-1

The practice question is below. I am having trouble understanding what assimilation means in this context. Also, I don't understand why unbelievable is pronounced umbelievable when spoken fast via linguistic rules but not the other two give - ungrateful and undeniable.

An assimilation rule applies for many speakers of “English” when speaking fast. They pronounce certain “n”s as “m”s. For example: “uNbelievable” is pronounced as “uMbelievable”

However, certain “n” s are pronounced as the “ng”-(engma) sound, e.g. “uNgrateful” is pronounced “uNGgrateful” (“NG” is a single phoneme/gesture/articulation, namely ENGMA---I just can’t type it!))

But, in other environments the n remains an n, as in for example. uNdeniable which is pronounced uNdeniable

Using distinctive feature analysis, explain what’s going on here. Use Underspecification of the feature values on the underlying, stored, nasal phoneme of the prefix “UN—“. What is the assimilation rule that applies in these cases? Do your best to use formal notation."

Araucaria - him
  • 4,002
  • 1
  • 15
  • 39
user13434
  • 13
  • 2
  • I guess you mean "Also, I DON'T understand"? – Natalie Clarius Jun 17 '16 at 06:56
  • 2
    Also, I think it would have been nice if you had tried to come up with an attempt by yourself and asked more specifically at which point you got stuck - the way you made your question it seems a bit like you are asking us to solve the whole exercise for you. – Natalie Clarius Jun 17 '16 at 07:19

1 Answers1

5

Did you make sure you have understood what assimilation means in general, not only, as you said, "in this context"? I think this task is pretty straightforward if you stick close to the definitions and go through them step by step.

Basic definition

An assimilation is a phonological process by which a sound segment is made more similar to one of its neighbouring sounds with respect to at least one feature.
(Assimilation mostly occurs to make pronounciation easier - however, this is not part of the definition, this observation is perceptually-driven, as @user6726 points out.)
For consonants, those features are:

  • place of articulation
  • manner of articulation
  • voicing

Assimilation in this example

To solve the exercise, you now just need to check those conditions step by step:

  • place of articulation:
    • the original [n] is alveolar
    • the neighbouring [b] is bilabial
    • the resulting [m] is bilabial

At this point you are actually already done: You can see that the [n] is made more similar to [b] with respect to place of articulation, namely that [m] is a bilabial sound which [b] is as well. This is why the change from /n/ to [m] is an instance of assimilation.
This makes sense, intuitively: It is easier to pronounce the sequence [mb] where you lips and tongue are already in the right place rather than [nb] where you first have to move your tongue and then close your lips to produce both sounds.

We could of course also check for other features:

  • manner of articulation:
    • [n] is a nasal
    • [b] is a plosive
    • [m] is a nasal
    • => the manner of articulation is not assimilated to [b]
  • voicing:
    • [n] is voiced
    • [b] is voiced
    • [m] is voiced
    • => voicing is not assimilated either because [n] and [m] already are both voiced
      (This is only the case for this particular instance; of course, voicing might look different in sound clusters like /N-p/ or /N-k/. But since voicing is in general a feature set with respect to which consonants can be classified and compared, I thought it to make the reasoning steps more complete.)

You can now also see why the assimilations are different in the other examples:

  • In the case of "ungrateful", we have a neighbouring [n] and [g], where [n] is alveolar, [g] is velar and the reulting [ŋ] is velar => [n] is again assimilated to [g] w.r.t. place of articulation; this time not bilabial but velar.
    Obiously, it wouldn't make sense to turn the [n] into a [m] here when the next sound you want to pronounce is [g]; correspondingly, it wouldn't help either to make [n] a [ŋ] when you then want to pronounce a [b], in fact a [ŋ] would be even worse as its place of articulation is even further away.
  • In the case of "undeniable", [n] and [d] are both already alveolar - assimilation is applied, but vacuously, i.e. without it making a visible change to the previous version.
    Again, in this word [n] is not made a [m] because this wouldn't make it easier to pronounce the sequence [nd].

Other instances of assimilation

This example uses only place of articulation as a feature for assimilation.
There are of course other types, where e.g. unvoiced consonants become voiced between vowels.
Also, assimilation can not only apply to consonants but also to vowels: Vowel harmony, which occurs e.g. in Hungarian, Finnish and Turkish, makes vowels more similar with respect to horizontal tongue position, i.e. within a word there should only occur either only front or only back vowels, and when appending an affix (all of those three languages are agglutinative languages), there is usually two variants of them, one with a front and one with a back vowel where the one which more similarity (front/back) to the other vowels in the word is chosen, or if you want to use underspecification, there is one underspecified underlying affix which may be realised in a different way if the context requires so.

