Talk:Old English phonology

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


From the text:

  Consonant allophones
  The sounds marked in parentheses are allophones:

I don't see any parentheses in the subsequent list, though. Only brackets. Is this what was meant? --Godtvisken 18:10, 17 March 2006 (UTC)

Yes. The symbols ( ) are called parentheses in American English, which this article is written in. Angr/talk 19:23, 17 March 2006 (UTC)

Voiced þ and ð[edit]

In The Cambridge Old English Reader by Richard Marsden (Cambridge 2004), it says that these letters are voiced ‘at the start of a word or medially’ (he gives the examples þis and hwæþer), whereas this article seemes to say that they were voiced only medially. Is Marsden wrong? Widsith 10:51, 11 September 2006 (UTC)

English is pretty conservative with consonants (except for c, g, and h). This means that you should be able to use your knowledge of Modern English to infer the voicing in Old English. So the OE ancestors of the, this, that, there, and then would have been voiced, and the ancestors of thane, thatch, thin, think, three, and through would have been unvoiced. Does Marsden say þ and ð are always voiced medially? My understanding is that they were not voiced when doubled or adjacent to an unvoiced consonant. (I can’t think of any NE examples, but siþþan and æfþanc are OE examples.) --teb728 09:01, 14 September 2006 (UTC)
the story i've normally heard is that the initial voicing of þ did not occur until middle english, at the time when þ and ð split phonemically. note that southern middle english dialects showed a similar voicing of initial /f/, which comes down to modern english in certain words like "vixen" (cf. "fox") and "vat". it's likely that the /f/, /v/ distinction became phonemic as a result of borrowing from french, and that this was reinforced by certain late middle english changes (loss of final vowels and double consonants) that also introduced phonemic /z/ and /ð/. Benwing 02:41, 9 January 2007 (UTC)

allophones of /h/[edit]

Another question. The article says that [ç] and [x] are allophones of /h/ ‘after front and back vowels respectively’. But surely /h/ also changed after liquids? E.g. a word like þurh ‘through’ must surely have sounded something like /θurx/, no? Widsith 16:46, 3 May 2007 (UTC)

be more specific![edit]

There was no thing such as "Old English" per se. We have records of various dialects, as well as cases such as the Beowulf which may have originally been set in one dialect (Anglian), but later preserved in another (West-Saxon). The phonology of modern English is largely derived from non-West-Saxon dialects, so at the very least it would be more responsible to indicate in the article that it is discussing just one dialect which acted as the standard written form (I doubt there was a standard spoken form) for several centuries. Of course a few lines about some of the major differences in phonology (West-Saxon vs. Kentish vs. Anglian) would not be amiss, even in a short article.Jakob37 05:51, 6 September 2007 (UTC)

Anglo-Frisian brightening[edit]

According to "a History of English Phonology" (Jones 1989) p. 85-86, long vowels did not participate in Anglo-Frisian brightening.Jakob37 02:04, 7 September 2007 (UTC)

I think there are two schools of thought on that. The vowel in question comes from Proto-Indo-European *ē and shows up in Gothic as ē but in Old Norse and in all West Germanic languages except the Anglo-Frisian ones as ā; in Old English it's ǣ and in Old Frisian it is (I believe, not 100% sure) ē. So it's usually reconstructed as *ǣ or [ɛː] or the like for Proto-Germanic so that it can be differentiated into Gothic ē on the one hand and Norse/WestGmc ā on the other. The question is whether it simply remained ǣ in Anglo-Frisian, or whether it got backed to ā in the ancestor to A-F as it did in all the neighboring languages and then got fronted again to ǣ by Anglo-Frisian brightening. It's probably not possible to prove it one way or the other, so people just have to rely on what they consider more plausible. Usually when this sort of question arises at Wikipedia, we report the majority view, and mention the minority view only if it's held by a significant minority and reporting it doesn't violate WP:UNDUE. In this case, though, I suspect the majority of Germanicists haven't bothered publishing their opinion because the question is (1) unanswerable and (2) trivial. —Angr 08:08, 10 September 2007 (UTC)
Trivial or not, it seems you are confusing apples and oranges here: the long vowel in question is Germanic ǣ, but the short vowel in question is Germanic a, which has no long equivalent; it could just as well be matched with long "o" as with ǣ. I think that Jones assumes that (traditional view) Gc ǣ had already gone to a central vowel by the time that Anglo-Frisian broke off, and that the later fronting in both languages was earlier and not part of AF Brightening. In any case, from what I have checked, the consensus is that AFB only affected short vowels. Where you say that the Gc ǣ: may have "simply remained ǣ in Anglo-Frisian" is not the received explanation, but to my mind it is simpler (as I recently expressed on some related page in Wikipedia), in which case, again, AFB would only affect short vowels.Jakob37 11:32, 11 September 2007 (UTC)
Gmc short a had no long equivalent *until* ǣ became ā. Once that happened, they're a pair and could have fronted to æ/ǣ together. I think that's the simplest explanation, but I don't know if there are facts inconsistent with it. —Angr 18:34, 11 September 2007 (UTC)
Jacob37, Perhaps you could elaborate on your statement, “from what I have checked, the consensus is that AFB only affected short vowels.” That is the only part of your comment that is potentially relevant. As for your opinion of which is the simpler explanation, Wikipedia doesn’t include the unpublished opinions of editors. Not that my opinion counts any more than yours, but:
  • ǣ is backed to ā in West Germanic, and then
  • a and ā are brightened to æ and ǣ (except sometimes) in Anglo-Frisian.
seems a simpler explanation than
  • a is brightened to æ (except sometimes) in Anglo-Frisian,
  • ǣ is backed to ā before w in Anglo-Frisian, and
  • ǣ is backed to ā in other West Germanic dialects.
--teb728 21:47, 11 September 2007 (UTC)


