Talk:Comparative method (linguistics)/Archive 1

Name and place for article

There are all sorts of comparative methods in all sorts of fields, aren't there? So, shouldn't this article live at comparative method in linguistics or something more specific? I really have no idea what's appropriate, actually, but this does seem to be too broad of a name. --LMS

Fair enough. I'll keep the article here for the time being, but if comparativists from other fields object it will be redirected. -- Piotr Gasiorowski

as soon as somebody wants to write an article about another sort of comparative method, he or she will move this article to comparative method (linguistics). problem solved.

However, you should consider merging this article with historical linguistics, since the subject treated is exactly the same-- Dbachmann 11:48, 11 Aug 2004 (UTC)

Comparative method question

Going by the following correspondence:

• English = Father, Mother, Daughter, Brother, Sister.
• Latin = Pater, Mater, ____, Frater, ____.
• Greek = Patros, Matra, Thygatra, ____, ____.

How might one find the proper Latin sound correspondence to for Daughter/Thygatra and Sister, or the proper Greek sound correspondence for Brother/Frater and Sister? This turns out to be a much more involved process than I thought. For a long while, I assumed "teiktra" (Latinized to "teictra") as the Greek equivalent to "daughter" (mainly by analogy to correspondences between "dryas" and "tree", for example, having made an obvious blunder, and between "night" and "noct-", the latter of which I had for some reason misremembered as having originally been a Latin borrowing from Greek).

Anyway, I know these talk pages aren't for personal tutoring, but I was hoping to better understand the dynamic between these three languages and their phonological histories. --Thorstejnn 09:54, 6 May 2006 (UTC)

Look at Indo-European sound laws - that will give you the basic correspondances --Pfold 10:25, 6 May 2006 (UTC)

Thank you very much! I appreciate the help.

Peace, --Þorstejnn 05:06, 14 May 2006 (UTC)

Table of Polynesian cognates

There was a missing form --- the Hawaiian for "enter". I supplied that word, ulu, which means "enter" as in "inspire", i.e., "enter one's soul". See Pukui and Elbert's Hawaiian Dictionary (1986:368-369, 436). I also made the font appearance consistent (IPA) for all Polynesian forms in the table. Then removed the unnecessary italics which created a bad visual effect in the table. It will look better yet if it switches the x and y axes, providing vertical visual comparison of the cognate forms. Agent X 23:30, 24 June 2006 (UTC)

Oh yeah, the Hawaiian /ʔe-/ number prefix is not necessary, and was only a distraction in the table. See Elbert and Pukui's Hawaiian Grammar (1979:158) "kahi, lua, kolu". Agent X 23:39, 24 June 2006 (UTC)

Hello. Would it be helpful to add colloquial Samaon to the table in which t becomes k, and n become velar nasal? —Preceding unsigned comment added by 164.214.1.54 (talk) 21:37, 12 December 2008 (UTC)

Well, to establish previously unknown kinships, the method uses the oldest known form of a language; so including colloquial Samoan (presumably a descendant of the formal variety) would not help illustrate the method, although it does show that [dental] > [velar] really happens and so can be used in reconstruction of other language histories. —Tamfang (talk) 21:18, 14 December 2008 (UTC)

Featured Article Candidacy

Right, this is already an excelent article, so: well done to everybody who's contributed so far. However, my bid to get it the Featured Article status it deserves has come unstuck over inline citations. These are particularly needed for the examples of the method being applied to various languages. I'm prepared to try to provide these for the IE examples, but I don't have access to the resourced to to find references for Algonquin languages, Polynesian languages, Finnish, Pirahã, Dravidian languages or the Uto-Aztecan tree. Can anybody out there help? sjcollier 21:24, 27 June 2006 (UTC)

I (aka Whimemsz) don't have most of the books I used with me at the moment, unfortunately, since I'm away from home for a few weeks. I think I can remember which work each example came from, but I obviously don't know the page numbers, and won't for about two more weeks. Should I just wait until then to add the in-text citations, or should I put in the references to the books themselves, with no page numbers, and then go back and add those when I know them? Take care --Red Newt 09:25, 28 June 2006 (UTC)

