ki-sito ‘harvest peanut’

ke-bləblə ‘harvest bamboo shoot’ ki-lpɨlpɨ ‘harvest peas’

ke ~ ki passive

ke-ɖoɖoː ‘be awaited’ ki-klukluɖu ‘be frightened’

ke-kθabɨ ‘peeled’ ki-tɨtɨ ‘be bitten’

3 Answers 3


You need to formulate two theories about the two things that alternate and compare those theories to decide which is best. One theory will be that the mid vowel is underlying and a high vowel is sometimes derived from it. The other theory will be that the high vowel is underlying and a mid vowel is sometimes derived from it. Formulate the rules required for the two theories, if possible, and compare the rules for naturalness (assimilations are good). If one of the theories cannot be made to work at all, because there is no way to predict the observed forms, your work is done -- the other theory must be correct.

If this doesn't work, you might have to resort to a more complicated theory in which the underlying form is different from both the surface vowels observed.

These are just general suggestions -- I didn't try to work your problem.

Good luck.


2) The rule for the Maga Rukai prefixes alternation is obvious: if the first syllable after the prefix has a high vowel (i, ɨ, u), you choose the high variety of the prefix (u-, i-, si-, ti-, ki-), but if the first syllable after the prefix has a nonhigh vowel (e, ə, o, a), you choose the mid variety of the prefix (o-, e-, se-, te-, ke-).

1) As for this question, it is your job to decide and justify. Use the answer by @Greg Lee, and also use this related question: How to determine which phoneme a group of allophones realizes?. Ths paper is also good, it has step-by-step solutions of problems similar to yours: Morphophonemics: determining underlying forms and rule interactions.

  • I think you mean high vowel, not high syllable. Commented Feb 6, 2015 at 22:39
  • @GastonÜmlaut - Yes, sure, a syllable with a high vowel. Do you think I should correct it?
    – Yellow Sky
    Commented Feb 6, 2015 at 23:54
  • Linguists will know what you mean, but I think it would be clearer that way so, yes. Commented Feb 8, 2015 at 2:01
  • @GastonÜmlaut - Ok, corrected.
    – Yellow Sky
    Commented Feb 8, 2015 at 2:09

I disagree with Yellow Sky's assertion that the answer is obvious. Given the accompanying instructions, that answer is also incorrect. [Change of heart]. Okay, maybe I misinterpreted the instructions and YS proposed rule is possible. It depends on what your instructor means by "rule". Conventionally, these kinds of problems require a specific formal mechanism to change the underlying form into the surface form, and a statement of correlation would not suffice. That is the point of requiring an explicit underlying form. But your instructor might also allow an observation of correlation without a specific "how" as constituting a rule. You can probably tell if you've gotten instructions in rule-formalism.[/Change of heart]

BTW, you should not be asking us to solve your homeworks.

My answer is an elaboration and generalization of Greg Lee's answer. The general strategy for solving these kinds of problems is to identify possible underlying forms, and pursue the consequences of each choice. If some root is always pronounced as [blah], then there is no reason to think it is anything other than /blah/. If it is sometimes pronounced [blah] and sometimes pronounced [balah], then you have a reason to think it might be /blah/ and a reason to think that it might be /balah/. Each assumption about underlying forms entails a different set of rules, so if the root is /blah/ then you need a rule of vowel insertion to account for the variant [balah]; conversely if the root is /balah/ then you need a rule to delete the vowel to get [blah].

Often, it turns out that one of these alternatives can be quickly ruled out because there are also roots that are always pronounced [blah] in the same context (e.g. [bunt] in German can be from /bund/ or from /bunt/). That kind of strong evidence disproves one alternative (though you have to check that the alternative representation isn't also likewise disproven, which would mean that you've missed an important fact). If you can't rule out one analysis on factual groups, then you have to appeal to conceptual reasons, such as the simplicity and generality of the rules that each theory entails. That then means you have to have a theoretical framework for evaluating such rules, and that is kind of instructor-dependent. The concept of formal simplicity is very simple, but naturalness is poorly defined and depends on a huge amount of factual knowledge about historical linguistics, articulation, perception, and observation of phonological processes across languages (kind of not available to you at this stage). Instructors differ as to whether they accept the idea of two equally-good solutions, and they may insist that you make a single concrete choice. My opinion is that is teaching bad science, but it happens, so find out whether "either A or B would work" is acceptable.

I would say that formalizing the rules isn't just an "if possible", it is an absolute necessity. Very often, students get an idea about an alternative rule but they can't formalize one of the the rules... because it isn't formalizable, with the given assumptions about formalization. That's a big clue -- an underlying representation that entails a "rule" that can't be formalized is wrong.

  • Do you mean, my rule is wrong? What's wrong about it? Is there an example in the task that violates my rule?
    – Yellow Sky
    Commented Feb 6, 2015 at 17:51
  • user6726 is saying that you didn't present the rule in the right way; you need to show the specific process that changes the underlying representation to the surface form. (For example, using features +high -high or something like that.) Commented Feb 6, 2015 at 21:09
  • @sumelic - After user6726 had edited his post I understood what user6726 meant. Still, I also stick to the point that we shouldn't do somebody else's homework here, so I just gave some guidelines, let Catie Baumgartner formalise the rule herself. Besides, I don't know what kind of schemes they use in their class to formalise the rules.
    – Yellow Sky
    Commented Feb 6, 2015 at 21:16

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.