It'd be nice if there were a section on OE phonotactics. Anyone know of any sources where someone could get this info? True (talk) 19:25, 2 June 2011 (UTC)


I made changes to the diphthongs section to provide a more balanced array of information on the iy/ie question. I have also found at least 2 examples in Bosworth and Toller of ie being spelt as ye, which could further sugggest that ie being spelt as y in latter OE texts did not mean anything of the sort that the diphthong was pronounced iy (because it is the primary element - the element not under question - that was represented as a y in these two examples, not the latter, which is the one being alleged to have been a y sound). Gott wisst (talk) 11:44, 8 May 2013 (UTC)


The vowels need to be hyperlinked. — Preceding unsigned comment added by (talk) 05:24, 26 January 2014 (UTC)

No, they don't. Nobody knows their exact pronunciation. — Peter238 (v̥ɪˑzɪʔ mɑˑɪ̯ tˢʰoˑk̚ pʰɛˑɪ̯d̥ʒ̊) 13:18, 24 March 2015 (UTC)

[dʒ] as allophone of /j/[edit]

Is the claim that "[dʒ] is an allophone of /j/ occurring after /n/ and when geminated" not a mixture of diachronic and synchronic approaches? Synchronically, I think one has to argue that OE /dʒ/ could be analysed as having phonemic status, as we can form minimal pairs, such as sencan /ˈsenkɑn/ vs. sengan /ˈsendʒɑn/. Moreover, I don't think the allophony (if there was any) would be centred on /j/, but rather on some kind of palatalized /ɡ/ as [ɡʲ] => [dʒ], which is far more phonetically plausible than [jj](?)=>[dʒ] imho. In any case, that would not be allophony in OE, but in Proto-WGmc at best. By the way, I like the entry as it is and many thanks for the many contributions; as with all wikipedia entries the information given has to be taken cum grano salis, which is fine. Symkyn (talk) 10:11, 11 September 2014 (UTC)

