Talk:Phonological change

From Wikipedia, the free encyclopedia
Jump to: navigation, search


This article needs much more context to explain how it relates to sound change and it needs better organization to present the topic in an structured way. Not all of it needs to be simple, but there at least should be a lead section that eases the uninformed reader into the topic and how it relates similar topics. For example, since I don't know what a phonological system is, I have no idea how this topic differs from sound change or why it is discussed as a separate concept. Is this just something that one person (Hoenigswald) came up with? - Taxman Talk 22:58, 21 November 2007 (UTC)

I have an observation and a counter-proposal. After a while, if I don't hear any objections, I might just start to act on this counter-proposal myself, in my copious free time.
  • Observation It seems to me, as a layman (more or less), that the content of the "Phonological change" article is describing an alternative way of analyzing a particular class of sound changes — changes to the phonemes, excluding prosody (such as tonogenesis, which is mentioned in the "Sound change" article).
  • Counter-proposal Create an article on Henry Hoenigswald, and dump the content of "Phonological change" into it. Before actually changing "Phonological change" into a redirect (to "Sound change", I presume), pare away the few links to "Phonological change" from other articles, deciding manually whether to send each wikilink to "Henry Hoenigswald" or to "Sound change". Pi zero (talk) 18:13, 6 February 2008 (UTC)

With due respect for the good intentions, I don't think that's a good idea. This isn't an article about Hoenigswald but about an immensely fruitful and much-honored scheme of analysis which brings simplicity and order into a field which had been largely pointilliste previously. And subsequently I might add. Any introduction to historical linguistics, from the massive Hock to slimmer volumes by Crowley and so on will cite eight or ten works by Hoenigswald, prominently among them the book (hard as it is to understand) and articles that propounded his analysis. Once you get away from "gee whiz" etymologies and so on, historical linguistics becomes a highly technical subject; self-described laymen are not necessarily in any better position to clarify things than a self-described layman is to clarify General Relativity or integral calculus. Nothing, however, in this article (if read in connection with articles on phonology and so on) is impossible for curious laymen to understand, I don't think. In fact, I read the objections, above, with some perplexity, since one of the great virtues of Hoenigswald's taxonomy is its clarity and simplicity. (I do rather wish he had called "primary split" conditioned merger, and called "secondary split" simply split. It was the one terminological matter that seemed to confuse at least a few of my students. Too late, now. Introducing heterodox terminology, except as a sort of "different way of looking at things" footnote, is bound to be mischievous.)Alsihler (talk) 01:57, 8 February 2008 (UTC)

Good, the material stays here. Now that that's settled, there's a problem here that needs to be fixed. The internal content of this article is all very well as far as it goes, ditto that of "Sound change", but the relationship between the two is aggressively baffling. This isn't about the clarity and simplicity of Hoenigswald's taxonomy, it's about the obscurity of the juxtaposition of these two articles, with their lead-ins as they current are. Fixing it doesn't require an expert in the subject — in fact, an expert might find it more difficult, because their expertise could make it harder for them to see what it is they're trying to fix — but it does require someone who knows enough about the subject to be able to figure out what the relationship between the two articles is, and who is able to explain it clearly.
How would you explain the relationship between these two articles? Note that "Sound change" is the article that most other articles link to; "Phonological change" is only linked from about half a dozen other articles, some of which may be just as confused about the difference between these two as most of us here are.
Also, I point out that your explanation of how Hoenigswald's taxonomy fits into the history of the subject would be appropriate content for the article proper — with suitable conversion into an encylopedic form, of course. Had the article explained this clearly, the question of whether Hoenigswald's taxonomy ought to be shuffled off into an article on Hoenigswald probably would never have come up. Pi zero (talk) 04:22, 8 February 2008 (UTC)

Great! Now that we seem to be sort of on the same page, we can talk.

It never occurred to me to put stuff like the beginning part of my Comments, above, into the article because, well, my own pedagogical experience is that my audience already knows what phonological systems are, etc., etc.

As for the relationship between Sound Change and Phonological Change, the latter refers specifically and explictly (and only) to systems, that is, the way sounds pattern in language such that some of them contrast (have the same or overlapping distributions) and some of them are in totally mutually-exclusive distributions, and how these distributional patterns are altered by sound changes.