Thanks! And again, fantastic work on this article. Page references are always nice, but are by no means vital; so if you add what you can now (providing you're confident you're citing the right source...) and put the page numbers in whenever it becomes possible, that'd be fantastic. I've already added what I can from books I have available, and Agent X has kindly offered to contribute references to the Polynesian stuff. I've also littered the article with hideous tags, which aren't ideal but at least make it clear what still remains to be done. Cheers, sjcollier 20:15, 28 June 2006 (UTC)

Okay, thanks! --Red Newt 23:20, 28 June 2006 (UTC)

That's great. I think we're getting there. sjcollier 23:51, 28 June 2006 (UTC)

One book I do have with me, actually, is R. M. W. Dixon's The Rise and Fall of Languages, which is basically about cases where the comparative method doesn't apply. So I'll add a section in the "Criticism" part about his Punctuated-Equilibrium theory--I think it's worthy of a section, since linguists have actually paid a lot of attention to it, and Dixon is extremely well-respected in the field. I know a bit about the history of the method (just the basics: Sir William Jones and the Neogrammarians, basically; a fuller mention of Greenberg and Swadesh and Mass Comparison and Glottochronology might be needed in such a section too), but not enough to feel comfortable actually putting what I think I remember onto Wikipedia. If I were at home, I could look in Fox, whom I seem to remember has some discussion of the history of the method. We might try checking the PIE article, the neogrammarian article (is there one?), and so on; there's probably a good deal of info and sources there. Take care, --Red Newt 00:25, 29 June 2006 (UTC)

That'd be good. I think Monocrat has a reasonable point about maybe a minor restructuring of the Criticism section - maybe a bit mroe of a general introduction? - though I personally don't think the subsubsections are too objectionable. I'm in a bit of a similar situation to you with books at the moment: I'm deprived of good library access for the summer vacation. Will see what I can do about a history section, though. May try to obtain a copy of Fox. Cheers, sjcollier 00:55, 29 June 2006 (UTC)

For anybody who can help, the citations still needed concern:

• History of the Persian language - borrowings from Arabic
• Evidence that the Finnish word äiti (=mother) is borrowed from Gothic (the cognate Gothic word would also be nice)
• Pirahã (x2) - seperate male/female dialects and borrowings from Nhengatu - (Encyclopaedia of Amazonian Languages, or something similar?)
• Evidence that Sanskrit grammarians were familiar with Grassman's Law
• Development of velar plosives in Kannada vs that in other Dravidian languages
• Etymology of the Spanish word palabra (< Latin parabola by metathesis)

sjcollier 00:55, 29 June 2006 (UTC)

Evidently, Panini mentioned Grassman's Law in his grammar of Sanskrit. That's the claim the page on Grassmann's law makes, anyway. I checked, and the user who added it is Opus33. On his talk page, he says he only edits on Sundays, though, so we might have to wait for then to get a reference. --Red Newt 03:56, 29 June 2006 (UTC)

Thanks. Will keep trying to track down something before then if I can. Sadly, I fear there is no hope of me obtaining a copy of Fox any time soon, so I'l try to find an alternative source on the history of the method. In response to suggestions from Agent X, I've added a bit on sub-groups to Genetically related languages and a paragraph on areal diffusion to the Criticisms; feel free to improve on them. sjcollier 11:37, 29 June 2006 (UTC)

Fully citated. I wish I'd though to check Campbell 2004 for the Finnish/Gothic citation earlier, though. Must be about FA standard now... sjcollier 17:54, 29 June 2006 (UTC)

w00tilicious. I added a section on Dixon's Punctuated-Equilibrium model (I don't know how well-written it is); all that's left, I guess, is a section on the history of the method. Great work! --Red Newt 21:09, 29 June 2006 (UTC)

Thanks! The Punctuated-Equilibrium section is great. I've tweaked the lead a bit, and moved the citations (and sentence about glottochronology/MLC) to the Criticism intro - having only just found out that the MoS reccomends against citations in the lead. As you say, a history section would be good at some point; can't see that it's absence should be a barrier to FA status, though. Fingers crossed... sjcollier 10:17, 30 June 2006 (UTC)

You just beat me to it on the History section - I was just about to start writing something based on Szemerényi. Thanks for saving me the work! Looks like you've done a great job; I'll see if I can add anything. sjcollier 19:37, 30 June 2006 (UTC)

Okay, I've added everything I can; I think that's fairly comprehensive (on the C19th stuff at least). Not sure that we need to add anything on the development of glottochronology or MLC; that really belongs in their respective articles (or a new lexicostatistics article). Excellent work - it must be nearly there now. sjcollier 21:15, 30 June 2006 (UTC)

Objection to FAC

I'm objecting to this article being promoted to featured status, but as my reasons will be somewhat long and somewhat technical, I'm bringing up them here rather than at Wikipedia:Featured article candidates/Comparative method.

