What do you think are the most important results in phonology? What have we learned and why is it important?
I spend a lot of time with non-academics and so I’m often pressed to explain what phonology is and why anyone should care. Beyond, “all knowledge is interesting” and vague statements about cognition or computer applications I sometimes have a hard time figuring out what to say. What are your thoughts?
Practical applications? Hmmm, yeah, well, that’s not my area. :)
What do you think are the most important results in phonology?
What a difficult question to answer… at the risk of sounding circular I say the most important result is that there is such a thing as phonology, and that languages use a set of abstract categories (i.e. phonemes) that mediate between sound and meaning. That’s not what I usually tell non-specialists but maybe it should be.
I sometimes mention how phonology helps explain historical change and dialectal variation. Those are phenomena that most people are familiar with and may have idly wondered about: why is ‘knife’ spelled with a ‘k’? Why do some people say ‘filum’?
Historical reconstruction is another area that many non-academics can see the value of. Many ancient languages would be less well understood if we didn’t know about regular sound change. For example, some obscure words in the Bible have been clarified through comparison with other Semitic languages. Comparative linguistics like this can’t be done without phonology.
When I have to answer the question I first reject the implication that I need an application. I’ve used my education to find beautiful patterns in places that a rational community agrees are worth exploring. We’ve just chosen a very small community.
And of course some of us were taught at a young age to worship the venerable St. Bede and we’ve started on the path to gain our own immortality. Because one day in the distant future our description of a current language may answer questions about pronunciation that helps to decipher a weird hieroglyphic Archie comic.
Sometimes I do use what I’ve learned to explain those common questions that make so many ears perk up. “Why do we have silent letters?” “Why is that accent so hard to fake?” “How come Americans are the only English speakers without an accent?” (I once had to stop an argument that was functioning on this premise; they agreed that there was no accent but disagreed about the reason for this “lack”.) It’s so much fun to tell people something about their native language that they never realized was happening.
And every once in a while you get that awed reaction to the revelation (often mixed with some relief — as if they’ve been waiting their whole life to know why this or that happens) that itself answers the question of phonology’s application.
Our currency might be a limited one. The exchange rate is woefully unfavorable when we travel out of our beloved land. But when I hear someone aspirate a stop when speaking Spanish — or I hear a light ‘l’ in the coda position — or I hear an unstressed vowel instead of a pure syllabified liquid at the end of a word — my own ears start to tingle. And I’ve chosen to scratch this itch. I’ve found a community that has agreed to pay me to do that.
I’ve had a relatively easy time justifying my academic interests to non-academics since I joined a speech and language therapy department in 2005.
The logic is rather obvious: phonology/phonetics are part of (clinical) linguistics, which provides speech & language therapists with essential assessment tools; speech and language therapy is a, to many people obviously, important discipline.
Similar arguments can be made for several other applied fields, including language (proficiency) teaching, and speech technology (though it helps to be a phonetician for those)
Whether or not academic pursuits should be justified on grounds of practical application is a wholly different matter, as is the question to what extent the results you or I deem important (currently) feed into any practical applications.
Wouter Jansen brings up a very good point. Speech therapy is a major area where I think the study of phonology has significant (and significantly interesting) applications — at least potentially, since (as I’ve been told by Jessica Barlow) there aren’t as many speech therapy students interested in linguistics as one might like/hope (and certainly vice-versa).
All very interesting and thougt-provoking answers. Although I feel I’ve muddied the question by introducing “applications.” Actually, what I’m more interested in is in the vein of Bob Kennedy’s answer. That is, “what have we learned from studying phonology?” in terms of phonology.
Well, let me troll (pun intended). One could imagine all kinds of answers along the lines of the inherent capabilities humans have for language, say. But it seems to me that this is much easier to show for syntax than it is for phonology.
So let me throw out a wild idea: what if (nearly) everything since the beginning of generative phonology turned out to be wrong? No phonological features, hence no arguments over underspecification, what is the right set of features, etc. No OT (at least not in phonology). Instead, things were pretty much as the American Structuralists imagined–there are rules (both allophonic and morphophonemic), but they’re expressed in terms of atomic segments (phonemes, maybe morphophonemes) rather than features; and maybe they can be ordered, or prioritized in some other way. (The structuralists’ allophonic rules were not ordered.)
You can do a more or less observationally adequate phonology of languages with the sort of machinery I’m suggesting. The arguments for generative phonology are more about descriptive or maybe explanatory adequacy.
Now I’m not sure I believe this, and I’m pretty sure it’s wrong when it comes to stress (but even Chomsky and Halle didn’t treat stress as a binary feature, so it’s always been the odd man out). And it’s taking us rather far from the original question–except that I’m asking whether any of the results in phonology in the last forty years or so are right.
BTW, computer speech recognition uses little or nothing of modern phonology, AFAIK. And the only computational tool that does phonological rules that I know of is the Xerox xfst toolkit (or similar programs), and that uses segments, not features. (The rules are extrinsically ordered.) Its main application is building morphological parsers.