Sound changes themselves are changes in pronunciation: phonetic assimilation, phonetic dissimilation, apocope, syncope, smoothing, diphthongization, haplology, epenthesis, and so on. These changes in pronunciation may have phonological significance in the Hoenigswaldian sense, but they don't have to. The most systematic of them (assimilations, say, or devoicing in word-final position) are typically primary splits or else mere changes. Apocopes (loss of final segments) can trigger secondary splits, if the lost segments conditioned something further to the left in the word. Dissimilations and other changes are typically without phonological effect, since they mostly occur in a word or two and at best alter the frequency of phonemes, and rather trivially, though, when regular, it's primary split. So a dissimilatory change like chimley for standard chimney does, to be sure, increase the count of /l/ and decrease the count of /n/, but not by enough to worry about. Nor does much of a rearrangement of the deck chairs take place when, say, in Spanish the sequence -mn- becomes -mr- with concomitant epenthesis of a -b- (so: nomne "name" > Sp nombre, since it only occurs in a handful of words, but it's a regular change, and an interesting one. (Come to think of it, Hoenigswald's taxonomy doesn't really have a slot for the epenthetic -b-, here; it's arguably a mere change in pronunciation.) Also, these changes are one-offs; more significant, phonologically, would be dissimilations that become part of the working phonology of the language, as in the dissimilation of aspirates in Sanskrit, wherein any reduplicating prefix to a root beginning with an aspirated stop has an unaspirated stop instead, a structure that applies still to roots that transparently came into the language long after the actual primary split (which is quite early) which the dissimilation fits into. Some phonologically-relevant changes don't actually have a name in a discussion of sound changes. Many of the innovations creating new allophones, for example (the aspiration of certain voiceless stops in English, say) are classifiable in Hoenigswald's system as "mere changes", but don't really have a name as a type of sound change, beyond something like "allophonic differentiation" (which is hardly ever so treated in a discussion of sound change).

I should explain that most phonological changes become phonological fossils, in a sense; a few become part of the actual phonology. An example of the latter would be the devoicing of final obstruents (stops and fricatives) in German, a primary split. That is, voiced obstruents in final position in German are literally unpronounceable. But most phonological changes do not become phonologized: the merger of intervocalic -s- with -r- in Latin, for example (a primary split), resulted in a highly visible pattern of alternation in Latin, but once the dust settled, speakers of Latin had no problem with pronouncing new intervocalic -s- arising in loan words (philosophia, basis, rosa) or created by other sound laws (causa, fũsus from caussa, fussus). A peculiarity of some wildly popular theories of phonology make the tragic mistake of equating the two, whereas it should be obvious with only a little thought that there is pragmatically annd empirically a vast difference between them.

Well, I don't want to overwhelm you with detail, but perhaps you see the point. If you don't say so, and I'll try again.Alsihler (talk) 02:03, 9 February 2008 (UTC)

"Recommended background for this course includes..." A familiar phenomenon, now that you mention it. Whereas, in contrast, Wikipedia articles migrate asymptotically, over time, toward aiding readers with no specific background in the subject.
If I'm quite following you, a "phonological system" is made up entirely of the arrangement of contrasts between phonemes, and is independent of what they actually sound like. In principle, the sounds of all the phonemes in a language could be completely rearranged without any phonological change at all, rather like a replacement code (well, okay, I see in Wikipedia these days it's called a "transposition cipher"). Is that right?
A point that still bothers me, a bit, is that the article on "Phonological change" refers several times to "sound laws". "Sound law" redirects to the article "Sound change", where there is a paragraph about the term that seems satisfactory in the context of the discussion of sound change; but it doesn't leap out, from that paragraph, why the term impinges on phonological change. I might hazard a guess that the term predates the sharp distinction between sound change and phonological change because it predates Hoenigswald's work, but that's only a guess. In untangling these articles, something will need to be said about that, because the word "sound" in the article "Phonological change" has contributed to the confusion. (I don't think eliminating the term from the article would be the right approach; I'd rather see it stay and be explained, since knowing that it might come up in further study of the subject is itself relevant.)
...and, speaking of guesses, let us not forget that both of these articles are tagged for not citing their sources. I can't really help with that, anytime in the next few months to a year, anyway, because I don't have ready access to (let alone familiarity with) reference materials on the subject. Pi zero (talk) 15:00, 10 February 2008 (UTC)

