On teaching phonology without the phoneme

I find the juxtaposition of the opening and closing sentences of Rick Wojcik’s comment on Bruce Hayes’ post about teaching allophony remarkable:

How can you possibly teach an introduction to linguistics without teaching the basic concept of the phoneme? […] Let them puzzle over those questions, not some dry lecture on distinctive features.

Bruce’s post asks a sincere question, and the causticness of Wojcik‘s comment, clear enough just from the two sentences quoted above, seems unwarranted as a response (to say the least).

First: I don’t appreciate the presuppositions snuck in by the “basic concept of” and “dry lecture on” — in my experience as an undergraduate, the noun phrases modified by these words could well be switched: “How can you possibly teach an introduction to linguistics without teaching the basic concept of distinctive features? […] Let them puzzle over those questions, not some dry lecture on the phoneme.”

Second: as if “the phoneme” and “distinctive features” were the only two options on the table! There are plenty of other phonologically-relevant topics that can be presented as a unit in an introductory linguistics course in such a way that deep and important issues in the field are conveyed in an accessible manner to students whose exposure to linguistics before that course (or ever again) is close to zero.

Below the fold, I take apart the middle part of Wojcik’s comment, breaking it into bite-sized chunks in order to either question the relevance of the phoneme to that chunk, or to point out the relevance of at least one topic other than the phoneme (or distinctive features, for that matter) that I think would make a fabulous phonology unit in an introductory linguistics course.

The point of an introductory class in linguistics is not to begin teaching people advanced analytical techniques, but to teach the rudiments of language.

As other commenters on Bruce’s post pointed out, it’s the notion of contrast that is a “rudiment of language” — phonemic analysis is an “advanced analytical technique” used to express the difference between contrastive and non-contrastive elements. So why not a unit on contrast, emphasizing the importance of phonetic context and so forth?

It is the first time most students will have encountered articulatory phonetics and become conscious of the essential difference between spelling and pronunciation.

I’m not sure how awareness of this “essential difference” leads naturally into phonemic analysis. On the one hand, if what you’re calling “pronunciation” is closer to the phonetic level, with allophones of phonemes distinguished, then talking about phonemes just adds a layer of complexity between spelling and pronunciation. (I’ve found this to be especially true when teaching Spanish phonetics and phonology using a typical structuralist textbook, which often includes rules for relating orthographic and phonemic representations as well as phonetic and phonemic representations.) On the other hand, if “pronunciation” is closer to the phonemic level, then the layer of complexity is added between pronunciation and finer-grained phonetics.

The most fundamental insight that a student can have about pronunciation is that there is a difference between the sounds that they think they are pronouncing and the sounds that actually come out of their mouths.

This can be taught showing (not necessarily allophonic) alternations, the most often used example (I think) being the past tense and plural suffix alternations in English.

It is a time for the class to become conscious of the differences in dialects.

Of course — but how is the phoneme necessary to achieve this?

That is why almost every introductory text starts out with the psychological phoneme (the fundamental insight of Baudouin de Courtenay in the 1870s) and progresses to a more classical version. The best that you can do in the short time that can be allotted to this subject is just to brush up against some of the problems with more classical versions of the phoneme.

I’d say that almost every introductory text starts out with the phoneme because … every introductory text has always started out with the phoneme. I take this to be the question Bruce’s point was meant to raise: should we teach things the way we always have?

(Also, I don’t really see what Wojcik’s initial “That” in the above chunk refers to — certainly not the immediately preceding “differences in dialects” bit?)

Leave it to an introductory course in phonology to get into more abstract phonological analysis.

The phoneme itself is a fairly abstract analytical concept; personally, I’d have a hard time telling you which is “more abstract”, the phoneme or the syllable — and I find that very basic syllabic analysis is easier to teach to novice linguistics students than phonemic analysis is.

Allophonic variation needs to be taught in an introductory class simply because it is an interesting linguistic phenomenon to become aware of.

Contextual neutralization needs to be taught in an introductory class simply because it is an interesting linguistic phenomenon to become aware of.”

Syllable structure needs to be taught in an introductory class simply because it is an interesting linguistic phenomenon to become aware of.”

Linguistic meter needs to be taught in an introductory class simply because it is an interesting linguistic phenomenon to become aware of.”

Utterance intonation needs to be taught in an introductory class simply because it is an interesting linguistic phenomenon to become aware of.”

etc.

Why don’t alphabetic spelling systems distinguish allophones?

Why do alphabetic and non-alphabetic writing systems both exist? Why does German orthography not encode word-final devoicing (Rad [rat] ~ Rade [radə]) while Turkish orthography does (kitap [kitap] ~ kitabɩ [kitabɨ])?