Underspecification

The underlying, underspecified phoneme is /N/ (note the capitalisation, reflecting its underspecified status) which is, by the use of a phonological assimilation rule, realised as the sound [m] in a context where it is followed by a [b].
Using undrespecification of the prefix, one could take "uN-" (/ʌN/) as the underspecified prefix, which then, by a phonological assimilation rule, gets realised as either [ʌn], [ʌm] or [ʌŋ] depending on the phonetic context.

Rule notation

For the rule notation, you could start out as follows:

/n/ -> [m] / _[b]

To make the rule a more formal and generative one, you should now try and abstract away from the concrete phoneme lables ([n m b]) by using distinctive features instead (such as [+cons], [LAB] etc.), but I hope you will be able to do this by yourself from what I wrote.

I hope this answers your question.
Again, I really think it would help to take a look at the definition of assimilation again and then just apply the feature checks step by steps, the difference between the examples becomes very clear then.

P.S.: To quickly type phonetic symbols (since one unfortunately can't TeX in the editor here...), you can use IPA typeit and then simply copy-paste the text.

Natalie Clarius
  • 6,499
  • 4
  • 21
  • 40
  • Errh, also note the instruction "Use Underspecification...", so maybe edit that answer. Since the prefix has no place, you'll have to reword your story. Also note that ease of articulation, which is actually irrelevant here (it's perceptually-driven) is not part of the definition of "assimilation". I'd get rid of the mention of voicing since assimilation also happens in /N-p, N-k/ clusters. And your "formal rule" is not a rule, formally speaking (what is a rule depends on the theory of formalism assumed by the instructor). – user6726 Jun 17 '16 at 15:34
  • @user6726 My intention was not to give a complete, readily copy-paste-able solution to the whole exercise but rather to address the OP's question of what assimilation means w.r.t. this example (and in general) so he could solve the particular instructions, also using underspecificaion, by himself. I can additionally write some more on that, but I intenionally tried to give a more general explanation so the OP would understand what assimilation amounts to (which is what he/she asked for) rather than giving the solution in full to all the single instructions. – Natalie Clarius Jun 17 '16 at 16:06
  • As for the ease of articulation part - yes, you are right, this is not part of the definition, I should have made this clearer. – Natalie Clarius Jun 17 '16 at 16:06
  • Getting rid of the voicing: I think you mean that my statement "voicing can/need not be assimilated because both sounds already are voiced" is not so generally applicable, because it would be different if it was a [p] or [k] that follows? Yes, I guess I should make this clearer as well. But I wouldn't remove the part on voicing completely, because this still belongs to the feature sets that consonants can generally be classified with and potentially undergo assimilation about - or isn't it? – Natalie Clarius Jun 17 '16 at 16:06
  • Formal rule: What makes my rule not an actual formal rule? As I said, one should replace the concrete phoneme labels by more abstract feature sets - but again, my intention was to provide a first approach, not the ultimate and final solution. Or is there something else that is formally incorrect? If so, please let me know. – Natalie Clarius Jun 17 '16 at 16:07
  • I edited my answer. Is it better now? – Natalie Clarius Jun 17 '16 at 16:27
  • What does it mean for a sound to be "more similar" to another? At first blush, it would be "removing differences", but is archiphoneme /N/ different from [m]? I'm suggesting fixing the answer so that it doesn't assert an underlying specification of place for the prefix. Much improved: get the piano into the apartment. – user6726 Jun 17 '16 at 18:17
  • In ordinary generative phonology or even phonemics, an assimilation does not have to be a change, so long as resultant sound segments are similar in whatever features are relevant to the assimilation. For instance, the example "undeniable" has assimilation of the first nasal to [d], even though the nasal needn't change to get from /n/ to [n]. (Underspecification theory might say something different.) – Greg Lee Jun 17 '16 at 18:23
  • @user6726 A sound being more similar to another means to me that they share more features. As [n] is featurally not equivalent to [m] w.r.t. the place of articulation, yes, I would say they are different. I don't really understand what you mean now. Is it about a phoneme /x/ not being comparable to a phone [x]? – Natalie Clarius Jun 17 '16 at 19:22
  • Also, I don't understand what you mean by " fixing the answer so that it doesn't assert an underlying specification of place for the prefix" - I mean I must choose some phoneme (and therefore some place of articulation) as underlying, which would most plausibly be /n/, and stated that the prefix /ʌn/ is only the underlying form which is then either realised as [ʌn] or something else, so how exactly would you like me to change the specification of the underlying prefix? – Natalie Clarius Jun 17 '16 at 19:23
  • @Greg Lee So you think that my answer suggested something wrong, or is this just another way to see it (as you wrote that underspecification theory, which is being asked for in the exercise, might have a different opinion on your point of view)? – Natalie Clarius Jun 17 '16 at 19:25