/ˈsenkɑn/ vs. /ˈsendʒɑn/ doesn't show that /dʒ/ is a distinct phoneme from /j/, though. The claim is that [ˈsendʒɑn] is phonemically /ˈsenjɑn/. At the phonemic level it hardly matters what the manner of articulation is; you can call the underlying phoneme /ɟ/ or /ʝ/ if you prefer as long as it surfaces as [dʒ] after /n/ and when geminated and as [j] elsewhere. —Aɴɢʀ (talk) 18:07, 12 September 2014 (UTC)
You are right; my minimal-pair-argument is probably dead-wrong here, thanks for pointing it out. I'm still not convinced that "[dʒ] is an allophone of /j/ occurring after /n/ and when geminated" or at least I don't quite get my head around this claim. So, I'll spend some more thought on it, first. Symkyn (talk) 05:14, 18 September 2014 (UTC)
Not sure why this is unlikely. [dʒ j] is rather similar to the stop-continuant pair [g ɣ], except in being postalveolar and palatal rather than velar. It's common for continuants and stops or affricates to be the same phoneme; this allophony occurs in Spanish, where /b d g/ are approximants in most cases, but stops after nasals (among other places). — Eru·tuon 16:36, 24 March 2015 (UTC)
I also don't get why would that be unlikely. [dʒ] sounds, arguably, more "defined" than [ɟ] or [ɟʝ], and is easier to pronounce. Indeed, in some dialects of Spanish, the non-lenited allophone of /ʝ/ is coronal [dʒ], rather than the usual dorsal [ɟʝ] (which to me sounds like [dʑ] in most speakers, but that's another story.) — Peter238 (v̥ɪˑzɪʔ mɑˑɪ̯ tˢʰoˑk̚ pʰɛˑɪ̯d̥ʒ̊) 18:51, 24 March 2015 (UTC)
I do not know whether there are perfect minimal pairs, but instances of /j/ after /n/ that are distinct from /dʒ/ do exist, e.g. in wenian "train, prepare, accustom" [1], gemanian "to exhort, to prompt, to remind" [2], wunian "to dwell, to remain" [3]. Contrasting with sengan "to singe", men(c)gan "to mix, to mingle" [4], glen(c)gan "to adorn" [5]. The trick is that palatalized /g/ merged with inherited /j/ in most positions, but they remained distinct after nasals, where /g/ must have been a stop rather than a spirant in Proto-Germanic. Because of this contrast /j/ and /dʒ/ must be kept as distinct phonemes in OE, the latter arguably with quite a limited distribution. Aucassin (talk) 10:44, 27 March 2015 (UTC)
I thought that i in the -ian ending was pronounced in its own syllable as /i/. You're saying it was pronounced as a semivowel /j/? That's possible, and maybe even likely, but is there a source that says this? — Eru·tuon 17:54, 27 March 2015 (UTC)
Actually, Hogg in the Phonology and Morphology chapter of The Cambridge History of the English Language (p. 113, 114) says nerian had /j/ in Proto-Germanic, though he mentions this only in passing, as a case where the /j/ that triggered umlaut was not deleted. Assuming that wenian, gemanian, and wunian also had [j], then the phone [j] does indeed occur after a nasal, and [dʒ] has to be a separate phoneme from /j/. Regrettably Hogg doesn't make the connection when he discusses the phonemic inventory of Old English, and unless we have another source that does make the connection, it would be OR if we added /dʒ/ as a separate phoneme from /j/.
I also encountered something saying that the phone [ɣ] occurred after a nasal through syncope, and if so, then this indicates that /ɣ/ had become a separate phoneme from /ɡ/. Regrettably, I don't know if any source has made this connection either. — Eru·tuon 02:43, 29 March 2015 (UTC)
According to Wiktionary, the 1st singular present of nerian is neriġe, which must indicate a syllabic -ij- and not just -j-. The infinitive is also attested once as nerġean according to Bosworth/Toller. If I'm not mistaken, there are also a few attestations of class 2 weak verbs with this -ġan, though I don't know which. I know of at least frēoġan < *frijōjaną < *frijōną but this stem is of a rather unusual shape. CodeCat (talk) 02:59, 29 March 2015 (UTC)


I think it would be beneficial to split the description of sound changes to another article, and to outline the more important changes in this article. The article is ridiculously long now, and too complex for a non-specialist. For description of phonemes, only brief information is needed: a list of the sound changes that gave rise to diphthongs, palatalized consonants, umlauted vowels, etc. — Eru·tuon 22:05, 18 February 2015 (UTC)

I've created Phonological history of Old English, and moved content from here to there. I'm planning on deleting the sections on sound changes from this article, and adding just enough on sound changes to explain the inventory of phones: for instance, a brief statement on where the diphthongs, postalveolars, and palatal originated from. — Eru·tuon 20:37, 27 March 2015 (UTC)

Celtic influence, and velarization and palatalization[edit]

Peter Schrijver has an interesting theory in his new book on Language Contact and the Origins of the Germanic Languages: that Old English breaking, backing, and i-mutation were phonological loans from the Celtic language of Britain. Schrijver points out that the vowel changes before velarized and palatalized consonants in Irish are similar to the vowel changes of breaking, backing, and i-mutation in Old English. He argues that Old English also had velarization and palatalization through language contact with British Celtic, and these phonological or phonetic features caused the observed Old English vowel changes.

It's generally agreed that Anglo-Saxons didn't replace the original population of Britain; rather, they became dominant over the native population and their language came to be spoken by British Celts. Schrijver argues that Britain had a dialect called Lowland British Celtic, distinct from Highland British Celtic, the language that became Welsh, Cornish, and Breton. Some Lowland British Celtic speakers moved to Ireland in the first century when the Romans invaded Britain, and later became Old Irish through a dramatic series of phonological changes when the native Irish adopted the language. But some Lowland British speakers remained in Britain, and since their language, Schrijver argues, already had velarization and palatalization, it loaned these features to Old English.

His theory needs a lot more development, but it looks promising. If phonemically velarized consonants existed, then the Old English backing diphthongs ie, eo, ea are really plain vowels followed by a velarized consonant, the same as i, e, æ; and when i, e, æ are actually spelled, this indicates they were followed by a plain or palatalized consonant. Then words that are traditionally taken as showing a contrast between monophthong and diphthong really contrast in whether the consonant following the vowel is velarized or not. And when velarization was lost, the supposed diphthongs merged with the single vowels, showing they were really the same thing all along. Differences between presence and absence of breaking could be explained by presence and lack of velarization in different words and different dialects of Old English. And the way in which a back vowel caused vowel changes was through the intermediary of the velarization of the preceding consonant (similarly with a front vowel and palatalization). But this is just me expanding on what Schrijver says; he doesn't develop his theory this much.