Why do words rhyme even when the phonemes are pronounced with different allophonic variants?

Rhymes are interesting for so many other reasons. For example: why must legitimate rhymes (in English) start from the stressed nucleus? (manila [mə’nɪlə] rhymes with vanilla [və’nɪlə], but neither rhymes with Pamela [‘pæmɪlə] or canola [k.ə’noʊlə].)

Or: why can some contrastive distinctions be ignored in rhymes? Recall (if you will) Tom Lehrer’s lyrics to “The Folk Song Army”, with Tom’s tongue firmly in cheek:

The tune don’t have to be clever,
And it don’t matter if you put a coupla extra syllables into a line.
It sounds more ethnic if it ain’t good English,
And it don’t even gotta rhyme–excuse me–rhyne.

In sum: I’m convinced the phoneme’s important, but it’s not that important. Someone will have to do better to convince me to teach it next time I teach intro.

10 thoughts on “On teaching phonology without the phoneme

  1. Nathan Sanders

    I’ve never had any significant trouble teaching the phoneme (and by this, I mean Bob Kennedy’s “abstract underlying contrastive segment of sound”) in intro, but I spend far less time on sociolinguistics, animal language, etc., so that I can spend much more time on phonology.

    Also, when I teach phonemes, I start off with a method I owe to Andy Wedel. I write upper and lowercase As and Bs on the board numerous times, in print, cursive, block letters, LED style, etc., repeating some of the styles with minute variation (for example, a slightly longer right leg on the capital A, then a slightly longer right leg). Students quickly accept that some of these symbols represent one thing (some abstract concept we call the letter A) and others represent something else (B). From there, phonemes (and allophonic variation) are pretty easy.

    This analogy seems so effective that I’m surprised it’s not in more common use. Have other people seen it? Used it themselves? Had particular success or problems with it? Find it pedagogically unsound?

    This doesn’t really address your concern about the importance of the phoneme, and I’m not sure I have a good answer. I try to give my intro students an accurate view of what they would be doing in the phonology course if they were to take it, and it seems disingenuous to have no discussion of abstract underlying segments.

    But if your goal for intro is different (say, keeping 200 undergrads with diverse interests sufficiently engaged with the course material for the duration of the term, with the valid expectation that the vast majority will never take a linguistics course again), then I can see why the phoneme isn’t necessary and can even be a hindrance.

    I guess I have a bit of a selfish and more cutthroat view, trying to groom the next generation of linguists and letting everyone else sink or swim.

  2. Bob Kennedy

    I think the original insight in Bruce’s post remains an interesting one – that in intro linguistics we could teach alternation without getting into phonemic analysis. Somewhere in that thread it became interpreted as teaching a phonology unit without referencing the phoneme … this is where I scratched my head, because I teach alternation (in intro linguistics) with the phoneme as an established concept.

    Like Eric, I’m also bemused at the assumption in Rick’s comment that an intro ling course that leaves out the phoneme would necessarily have to replace it with a discussion of distinctive features. I just don’t see how that follows Regardless, I think if you go back to Bruce’s original post, the only thing he suggests leaving out is phonemic analysis (but refers to this approach unfortunately as the “no-phoneme approach”). Taking this liberty in interpreting Bruce’s suggestion, I guess I’m willing to entertain having a phonology unit in intro ling that covers only alternation – with the phoneme as a tool, but without phonemic analysis beforehand. Then in intro phonology, I’d start with more alternation, and still more alternation, and maybe even do syllable structure before phonemic analysis. I’d even consider leaving distinctive features until after all of this.

  3. Bob Kennedy

    I should add a bit to the last part of my previous comment … this quarter I’m following the order of phonemic analysis, then features, then alternation and neutralization. I think the struggle with features (for some students at least) has obscured the big-picture point of looking at alternations.

  4. John McCarthy

    Some of the comments have raised the matter of teaching distinctive features. Even when I teach the phonology course for ling majors, I barely discuss the features. They occupy one lecture, the point of which is “Here’s a handout with the features. You will probably see them when you read things. I don’t expect you to use them.” I deeply regret the time I wasted in the past trying to get undergrads to use the features correctly.

    In my view, the features embody only one important insight about phonology as a whole: that it manipulates categories that are defined in phonetic terms. This point can be made in the transition from the phonetics to the phonology part of the course, without the features as intermediary. Historically, the features were important for another reason, the Evaluation Metric, but I doubt whether the EM plays much of a role in contemporary phonology courses.

    In a more advanced undergraduate course, the features can be introduced in the context of feature geometry, which offers additional insights that more advanced students can absorb and appreciate.