  • @lemontree, the OP's instructions are to frame this in underspecification theory, meaning the nasal is not specified for POA. To be consistent with that requirement, you must not pick an underlying POA value, so the prefix has /N/ and not /n/. And GL is correct that rules can apply vacuously. Even assuming underlying /n/ rather than /N/, it is wrong to say that assimilation does not take place in /undeniable/: there is no difference between input and output, but the rule applies. The original instructions are kind of poor, I have to say. 'the n remains an n should be 'the "n" appears as n. – user6726 Jun 17 '16 at 19:56
  • Oh, okay, I see the problem, both of them. I didn't realise your remark was about /N/ vs /n/ notation, and for the vacuous rule application, I should have known that. I'll change it. – Natalie Clarius Jun 17 '16 at 20:46
  • Notice also that in generative phonology, your formulation "/n/ -> [m] / _[b]" is inappropriate, since general formulations in features are used. and since phonological rules do not apply to phonemes. Instead of "/n/", the rule would apply to any of a set of sounds which includes [n] and other sounds as well. – Greg Lee Jun 17 '16 at 20:50
  • @Greg Lee I think I made it clear that my rule formulation is not the final formal form, I provided it as a basic scheme for how such a rule would generally look like (because I wasn't sure whether the OP is familiar with such rule notation at all) and said that one should in the next step abstract away from the phonetic labels by using feature sets instead. I made it more explicit now in my post. – Natalie Clarius Jun 17 '16 at 21:50
  • As for your second comment, I am confused - wouldn't this somehow contradict the point of having underlying phonemes /x/ which are then realised by different allophones [x] rather than assuming that a phonetic sound [x] changes to a different sound, as it would look like if we wrote [x] on the left-hand side of the rule? – Natalie Clarius Jun 17 '16 at 21:52
  • As to why I say that in generative phonology rules don't apply to phonemes, consider an entire phonological derivation with many steps, with each rule applying to the output of the preceding rule. Even if we use "phonemic" to mean "underlying form", obviously only the very first rule could apply to a phonemic form. Right? (But there is a non-standard variety of generative phonology in which allophonic rules can be restricted to apply only to phonemic forms. (cont ...) – Greg Lee Jun 18 '16 at 04:25
  • (cont ...) That is my theory, described here: http://linguistics.stackexchange.com/questions/15148/eliminating-intermediary-forms-to-account-for-production-and-perception ) – Greg Lee Jun 18 '16 at 04:26
  • I see the problem about the status of intermediate steps but don't want to read through the whole post you linked now; I was taught that in a sequence of derivations, the very first step is to be written like a phoneme /x/, the final result like a phone [x] and everything in between, i.e. those forms that serve both as the output of a derivation and as the input of a new derivation, should simply be written in plain without any slashes or square brackets. I found this a little weird too, to me a sequence of sounds should either be phonemic or phonetic, but that's what I found it textbooks. – Natalie Clarius Jun 18 '16 at 11:23
  • Writing the input to a rule with slashes in GP is a mistake, because it implies that the rule would not be applicable to non-phonemes. Actually, in GP, the phonemic (that is, underlying) status of a sound is irrelevant to the applicability of a rule. (Similarly, distinguishing between the output of the very last rule to apply and the outputs of other rules is also mistaken.) – Greg Lee Jun 18 '16 at 14:32
  • Okay, I think I understand your argument, and believe you that some generative phonologists would criticise my rule formulation. I'd still like to stick to my version now - because this is the way I have gotten to know it in phonology and don't consider my formulation to be wrong according to those theories, because you pointed out that underpec. theory (which is what the OP asked for) is not exactly the same as GP, and because I've made so many edits now that I fear this discussion won't ever come to an end where both of as are completely satisfied with the result. – Natalie Clarius Jun 18 '16 at 14:45
  • (I'm already prompted all the time to move this disuccsion to chat, so we should try to stop sooner or later). If you believe that my way is not the best way to do it, maybe post your own answer and explain why you would do it differently? As I said, I understand your point, but I think my formulation is not blatantly wrong according to underspecification theory and would like to keep it as-is, now that I have alread edited quite a lot. – Natalie Clarius Jun 18 '16 at 14:47
  • You should listen to the prompt. :P Comments are not for extended discussion; this conversation has been moved to chat. – Alenanno Jun 19 '16 at 14:59
  • 1
    Just wanted to thank you all (especially @lemontree). I was really overthinking a lot of this, but appreciate how in-depth you went - it actually helped me better understand not only this practice problem but other problems I had to solve later on. – user13434 Jun 20 '16 at 05:41