In the past few years, I have been teaching a course on the History of Linguistics, and in the last class of this course I discuss the history of generative phonology by following one topic — vowel harmony. Every time I am teaching about it, it strikes me that you can really see that there have been ‘results’ here, i.e. there has been progress: if you read a little bit about it you end up knowing more and understanding more than the greatest experts of 50 years ago. There are still many mysteries and probably no question has really been completely solved so far, but we really have more insight into at least this one phenomenon than fourty years ago. For instance, it seems to me that we have a better understanding of the typology of vowel harmony systems (what is possible, what is impossible), of the reasons why there is more vowel harmony than consonant harmony, of the possible interactions with morphological constituency, etc. Further, it seems that both representational theories (autosegmental phonology, but also various versions of grounded phonology) as well as theories of language variation, such as OT, have had something to contribute to this insight. Maybe it is possible to describe the facts of, say, Hungarian vowel harmony in terms of structuralist phonology, but it seems to me that an average paper on ROA on the topic really would show more insight, even if it were written by a not-very-advanced graduate student.Furthermore, with the exception of a brief period in the 1980s, vowel harmony never was at the forefront of phonological research. There are other topics where probably even more advances have been made.
To me, it seems like some salient points of the field are: that sound change is regular (both diachronically, and from UR to surface form); that some sound changes are more common and natural than others; that sometimes there is an explanation for naturalness in terms of acoustics or perception or articulatory demands; and that sound structure seems to involve abstract units like segments, moras, syllables, and feet, rather than just periods of noise.
As long as I’m playing Devil’s Advocate here, let me just make a comment (more of a query, actually) on part of Marc van Oostendorp’s response:
Are the insights in any way dependent on representational theories (autosegmental phonology,…) or on OT? That is, can the generalizations be expressed in theories other than those they were discovered in, or successors to those theories? My suspicion is that they can be expressed in other theories, maybe even structuralist approaches, although the explanations for the generalizations might be different–e.g. Blevins’ evolutionary approach.
Putting this differently, a theory s.t. like structuralist phonology could well be observationally adequate, and maybe descriptively adequate (although descriptive adequacy seems a bit vague in phonology to me). Structuralist phonology would probably not come up with an explanation for the phenomena, but maybe it doesn’t need to–if the explanation can be found in some other domain (I’m thinking of Blevins again, although I have to admit that I just ordered her book the other day).
I wait to be enlightened.
Thanks to Ed Keer for posting this interesting question and to Mike Maxwell for leading it in a provocative direction. Could I take a moment to defend theory? I have two defenses in mind.
First, I think there are interesting factual generalizations that almost certainly would not have been discovered in the absence of a theory to guide the inquiry. I cite two from Donca Steriade’s work:
Steriade’s Law of Rounding Harmony: When rounding harmony is blocked by a height shift (as in Turkish), the target vowel surfaces as [-round].
Steriade’s Law of Translaryngeal Assimilation: When a rule of vowel-to-vowel assimilation applies across glottal consonants and no others, it also applies in hiatus.
Of course, I’m not sure these laws are true, but I know no counterexamples and think they are nontrivial. The point I want to make is that Steriade probably wouldn’t have discovered them had she not been using a theory, with rules and derivations, to guide her inquiry. Indeed, these generalizations presuppose rules in the general sense, and thus aren’t even statable without some sort of phonological theory.
A second reason not to be theory-averse is that, if you’re interested in modeling human phonological learning, it becomes very difficult to make any progress at all without the use of theory. Even the famous 1986 Rumelhart/McClelland past tense simulation, held to be a case of pure phonological inductivism, actually included a (not very good) phonological theory, as Pinker and Prince pointed out in their 1988 Cognition review.
This point is currently on my mind because (plug, plug!) I’ve just finished a paper on phonotactic learning with Colin Wilson, posted on ROA. In our work, we repeatedly found that theoretical ideas from the last few decades make it possible to do phonological learning that would otherwise be impossible. In particular, we found that under our approach, phonotactic learning requires use of feature theory, underspecification theory, autosegmental theory, and metrical grid theory.
Of course, we’d be fascinated if others can prove us wrong, and to this end we would be happy to share our simulation data with anyone who is interested. My point is that when you work on a learning simulation, where there is a clear criterion of analytic success or failure, it can give you a vivid and renewed appreciation of theory, because theory often provides you with the tools that you need.
I guess I should clarify my posting, in light of Bruce’s comments. I completely agree that a theory is necessary to make progress (I could say a lot more about typological linguistics, but I’ll refrain). My question about Marc van Oostendorp’s posting was rather whether the discoveries about vowel harmony etc. were dependent on the particular phonological theories that people were working in (variants of generative phonology). In other words, might not these discoveries have been made if people were still doing structuralist phonology?
So I’m not trying to be theory-averse, just (at least from my devil’s advocate position) averse to a particular theory, namely generative phonology with innate/ fixed distinctive features, theory-particular notions of natural classes as being defined by a single node in a feature hierarchy, etc.
I’m sure we could go on speculating for a long time about alternative realities, and what discoveries might otherwise have been made if SPE hadn’t taken over the world. I don’t think we should go there. Rather, I’m asking to what extent the “important results” (Ed Keer’s original term) that have been made in phonology are dependent on the particular theories that were being developed. Bruce does address this in the paragraphs about phonological learning. So I’m off to see his ROA paper!
Bruce,
Thanks for your post! I’m looking forward to reading your and Colin’s paper.
I see two types of answers to the “major results” question I asked. One is that we’ve learned (or have we shown?) that there is such a thing as phonology–abstract representations of knowledge about sound systems. And the second is that we’ve really learned a lot more about the types of processes that occur in phonology–for example vowel harmony.
Does that sound like a fair summary?