  5. Lameen

    “Why don’t alphabetic spelling systems distinguish allophones?”

    Urdu has a separate letter for retroflex-r, even though it’s an allophone of retroflex-d. Tashelhiyt poems in the traditional Arabic orthography write stressed phonetically long vowels long, although length is not a phonemic feature in Tashelhiyt (and all vowels were marked in any case.) Tiberian Hebrew vowel marking can be argued to distinguish allophones, although the fact that Hebrew had no native speakers when the Tiberian system was devised makes things a bit tricky. Even allowing for possible debate over the precise definition of “alphabet”, I’d be extremely wary about teaching such a claim.

  6. ellen broselow

    One method of introducing phonemes that I have not used, but would like to try, is to first have students participate as subjects in a categorical perception experiment, then present the results and ask students to try to explain the patterns. Ideally, the responses would be sorted by native language background, so that, for example, native speakers of English and Japanese would perform differently on a /ra-la/ continuum.

  7. Bruce Hayes

    Hmm, I think my remarks pressed some buttons for Rick Wojcik that I didn’t intend to be pressed. In fact, I think my intro teaching (both intro linguistics and intro phonology) is rather conservative. I try to get across what I take to be core, well-seated ideas, without including much current theoretical speculation. My original post concerned only an issue of pedadogical order; i.e. whether it would be more effective, in a very tight time frame, to use alternations as the way to start students out in phonology.

    Re. features: I find they are quite tough to teach, particularly since I think that there is no set of features that succeeds in doing everything we would like the features to do. So in intro, where there is so little time, I just employ the traditional phonetic labels ([stop], [vowel], etc.) as makeshift features. This doesn’t work too badly, and gets you into empirical material much faster.

  8. Rick Wojcik

    Hi, all. Sorry for the long delay in answering, but I just became aware of these responses. I’m going to just devote this to Eric, because it would be too lengthy to reply to everyone in this one post.

    I find the juxtaposition of the opening and closing sentences of Rick Wojcik’s comment on Bruce Hayes’ post about teaching allophony remarkable:

    How can you possibly teach an introduction to linguistics without teaching the basic concept of the phoneme? […] Let them puzzle over those questions, not some dry lecture on distinctive features.

    Bruce’s post asks a sincere question, and the causticness of Wojcik’s comment, clear enough just from the two sentences quoted above, seems unwarranted as a response (to say the least).

    Sorry if I sounded indignant, but Bruce’s question evokes some sharp controversies from the past. Sometimes a little passion is not a bad thing, but I apologize for any hard feelings that it may have caused.

    First: I don’t appreciate the presuppositions snuck in by the “basic concept of” and “dry lecture on” — in my experience as an undergraduate, the noun phrases modified by these words could well be switched: “How can you possibly teach an introduction to linguistics without teaching the basic concept of distinctive features? […] Let them puzzle over those questions, not some dry lecture on the phoneme.”

    My response is that the concept of the phoneme is too basic to leave out of an introductory course. Distinctive feature theory is probably best de-emphasized in favor of more time spent on articulatory and acoustic properties of sounds. Whether or not features need to be taken as “binary” or not ought to be left for an introduction to phonology, since the theoretical justification requires more than a casual acquaintance with phonological rules. Remember that you only have a few days to cover each basic topic in an introductory course. The idea is to give an overview, not to ponder fine points of generative theory.

    Second: as if “the phoneme” and “distinctive features” were the only two options on the table! There are plenty of other phonologically-relevant topics that can be presented as a unit in an introductory linguistics course in such a way that deep and important issues in the field are conveyed in an accessible manner to students whose exposure to linguistics before that course (or ever again) is close to zero.

    I don’t disagree about distinctive features, but that is your strawman here. On the phoneme, I draw the line. That concept has been with us since Panini’s time, and it dominated synchronic studies for well over a century. What makes me so mad is the realization of how badly Morris Halle’s flawed argument against the phoneme has bent phonological theory out of shape. (I know that many or most will disagree with me on this, but I’m prepared to defend the opinion.)

    The point of an introductory class in linguistics is not to begin teaching people advanced analytical techniques, but to teach the rudiments of language.

    As other commenters on Bruce’s post pointed out, it’s the notion of contrast that is a “rudiment of language” — phonemic analysis is an “advanced analytical technique” used to express the difference between contrastive and non-contrastive elements. So why not a unit on contrast, emphasizing the importance of phonetic context and so forth?

    I’m not opposed to this suggestion, but I would do it in the context of talking about phonemes. Phonemes are not an “advanced analytical technique”. They are a basic building block of linguistic theory. What about rhyme and alphabetical writing systems? Do you think that those concepts can be addressed adequately without referring to phonemes?