1. I really wish the notes were given in the style recommended at WP:FN (i.e. using the <ref> tag) and that the books were cited using the {{Cite book}} template. I know that the current style is standard in linguistics journals, but well, this isn't a linguistics journal, it's an encyclopedia.
2. There's something of an internal inconsistency in the article: is the comparative method a tool to show that two or more languages are related, or is a tool to reconstruct the proto-language of languages known to be related? The opening paragraph suggests the former; the "Application" section suggests the latter.
3. The lead ends with the words, "reconstructions obtained by the comparative method are now generally treated with a degree of skepticism." What, all of them? Hardly. Advances in our understanding of how historical linguistics works have shown ways in which the comparative method needs to be tweaked and added to, but they haven't brought the entire method under suspicion. This comes upon again in the first paragraph of the "Criticism" section. The sentences: "...linguists continue to use the comparative method; other proposed approaches to determining linguistic relationships and reconstructing proto-languages ... are considered flawed and unreliable by most linguists. In contrast to previous generations of historical linguists, present-day linguists recognize that results obtained with the comparative method are to be viewed skeptically" practically contradict each other: in one breath you say that modern linguists still use the comparative method and its competitors are flawed and unreliable, and in the next breath you say its results are viewed skeptically. The reader is left scratching his head wondering if the comparative method is a tool used by reputable linguists or only by crackpots.
4. In the Application section it would be nice if we could stick to the same set of examples throughout. We begin with a nice chart of Polynesian languages, but the reader is disappointed because the Proto-Polynesian forms of the words in this chart are never reconstructed.
5. In the section "Establish correspondence sets" I don't understand why some correspondences need to be nontrivial, and your example of ŋ : b makes me wonder whether that is an attested correspondence set in some language family, and if so, where.
6. In the section "Discover which sets are in complementary distribution", what does the Dravidian example have to do with a sound change occurring in an environment that was later lost?
7. In the same section, the Romance example needs data (cuore/corazon/coração/cœur vs: caro/caro/caro/cher); also, the a didn't always become [ɛ] in French (cf. château, chambre)
8. The "Criticism" section seems to me to be mostly criticism of things associated with the comparative method, but not of the comparative method itself. The Neogrammarian hypothesis that sound laws operate without exception is probably untenable in its strongest form, but that isn't actually the comparative method's problem. Analogy isn't really a problem for the comparative method either; even the Neogrammarians recognized that analogy can disrupt the regularity of sound change. And the example of the Slavic word for nine beginning with /d/ under the influence of the word for ten isn't really the best example of analogy anyway. Perhaps a better example would be Pre-Greek *sek^wetai > Greek hepetai "follows" (although *k^w normally becomes t before e in Greek) on the analogy of other forms in the same paradigm where *k^w became p regularly.
9. The section "Problems with the Tree Model" doesn't establish that the comparative method depends on the Tree Model and that the comparative method can't be used if the Wave Model is right. It also incorrectly suggests that the Tree Model and the Wave Model are mutually exclusive, which of course they aren't. It also lacks references on who first developed each of these models.
10. Sources are needed to support the claim "This Punctuated-Equilibrium Model has received a great deal of attention from linguists, and many are inclined to accept the model as accurate."
11. In the section "Subjectivity of the reconstruction", readers are going to wonder what you mean by "Such dramatic asymmetries in the growth of different branches of the same tree are in fact common; contrast for example the Romance and Celtic branches of Indo-European." I know I do.
12. More generally, I feel the article places too much emphasis on the mechanisms and limitations of the comparative method and not enough emphasis on its significance. I would like a nonspecialist reader to come away from an encyclopedia article on the comparative method knowing that comparative historical linguistics simply cannot be performed without the comparative method. The details of how it works and the limitations are also important, but I think that is the most important thing of all: the comparative method is indispensable in comparative linguistics.