Before getting into a discussion of phonology, or of a way of looking at phonology, let me say that your hunch about about chronology is spot-on: looking at sound changes in terms of phonological structures is a very recent development in the history of diachronic linguistics. Indeed, our worthy ancestors, the real pioneers, as often as not spoke of diachronic developments in terms of letters. It's a nearly universal trait of linguistic naivete, I guess, exactly like speaking of pronunciations of the goin', comin', lookin' sort in English as "dropping the g", when in reality it's phonetically a matter of the fronting of a point of articulation. (By the same token, in amphi-Carribbean Spanish, the change of word final [n] to [ŋ] isn't a matter of "adding g's" but of backing of a point of articulation.)

In the beginning of historical linguistics, in fact, there WAS no theory of phonology, though some historians and students of language intuited some rudiments for themselves.

Now, as to phonology, the term is a blanket term for any theory of how speech sounds function as part of a code to form intelligible speech forms ("words" but also parts of words ... affixes and the like). There are all sorts of varieties of such theories, which have experienced, or are experiencing, major vogues and/or major desuetude. Most of the more sophisticated ones aren't actually of any use to historians, because they implicitly claim to be mapping "native speaker intuitions" and the like, which can't readily be checked with dead people. It can't readily be checked with living subjects, either, and the strain of linguistics I was raised in eschewed such "psychologizing" as adding nothing useful to objective observation. It has also been pointed out, to the annoyance of the acolytes of the more sophisticated theories, that you can't actually bring them into play until you perform some sort of "traditional" phonological analysis, since the elements manipulated by the more sophisticated theories bear an unmistakable resemblance to traditional phonemes (term defined below).

One fact accepted by all theories is that the actual inventory of describable speech sounds that come out of the mouths of native speakers is very large. Hundreds, thousands. Thus, in my American English there is an impressive collection of sounds that might be called "r": vocoid glides (oral and nasal both, and both rounded and unrounded), fricatives (in words like truck and drink) a voiced tap in three. But it is accepted on all hands that the meticulous piling up of phonetic details in this manner doesn't get you any closer to "what is going on" when people speak and understand one another; rather it actually seems to be getting further away from the nub of it.

All speech sounds consist of a bundle of simultaneous phonetic features: voicing, point of articulation, manner of articulation, duration, and so on. Crudely, and this formulation breaks down without further refinement (don't worry, I won't go there), some phonetic features in a bundle are predictable, i.e. you know they will be there ahead of time, and therefore have little communicative function (people used to say "no communicative function", but that is wrong; I'll touch on this again below), while other features are vital for distinguishing what was said from what might have been said instead. So, for example, take the features mentioned above in connection with English "r": when bundled with certain other features, voices vs voiceless is vital; oral vs nasal is vital; but when combined with the features that add up to "r", they're wholly predictable on the basis of which sounds occur in the immediate vicinity of the "r" (what is technically known as its environment). Thus, the difference between the words Mable and maple consists in whether the vocal cords are vibrating during the bilabial stop in the middle of the word. This isn't predictable, since there's nothing in the environment that "triggers" (to speak crudely) the phonetic difference. Similarlly, the difference between the utterances bat and mat inheres in whether the velum is open (mat) or closed (bat) during the segment defined by voiced bilabial occlusion. Compare this to the story of "r", which is nasal in words like barn, nearing, ran and any other word with a following nasal; is voiceless in truck, prick, cram and in any other word where it is immediately preceded by a voiceless stop ... unless there is an s before it, so in words like spring and string, the "r" is voiced.

Now: all these statements apply to one specific phonology, that of American English. Whether a phonetic feature distinguishes between utterances is a matter of its "distribution", which is by definition language-specific. A familiar example is the phonetic feature of nasality (open velum) in vocalic (vowel-like) segments. (What are familiarly called "sounds" are usually called "segments" by linguistics.) Both English and French have a distinction between oral and nasal vowels; but in French this distinction is contrastive, in English it's predictable (i.e., automatic). Thus, in English, ban, bum, bing, boon and so on all have nasal vowels; the vowels of bat, bud, big, boot are all oral. (There are complications, here, from the fact that the distribution of nasality in English is a dialect matter. In some Midwestern regions, for example, ALL tonic -- accented -- vowels are nasal, not just those followed by or preceded by a nasal consonant.) But in French sets like bon [bõ] "good [masc.]" and beau [bo] "handsome [masc.]", the only thing that distinguishes the two utterances is the open velum in the former and the closed one in the latter. That is, the communicative function is literally carried by the single feature of nasality.