    I’m not sure how awareness of this “essential difference” leads naturally into phonemic analysis. On the one hand, if what you’re calling “pronunciation” is closer to the phonetic level, with allophones of phonemes distinguished, then talking about phonemes just adds a layer of complexity between spelling and pronunciation…On the other hand, if “pronunciation” is closer to the phonemic level, then the layer of complexity is added between pronunciation and finer-grained phonetics.

    Alphabetical writing systems ideally represent symbol-phoneme correspondences in the standard dialect of a language, not symbol-allophone correspondences. Why is that? Does English spelling violate the phonemic principle behind alphabets? How does your phonemic system differ from that of the person sitting next to you? These are interesting questions for class discussion. The idea is to get students listening to how they and others pronounce words differently–to raise phonological consciousness. That should be one of the major goals of any introductory course, not the study of theoretical constructs that will only be relevant to linguistics majors.

    The most fundamental insight that a student can have about pronunciation is that there is a difference between the sounds that they think they are pronouncing and the sounds that actually come out of their mouths.

    This can be taught showing (not necessarily allophonic) alternations, the most often used example (I think) being the past tense and plural suffix alternations in English.

    I think that there are some subtle problems with teaching the plural suffix as purely phonological, but that’s not something that bothers me here. What bothers me is that one really wants to focus on how Spanish speakers and English speakers perceive sounds. Voiced medial stops in Spanish are pronounced as fricatives, but Spanish speakers don’t perceive an interdental fricative, for example, as a sound distinct from /d/. English speakers do. How would you teach this fact without making reference to phonemes? More importantly, why would you want to?

    It is a time for the class to become conscious of the differences in dialects.

    Of course — but how is the phoneme necessary to achieve this?

    Different dialects have different phoneme inventories. If you want to understand a dialect map, this is fundamental information. Do people not teach students about geographical and sociological distributions of dialects anymore in introductory courses? How does one do that without talking about phonemes? I suppose you could just not talk about pronunciation differences between dialects, but I’m having real trouble understanding why anyone would leave out a key concept to understanding so much of the basic linguistic literature out there. The phoneme isn’t just for phonology classes. It has relevance in psycholinguistics, historical linguistics, sociolinguistics, field studies, etc.

    I’d say that almost every introductory text starts out with the phoneme because … every introductory text has always started out with the phoneme. I take this to be the question Bruce’s point was meant to raise: should we teach things the way we always have?

    That question can just as reasonably be asked of any linguistic concept taught in an introductory course. The question here is why the phoneme is singled out for capital punishment.

    The phoneme itself is a fairly abstract analytical concept; personally, I’d have a hard time telling you which is “more abstract”, the phoneme or the syllable — and I find that very basic syllabic analysis is easier to teach to novice linguistics students than phonemic analysis is.

    That’s a false dichotomy. Both phonemes and syllables need to be covered in an introductory course.

    Why don’t alphabetic spelling systems distinguish allophones?

    Why do alphabetic and non-alphabetic writing systems both exist? Why does German orthography not encode word-final devoicing (Rad [rat] ~ Rade [radə]) while Turkish orthography does (kitap [kitap] ~ kitabɩ [kitabɨ])?

    Great additional questions to pose to the classroom. Do you see these as questions that won’t get asked because you’ve taken the trouble to introduce the concept of the phoneme? I’m not sure what point you think you are trying to make here. When a student leaves an introductory course, that student should have more questions about language than he or she walked in with.

    Why do words rhyme even when the phonemes are pronounced with different allophonic variants?

    Rhymes are interesting for so many other reasons. For example: why must legitimate rhymes (in English) start from the stressed nucleus? (manila [mə’nɪlə] rhymes with vanilla [və’nɪlə], but neither rhymes with Pamela [’pæmələ] or canola [k.ə’noʊlə].)

    Again, why would the existence of additional questions prevent you from pointing out that rhyme is essentially phonemic. This was one of the fundamental insights that led to phonological theory (aka “phonemic theory”) in the first place.

    In sum: I’m convinced the phoneme’s important, but it’s not that important. Someone will have to do better to convince me to teach it next time I teach intro.

    That’s too bad. I think that you do your students a great disservice. At some point, at least some of them will wonder why you never covered the subject in your introductory class, and not all of those will be linguistics majors.

  9. Rick Wojcik

    Bob Kennedy said:

    Like Eric, I’m also bemused at the assumption in Rick’s comment that an intro ling course that leaves out the phoneme would necessarily have to replace it with a discussion of distinctive features. I just don’t see how that follows…

    Bob, it doesn’t follow, and I never said such a thing. You were confused by Eric’s attempt to set up a kind of reductio ad absurdum argument. See my response above to get a better understanding of my position. I actually think that is less worthwhile to talk about distinctive features than to talk about phonemes.