Not sure if or in what way Schrijver's theory should be mentioned in the article. Perhaps I'll add a note on the possibility that the diphthongs were really simple vowels followed by a velarized consonant, in the case of breaking at least (perhaps not in the case of diphthongs inherited from Proto-Germanic *iu, *eu, *au). This is a possibility already considered in other sources, I think. — Eru·tuon 20:29, 30 March 2015 (UTC)

I definitely think the interpretation on velarisation is interesting and is certainly worth mentioning. But I'm more skeptical about using the same reasoning for i-mutation. After all, i-mutation was far from unique to Britain; it spread across all Northwest Germanic languages. It appears in Old High German and Old Norse already in the oldest texts. It would be highly unlikely for the change to spread from Britain to all of Germanic Europe in such a short time span. Old High German could, of course, have a similar Celtic substrate, but that's much less likely as large parts of High German territory had been Germanic for a much longer time, and in any case it would mean that velarisation existed not just in British but also in Gaulish, for which there is no evidence. And of course Old Norse never had any Celtic substrate, making i-mutation there even harder to explain. The only explanation that makes sense to me is that i-mutation already existed allophonically throughout Northwest Germanic. CodeCat (talk) 22:02, 30 March 2015 (UTC)
I'm not sure if Schrijver discusses i-mutation in Old Norse and Old High German as well as Old English. He has a chapter on OHG at least, but it's not viewable on Google Books. I'm also skeptical if he's claiming that i-mutation originated in England from British Celtic influence; it seems to be an inherited feature from well before Germanic languages came to England, at least in OE and ON. (I'm not familiar with Old High German.) However, the explanation of i-mutation as originating from palatalized consonants is phonetically plausible, more so than the notion that frontness jumped a consonant barrier from vowel to vowel.
The question is where this palatalization came from. Probably not British Celtic, as you say, but maybe from continental Celtic or Uralic languages. As you say, I'm not sure what evidence there is. Actually, there may be evidence for Gaulish palatalization: in French as in Old English, a was fronted, and Latin c was palatalized before it: cattus > chat. This may indicate phonological similarity in the substrate languages, British Celtic and Gaulish. Perhaps Schrijver has more on this, but I'd have to get his book. — Eru·tuon 22:56, 30 March 2015 (UTC)
You must not have heard of Vowel harmony before? There is no particular reason that i-mutation must have occurred with palatalisation of a consonant as an intermediate step. Furthermore, an explanation using consonant articulation as "transmission" does not work when considering a-mutation and u-mutation as well. U-mutation causes rounding, but there is no evidence that consonants were rounded by this process. CodeCat (talk) 01:18, 31 March 2015 (UTC)
Also, not all French varieties have palatalisation of c before a. The dialects of Northern France don't have it, nor did Anglo-Norman. The fronting of "a" is a common Anglo-Frisian feature, found also in Frisian and therefore cannot be Celtic in origin either. CodeCat (talk) 01:21, 31 March 2015 (UTC)
Yeah, I'm aware of vowel harmony, and I once thought it might be an explanation for i-mutation, but modern Finnish, Hungarian, and Turkish vowel harmony operates the opposite way from i-mutation: following vowels are assimilated in backness or rounding to preceding vowels, rather than root vowels changing to assimilate with following consonants or unstressed vowels. So, if old Uralic languages had the same pattern of vowel harmony, then vowel harmony is an unlikely explanation for i-mutation. However, the suprasegmental palatalization in Skolt Sami sounds much more like i-mutation.
I had forgotten or wasn't aware that palatalization before a didn't occur in Norman and other northern French languages, and as you say, Old English palatalization of c and g must have occurred on the Continent. I suppose that is why Schrijver doesn't claim palatalization of c, g as evidence of Celtic influence.
I didn't say i-mutation, or u-mutation for that matter, had to occur through the intermediary of consonants, merely that it seems more likely. It's odd if a phonetic feature is transferred from vowel to vowel without passing through the consonant in-between. And once rounding is present on two vowels in a word, it's highly unlikely that the consonant in-between will not be rounded (labialized): for instance, the ð in fǫður would almost certainly be labialized, at least in fast speech, since it's flanked by two rounded vowels. If this is the phonetic result, there are two ways to explain it: either rounding was transferred to the vowel and then the consonant, or to the consonant and then the vowel. There may be no specific evidence for consonant rounding, but neither is there for lack of rounding, so both options should be considered. I think the option that doesn't involve leaping phonetic features is more plausible, although plausibility may be best determined by looking at modern languages with vowel mutations and seeing how they do it. — Eru·tuon 03:00, 31 March 2015 (UTC)