There are some other differences, too; all English vowels and diphthongs are either nasal or oral, depending on environment; French has an elaborate oral vowel systems but only four nasal vowels. More importantly, while vowels in English are nasal in the environment of nasal consonants, in French, vowels before nasal consonants are oral, unlike English: there are a number of phonetic differences between French Beaune [bon] and English bone, but among them is the fact that the vowel nucleus is nasal in English and oral in French.

The traditional terminology here is that the specific phone types -- all those different r's in English -- are called allophones. The ones that are phonetically similar and in complementary (= mutually exclusive = non-contrastive = predictable) distribution make up a set defined as a phoneme. Note: you can't by definition pronounce a phoneme; phonemes are abstractions, sets of phonetically different segments. You can only pronounce bundles of specific phonetic features, i.e., allophones. What do phonemes, then, do? Phonemes are essentially units of perception. It's not really QUITE that simple, nothing in Linguistics is, but in a nutshell, we pronounce allophones but hear phonemes. When I say words with all these different allophones of "r" in English, what you HEAR is "the sound 'r'". (NOTE: for a long time it was stated or assumed that the phonetic features distinguishing allophones had no communicative function, all the "work" was being gone by the contrastive (non-predictable, non-complementary) -- briefly "phonemic" -- phonetic features. Not so. The allophonic phonetic details contribute massively, and critically, to the redundancy of the speech signal, and redundancy -- lots of it -- is vital. If it were even physically possible to speak a language with only one allophone per phoneme (it isn't), it wouldn't be a good idea.)

Now, obviously, the systematic shuffling around, addition, loss, of many predictable phonetic features will have no effect on distinguishing one utterance from another: more or less nasality in English vowels, more or less aspiration of voiceless stops, the spread of aspiration from an earlier inventory of environments to a larger set, and so on. Similarly, what Hoenigswald calls Primary Split only rearranges the deck chairs: makes more of one phoneme and less of another. As when all obstruents (stops and fricatives) in German become voiceless in final position. /ptks/ were already phonemes in the language, and there are still /bdgz/ phonemes, after the change, merely more of the former and fewer of the latter.

In theory, as you say, it would be possible for a phonological system (meaning, n number of phonemes) to become completely rephoneticized without actually changing the number or distribution of the contrast points (phonemes), as if, say, all /ptk/ became fricatives, all /bdg/ became /ptk/, all /mn/ became /bd/ (such things have happened, if rarely), /s/ became [þ], /f/ became [χw], and the vowels all played musical chairs. But no language has ever, ever managed to achieve anything like that, not for a whole phonological system. WAY too tidy. Human language is probably the most fastidious and highly-structured form of human behavior, but in the end it is still human behavior.

For that reason, in fact, sound laws that uniformly and without condition or variation of any kind affect all tokens of a given phoneme in a language are decidedly uncommon, by comparison to sound laws that affect a subset of phonemes. This is partly, I suspect, a reflection of the fact that "sounds" in a language are actually congeries of allophones, i.e., phonemes actually come from the factory with subsets ready-made.

Anyhow, by now, if you haven't given up reading this in disgust, you can see that many "sound changes" (which include such things as anaptyxis, such as the "intrusion" of stops in such histories as English stream < *srow-mo-, timber < *timriyo-, and so on) have no real impact on the inventory and distribution of phonemes in a language. Primary split, a very common type of sound change, reshuffles frequencies and distributions, occasionally reduces the number of contrasts (if all split products merge with existing phonemes), but it is true that even the distribution (where they occur in words relative to other phonemes) of phonemes is a phonological issue (e.g., the lack of voiced obstruents in final position in German words).