In addition to these specific objections, I have a much vaguer one that probably cannot actually be fixed: the tone of the article is much more that of a textbook chapter than that of an encyclopedia article. I'm not sure how to fix that, or even how to be more specific about what gives me that impression, and I wouldn't object to FA status for this reason alone. But the fact remains that the reader is left with the impression of having been taught something rather than having learned something on his own, which is as close as I can come to defining the difference between a textbook chapter and an encyclopedia article. User:Angr 19:03, 5 July 2006 (UTC)

With regard to Angr's point 3, I think that there isn't really any contradiction. On the one hand, as a method for establishing genetic affiliation, the comparative method is well established and is considered reliable, in contrast to such alternatives as superficial lexical comparison. On the other hand, the reconstructions are regarded with some skepticism in the sense that we can never be sure how accurate they are. For example, there is no way of reconstructing a contrast in the proto-language if it was completely neutralized in all of the attested daughter languages and no other changes interacted with it and provide indirect evidence. The fact that we can't be sure of the phonetic form of reconstructions, or even of some aspects of their phonology, does not detract from the utility of the method as a means of establishing relationships. It may be, however, that this distinction is not sufficiently clear and that the article should be revised to be clearer.Bill 07:06, 9 December 2006 (UTC)

I agree with pretty much everything Angr pointed out above. This is a pretty good article, but it still needs detail work to make it great. I'd like to add some more comments and suggestions.

• The section "Crticism" needs to be heavily edited. There's a lot of stuff here that doesn't belong under that heading, but could be included in the article in another way. E.g., "Analogy" clearly belongs somewhere else, and I think also "Creoles" should be removed; these have nothing to do with "critcism". Moreover, I fail to see why creole formation should pose a problem to the comparative method. Even if the method were not applicable to creoles (which, I think, can be disputed), this would not invalidate it. No scientific method is omnipotent.

Agreed. I'm not sure the section on Criticisms has the right heading. Most of the issues raised here are about unwarranted assumptions made by (particularly the early) practicioners of the method, not flaws in the method itself. For example, correct use of the method has revealed cases where no single root can be constructed, so how can it be a flaw of the method that it reveals variation in the proto-language? I think it would be preferable to talk of Limitations. As criticisms, these really belong in comparative linguistics. --Pfold 22:56, 10 July 2006 (UTC)

• "reconstructions obtained by the comparative method are now generally treated with a degree of skepticism." -- As noted by Angr, this is an overgeneralization. It is well-known that in certain types of cases reconstruction through the comparative method may produce a proto-form known to be wrong. But for balance, I think it should be mentioned somewhere in the article that there are also numerous cases where reconstructions are confirmed as correct by independent evidence. A good example is the striking confirmation of Proto-Germanic and Proto-Baltic reconstructions in Proto-Finnic in the form of loanwords: cf. such well-known examples as Germanic *kuningaz- 'king', *skauniz 'beautiful' etc. ~ Proto-Finnic *kuninkas 'king', *kaunis 'beautiful'. "Subjectivity of reconstructions" might be a good place to place this kind of information as counterevidence to subjectivity.
• "This Punctuated-Equilibrium Model has received a great deal of attention from linguists, and many are inclined to accept the model as accurate." -- I find this passage problematic as well. Of course, what counts as "many" is unclear, but my impression is just the opposite. True, there has been a lot of attention, but partially in the form of heavy critiques (e.g. by Campbell), and I haven't seen much signs of the punctuated equilibrium hypothesis being widely accepted as yet. There's much discussion on the idea in Areal Diffusion and Genetic Inheritance: Problems in Comparative Linguistics ed. by A. Aikhenvald & R.M.W. Dixon, and most of the contributors in this volume seem to have a somewhat cautious or critical attitude towards it.--AAikio 07:53, 6 July 2006 (UTC)

Proposed revision

Hello all. I've been drafting a significant revision to the article to try to address a few of the problems highlighted by Angr and others. What I've got so far is on my subpage, User:Miskwito/Notes (ignore the two links and the horizontal rules at the very top). Some of the changes I've made there are:

1. Tried to make it clearer that the comparative method is reliable and not distrusted. I renamed the "criticisms" section to "problems" "potential problems", which seems to have some precedent in other Wikipedia articles, and which in my opinion anyway is somewhat less harsh (perhaps Pfold's suggestion of "limitations" would be better?). I also did some rewordings of certain paragraphs and sentences to remove or clarify some of the more POV or inaccurate language, and also provided several quotes from Lyle Campbell and Hans Henrich Hock to further highlight the reliability and strengths of the method.
2. I removed the section on "creoles", which isn't really a challenge to the comparative method at all. I also removed the section on the punctuated-equilibrium model, which as has been pointed out by others above, isn't well-accepted.
3. I changed the citation format from inline to using <ref> tags, which I think cleans up the article a bit (and following Angr's suggestion).
4. I removed the Dravidian example Angr mentioned, which was irrelevant to the discussion at hand.
5. Altered the mention of "non-trivial" changes somewhat, and removed the hypothetical one of b:ŋ.
6. Made it clearer that the reconstructed phoneme inventory in the section on "examin[ing] the reconstructed system typologically" is hypothetical, and not necessarily an actual reconstruction of any real proto-language.
7. Removed the "Such dramatic asymmetries in the growth[...etc.]" sentence Angr mentioned.

In my opinion, anyway, it's a good start toward addressing many of the weaknesses of the article. It's still got a long way to go, though. And I'm not sure there's any real way to fix the sense of being "taught" Angr brought up--short of pretty much rewriting the entire article with new wording. I also haven't addressed his points 4 or 7 yet, nor the point about the example used to illustrate "analogy". Maybe tomorrow. Or something. Anyway, since these are some significant changes, I wanted to check with all interested parties that they're okay, and that everyone agrees for the most part with the changes. If you need to view all of the changes at once, here you go: [1]. Comments? Ideas? Criticism? --Miskwito 02:33, 28 January 2007 (UTC)

That all makes good sense. Good work, that man. Dewrad 22:26, 28 January 2007 (UTC)

Excellent revision, Miskwito, just go ahead and replace the earlier version with it. I have a few edits in mind I could make after that, e.g. I'd like to balance the "Subjectivity of the reconstruction" section by adding something on cases where reconstructions are confirmed by loanwords in another language family (such as Germanic > Finnic) - what do you think? And, I also think that "Limitiations" would be a better heading than "(Potential) problems". --AAikio 07:24, 29 January 2007 (UTC)

Sounds good to me--the Germanic>Finnish loanwords are a good example of that. "Limitations" sounds good; I added potential because in that section, as the revised version now stands, the basic premise is that most aren't actually significant problems. But "limitations" for me does carry pretty much the same connotation. I'll go ahead and make the switch. --Miskwito 18:49, 29 January 2007 (UTC)

'punctuated equilibrium' redirects to this page

When clicking on the link for 'punctuated equilibrium' on the page on R. M. W. Dixon I got redirected here, yet there is no mention of that hypothesis here (well, ok, on this talk page there is). Is this some remnant of the section 'Criticism' that apparently once has existed, and where, apparently, Dixon's hypothesis was mentioned? -Stephan Schulz 80.186.191.43 21:17, 21 March 2007 (UTC)

The section can be found in the page's history. Clicking this link should take you to the old Punctuated Equilibrium section. --Miskwito 23:08, 22 March 2007 (UTC)

But should it really be linked here, when there's no mention of it on the main page? I was just wondering, I don't know the policy. -Stephan Schulz 137.163.18.151 09:14, 23 March 2007 (UTC)

I removed the link --Miskwito 23:36, 25 March 2007 (UTC)

Further improvements

Dammit. I had a whole nice long thingy written here and one mistaken click later it was lost forever. Argh. So, this isn't as brilliant as the original version, but use your imaginations to picture how friggin' inspired my prose was:

I've been thinking lately about how we can improve the article further. One minor change that can be made is to get a better example of analogical change. The current example is useful because it demonstrates how individual words can undergo irregular changes due to the shape of some related word, but that's not exactly the same as analogical change brought about by paradigm shapes and structures and what have you. But I don't think we should get rid of what's there now. I can also go through the article and just make sure all the references are formatted correctly.