    Lameen responded to the following comment by me: (“Why don’t alphabetic spelling systems distinguish allophones?”)

    Urdu has a separate letter for retroflex-r, even though it’s an allophone of retroflex-d. Tashelhiyt poems in the traditional Arabic orthography write stressed phonetically long vowels long, although length is not a phonemic feature in Tashelhiyt (and all vowels were marked in any case.) Tiberian Hebrew vowel marking can be argued to distinguish allophones, although the fact that Hebrew had no native speakers when the Tiberian system was devised makes things a bit tricky. Even allowing for possible debate over the precise definition of “alphabet”, I’d be extremely wary about teaching such a claim.

    Lameen, I’m not familiar with the particulars of all of your examples, but I’m not surprised to find apparent exceptions to the rule. Who knows better than English speakers that spelling conventions can appear arbitrary and counterintuitive to speakers? However, the phonemic principle does explain many misspellings, and it is the basic explanation for the detection of many systematic scribal errors in old manuscripts. One can often tell when a manuscript was written by the systematicity of those errors. Urdu is not alone in that practice. Sanskrit devanagari script contains even more examples of apparent allophony, but that may be because ancient Hindu linguists were too good at phonetics. :) Anyway, you must admit that examples of allophonic variation in alphabetic writing are extremely rare and often questionable (as you admit in the Hebrew example). If one were to create an alphabet from scratch for an illiterate language, it would be nonsensical to include allophonic variation. Don’t you agree?

    Ellen Broselow wrote:

    One method of introducing phonemes that I have not used, but would like to try, is to first have students participate as subjects in a categorical perception experiment, then present the results and ask students to try to explain the patterns. Ideally, the responses would be sorted by native language background, so that, for example, native speakers of English and Japanese would perform differently on a /ra-la/ continuum.

    I think that any proposal along these lines is right on the mark. One needs to teach the concept of the psychological phoneme first before diving off into theoretical abstractions. Students need to be grounded in the basic insight before they try to explain it.

    Bruce Hayes wrote:

    Hmm, I think my remarks pressed some buttons for Rick Wojcik that I didn’t intend to be pressed. In fact, I think my intro teaching (both intro linguistics and intro phonology) is rather conservative. I try to get across what I take to be core, well-seated ideas, without including much current theoretical speculation. My original post concerned only an issue of pedadogical order; i.e. whether it would be more effective, in a very tight time frame, to use alternations as the way to start students out in phonology.

    Sorry if I misunderstood you, Bruce. Attacks on the phoneme do push some of my buttons. :-) I consider the loss of phonemic theory to be one of the greatest failings of the generative era. Halle’s flawed argument against the phoneme, and Chomsky & Halle’s attempt to co-opt Sapir set linguistic theory back almost a century, and that was in the 1960’s. We still haven’t recovered, and I suspected that is why you were even asking this question.

    Morris Halle should have known better, but he did not have a very good grasp of the history of Russian phonemic theory. Nevertheless, he published a very influential, but flawed, paper on the subject of the Moscow and Leningrad schools of phonology back in 1960. He was set straight by Reformatskii, one of the founders of the Moscow school, but the correction never made it into the collective consciousness of Western linguistics. Reformatskij’s rebuttal was never published in English, but it can be found in “Iz istorii otechestvennoi fonologii”, for those who can read Russian. Not only did the article gently crush Halle, but it laid the basis for rejecting his argument against the phoneme.

    Alternations are the basis of modern phonological theory, and they harken back to Baudouin’s original discovery of the difference between phonology (physiophonetics) and morphophonology (psychophonetics). So I would actually encourage you to teach the concept in an introductory class. Distinctive feature theory–that I would have a problem with.

    Re. features: I find they are quite tough to teach, particularly since I think that there is no set of features that succeeds in doing everything we would like the features to do. So in intro, where there is so little time, I just employ the traditional phonetic labels ([stop], [vowel], etc.) as makeshift features. This doesn’t work too badly, and gets you into empirical material much faster.

    Right. I very much endorse your judgment on this point, just not the elimination of the concept of the phoneme from an introductory course.

  10. Charles Jannuzi

    I really can’t understand the rationales on insisting that the phoneme is somehow fundamental to phonology (or phonetics). If it can not be found in articulation, speech perception or speech acoustics, its status as some sort of invariable unit of speech is laughable.

Leave a Reply

Your email address will not be published.