Different subject. This matter of citations in Wikipedia articles. It's not your lookout, but I don't have the impression that encyclopedic reference works in the Real World cite sources for assertions and statement of fact unless there's some, ah, specific question. In a contribution to Wikipedia from my hand (but by now tinkered with and disimproved to the point I would not admit to having anything to do with it) someone fretted over the fact that my examples weren't footnoted, meaning that if they didn't occur in some previously printed source they counted as "original research". Hah? Say what? For openers, lots and lots of "examples" that appear in published works are simply wrong. And for any historical (or other linguistic) fact, there are typically many possible valid examples, one doesn't have to keep recycling a sort of canon of Approved Examples. (Sheesh.)

But I'm sure I've trespassed on your good will. Please forgive me. Particularly since, for a variety of reasons, I'm not sure my original interest in contributing to Wikipedia, on matters I can claim to know a lot about, can be sustained. Too bad. Anyone who wants to know anything about these subjects should buy a book, not necessarily one of mine; there are several excellent treatments out there, both of diachronic linguistics in general and of specific languages.Alsihler (talk) 02:50, 12 February 2008 (UTC)

The great value of experts as Wikipedia editors stems partly from the fact that use of citations in Wikipedia is not like their use in physical encyclopedic reference works, and also partly from the fact that some published materials are simply wrong.

Citations here are, in significant part, a way to help other editors judge the validity of edits to an article. If an unsourced example in some highly specialized subject is changed by, say, an unregistered user who has never edited Wikipedia before, most editors have no way to know whether the edit is a valuable correction or subtle vandalism. They'll pretty much have to let the edit stand, because the assumption of good faith is one of the principles that allows Wikipedia to function; and that makes an unsourced article highly vulnerable to degradation over time. If the example has a specific citation, though, then it becomes somewhat possible for a non-specialized editor to reconsult the source to gauge whether the edit is on the level. Preservation of previously sourced material can therefore draw on a large pool of general editors — leaving the expert free to work on the exceptional tasks, of generating the material to begin with (which you've been doing here), and generating and improving citations for it. A non-expert would have to go hunting through unfamiliar material for these tasks; and, in particular, a non-expert would have great difficulty reckoning which published materials aren't right, where an expert could tell readily and share that knowledge with the rest of us. Pi zero (talk) 17:37, 20 May 2008 (UTC)


If both this article and sound change are sufficiently well developed, I can see maintaining separate articles. It would be nice if we listed this as the main article for a section or three of that, so there wasn't too much overlap. However, there is also the article Phonemic differentiation, which seems to be completely redundant.

Also, since this article is not specifically about Hoenigswald, I think we should stick to intuitive terminology such as "conditioned merger", and leave "primary/secondary split" to parenthetical notes. Alsihler stated that he didn't want to confuse the issue by creating new terminology, but "conditioned merger" is quite common in the lit, whereas AFAIK "primary/secondary split" is usually used in association with Hoenigswald. Many texts do avoid it, presumably because it's unintuitive. kwami (talk) 18:47, 15 August 2008 (UTC)

I agree with the merger, in principle. Both articles should be considered carefully as they both seem to contain valuable content, but they do seem, to me, to completely overlap in scope (which Sound change does not, in my opinion, as a phonological change is a notable subset of the former). Care should be employed also because there are additional tightly related articles such as phonological rule (which is more about the formal framework in which to express any sound change, but should definitely at least be linked to in this article). LjL (talk) 01:41, 17 June 2009 (UTC)
I absolutely, verily agree with the merger proposal, adding my encouraging calls. The current state is somewhat confusing, to say the least. ... said: Rursus (bork²) 07:58, 4 August 2009 (UTC)
Disagree. The definition on phonemic differentiation as it stands reads: "Phonemic differentiation is the phenomenon of a language maximizing the acoustic distance between [the range of phonetic values of] its phonemes" not differentiating/splitting phonemes. The terminology is confusing.
Chain shifts illustrate phonemic differentiation (and there is thus a section on them on the article), though they hardly illustrate phonological change--chain shifts are defined as non-phonological change, because there is no change in phonemes.

If the articles phonological change and Phonemic differentiation are kept separate, most of the redirects currently pointing to Phonemic differentiation should instead point either to phonological change or to one of the articles in Category:Splits and mergers in English phonology. jnestorius(talk) 16:21, 22 October 2010 (UTC)

Agree with the merger for the reasons stated by the nominator. Onthegogo (talk) 01:54, 23 March 2011 (UTC)