For the more significant change, I've been trying to decide how we might address Angr's concern about the prose style of the article (it reading more like a textbook than an encyclopedia article). In part, I think this is due to the use of "we" a lot, which shouldn't be terribly hard to fix. In part, though, I think it's due to the types of examples used and how those examples are presented: it's kind of like the article guides the reader through a process of learning how to use the method, which I guess isn't exactly the same as simply presenting the methods linguists use and saying "these are the methods." But I'm having trouble figuring out how we'd be able to change that while still retaining the current examples (which I think are useful and nice, for the most part; there's also the point about switching between different languages to illustrate each point, rather than, say, sticking with Polynesian languages, but I'm too lazy to worry about that now).

So, I think the questions are: (1) is there really a problem with the prose that needs to be fixed (or, if not a "problem", is there a way the prose can be improved)? (2) if yes, how do we improve the prose? As I said above, for starters, just getting rid of the "we"s might do a lot. If it still seems "textbooky", perhaps we'd want to get the League of Copyeditors involved? I'm not sure...

What do others think? Agree? Disagree? Other thoughts or ideas? --Miskwito 09:29, 25 March 2007 (UTC)

A...anyone? --Miskwito 05:47, 8 April 2007 (UTC)

Tree/Wave Model Sections

These sections should not be in this section as they can be used in connection with other than the comparative method. A better home would be Historical Linguistics. Adresia 20:17, 25 March 2007 (UTC)

My reversion of Adresia's edits

I'm going to give the reasons I reverted your recent edit here, Adresia, so we can discuss it on the talk page...and, you know, stuff:

1. "in social science" is out of place, I don't know why you added it there or what it's supposed to tell people
2. There's no need to add terms like "qualitative" in the lead, which is supposed to provide a very basic, easily-understandable summary for lay readers. "Qualitative" doesn't really add any valuable information there, it just makes the writing slightly denser, I think.
3. Same goes for the mention that "the latter may consist of grammatical, morphological or lexical characters", and adding "lexical" after "cognate lists" and all that; those terms aren't really necessary in the lead (although I do think greater mention of the comparative method's use in areas besides phonology would be helpful for the article, but I'm not sure the lead is the place for it. I'm kind of ambivalent on this, though).
4. "The basis of the method is the Neogrammarian Hypothesis that "sound laws have no exceptions" and the tree model of language development." Again, unnecessary addition of information; if I was someone unfamiliar with historical linguistics already, seeing this sentence randomly at the end of the first paragraph of the lead would confuse the heck out of me.
5. Changing "mainstream" to "traditional" is inaccurate--the linguists who reject the comparative method and/or use methods like mass lexical comparison are definitely considered fringe and outside the mainstream. "Traditional" kind of makes it sound like the people with these new methods are more progressive, while the "traditional" linguists are conservative and refusing to yield to new, better ideas. Probably not what you intended, and I don't know if readers would actually consciously think that, but... Hm, in any case, "mainstream" is a bit of a loaded term as well; maybe we can try to change it, but I don't think "traditional" is a good solution.
6. "Modern historical linguistic methods are based both on the comparative method and the lexicostatistic method, but use statistical hypothesis testing." This is a very clumsy sentence now, and if I'm reading it correctly, fairly inaccurate. Most historical linguists don't use lexicostatistics for anything other than presenting its results as a reeally general guess. They don't generally treat it as something reliable or on par with the comparative method.
7. I don't know that either of "Application" or "Details of the Method" are better than the other, although I think "application" describes the content of that section more accurately. I'm ambivalent about this too, though.
8. This isn't a real problem at all, but "sound" in "Establish sound correspondence sets" isn't really necessary either, I don't think.
9. Adding "problems with" to the headings is both unnecessary (because of the larger heading "limitations") and misleading (see the discussions above).
10. "but still relies on judgement by linguists" totally contradicts the previous phrase, claiming that "the identification of systematic sound correspondences between known languages is fairly objective". The correspondence sets themselves generally are objective, once you find them--if there are enough examples of the correspondence; it's just the reconstruction of those sets that's really subject to the biases and opinions of the researchers.
11. The "Method is non quantitative" section doesn't explain what "quantitative" or "qualitative" really mean in this context, it brings up examples without explaining them (the lay reader will be asking "what the heck is 'Italo-Celtic'?"), and it inaccurately suggests that lexicostatistics "compliments" the comparative method--to my knowledge, that's not the view adopted by most historical linguists, who, if they use it at all, use lexicostatistics just to give them a broad guess about when the protolanguage was spoken, and get them started on possible segmentation of the family tree (actually proving subgrouping requires a number of shared innovations, phonological and/or morphosyntactical). Finally, the point that the reconstructions are subjective is already addressed elsewhere.

So, those are my reasons. I'm interested in your responses, though, and in what others think. Take care, --Miskwito 23:35, 25 March 2007 (UTC)

Sorry Miskwito to have not responded to you earlier but I havent looked at this page for a while. I will briefly respond to each of your points :- 1. This merely added information as to the field in which there was another comparative method. This is taken up in the rewrite. 2. "Qualitative" does add information in that it shows that the method can not determine the degree of relatedness. 3. I'm content if the method's use in non-phonological areas is mentioned later. 4. In a summary of the method at least the tree model ought to be mentioned. 5. I agree that mass comparison is non-mainstream (and dead). By non-traditional I was thinking of cladistic methods introduced in the last decade by Ringe, MacMahon, Gray and Atkinson, etc. 6. I was not referring to 1950s lexicostatistics as a modern method but the ones introduced by the above people. They produce more than a general guess IMHO, producing similar tree to the traditinal method. 7. "Application" to me is when the method is used on (say) Australian languages. Defining the steps involved is an algorithm. 8. There could be other correspondances other than sounds. 9. It may not be necessary but is surely not misleading. 10. Cognacy judgements are only opinion and do vary fron expert to expert. See for example the revisions of Dyen's judgements. I think that this phrase should be reinstated. One of the questions is how many correspondances is enough. 11. Qualitative and quantitative are defined in any dictionary, and dont really require further explanation. Italo-Celtic can be looked up in Wiki if interested. Lexicostatistics uses cognacy judgements (when used properly) and does provide additional information and so complements the method. Adresia (talk) 14:15, 1 April 2008 (UTC)

palabra and девять

The word palabra is an instance of metathesis, but it's not the only similar one: peligro is metathesized from periculum "peril". A similar-sounding word culebra "snake" is not metathesized; the Latin form is coluber. I propose the rule: r + stressed vowel + voiced stop + l -> l + stressed vowel + voiced stop + r. Can anyone provide other examples or counterexamples?

Thought of a couple. milagro is metathesized from miraculum; roble (from robur) was not metathesized, so there has to be a vowel before the r for it to metathesize. -phma 22:04, 22 April 2007 (UTC)

The change n->d couldn't have happened between Proto-Slavic and Russian, as it also occurs in Polish dziewięć, Bosnian devet, and even Latvian deviņi. It appears to be a Balto-Slavic innovation. phma 15:02, 22 April 2007 (UTC)

Comparative method as heuristic

Is the whole emphasis of the initial definition of the comparative method given here not wrong?

Surely the comparative method is used to show how languages which are related have developed from a common ancestor, and alone is not proof that they are? Regular sound correspondences in lexical material, no matter how extensive, can be shown by borrowings; in extreme cases the entire lexicon of a language can be replaced. Genealogical linguistic relatedness is proved, not by lexical comparisons, but by the inheritance of shared paradigms and irregularities which are so unusual that they could only be accounted for by relatedness. Comparison is what happens after relatedness has been proved (or is so obvious to be beyond doubt). I think this distinction is an important one, given the prevailing tendency for tenuous linguistic relationships to be alleged on the basis of a bunch of lexical resemblances with apparently regular sound correspondences gathered by supposedly following the comparative method.

See Johanna Nichols (1996): "The Comparative Method as Heuristic" in Mark Durie and Malcolm Ross (eds.) "The Comparative Method Reviewed: Regularity and Irregularity in Language Change", Oxford University Press

Bofoc Tagar 12:56, 8 May 2007 (UTC)