Talk:Swedish phonology/Archive 1

Phonetics

When I did my edits, I just fixed the formatting - I apologize if I accidentally introduced any errors. I actually don't know much about the language, but in reading the article I noticed that it was asking someone to do an IPA transcription (which I do know how to do).

Anyway, I'll fix the phonetics. "put" for short 'a' should be "putt" instead. That's clearly a typo. Also, I don't agree with "weird" as being a model for IPA /e/. At least in American English, the sound in "weird" is /i/, not /e/. The closest sound to /e/ (at least in American English) is the 'ai' in "pail", as vowels preceding 'l' in English tend to automatically un-diphthongize. But I realize that in British RP the sound may be different...

ADDENDUM: I've clarified that pronunciations are for General American and not (for instance) RP.

ADDENDUM: I think the confusion over using "marsh" (not my idea!) as the English equivalent for 'sj' is that in most U.S. pronunciations - especially those using a retroflex 'r', as in the Texan dialect - the sequence 'rsh' gives a curvature to the tongue that creates a sound more similar to the 'sj' sound than anything else in English. In other words, it's a poor approximation that doesn't work for non-rhotic English dialects like RP and ENE/NY/Coastal Southern American.

Dave 06:14, Feb 4, 2005 (UTC)

Swedish pronounciation is similarly differentiated as that of English, given the differences between Australian, American, RP, and not to forget Indian English. I think the best thing we Swedish speakers can do, is to ensure that the IPA transcriptions are right, then it's your call as English speakers to make further transcriptions to English.
;-) --Johan Magnus 12:07, 4 Feb 2005 (UTC)

Indeed. If you make any changes to the phonetics, I'll fix the English.

Dave 17:42, Feb 7, 2005 (UTC)

More phonetics

Hello, native Swede here.

I see many (and sometimes rather large) problems with the IPA transcriptions, some of which I have altered. I discuss these and my other issues below, but I will not make upheaving changes myself; this post is intended as background information for future actions. Since I have no simple way of writing IPA characters directly here, I will use SMACKIE throughout this post.

Vowels

Firstly, there exist open allophones of 'ä' [E:] and 'ö' [ø:], namely [æ:] and [&:]. The shift is usually triggered by a following 'r', but some dialects ignore this or use the second variant all the time. I've added this information.

Secondly, I would not say that the short vowels are as uniform as they appear in the chart. For example, I don't accept out of hand that 'a', 'i' and 'y' must lose their tension when short; I can usually detect a clear difference between Swedish "sitt" and English "sit". Furthermore, 'e' can also retain its closed quality (and this isn't even necessarily a dialectal difference like the 'r+s' issue; there are many who would normally employ [E] that still differentiate between the two vowel sounds in, for example, "bekväm"). And short 'ä' should be [E], not [æ].

Consonants

Lots of strangeness here!

First of all: 'u' is not a back vowel. It is central, and even a bit closer to [y] than it is to [u].

On to the consonants:

'g' has three sounds: /g/ (hard), /j/ and /sj/ (soft). It's /sj/ in loan words like "garage", "genre", etc.

'k' does absolutely not have the sound of /sj/ -- and especially not the two variants (of at least three) listed! The soft 'k' phoneme is /tj/, which, like /sj/, is hard to define. I myself have found the best way to do it is to regard it as the unvoiced counterpart to /j/: [j°]. That is, an unvoiced approximant, which, like [w°], requires an increased airflow to be strong enough to be heard among the other sounds, and so can for all intents and purposes be labeled a fricative (but not one equal to [ç], since the position of the tongue does not change). The reason for this classification is that I feel that both [ç] and especially [C] (which are both used to denote /tj/ in various places) have too much turbulence for the actual spoken sound. Either way, equating soft 'k' with /sj/ is wrong -- the /sj/ and /tj/ sounds are wholly distinct.

Now, /sj/. As mentioned, there are three main allophones of this phoneme: [S], [s_] and [%]. How it is realized is heavily dependent upon regional customs, but there is also considerable overlap between the sounds. [S] is typically employed in loan words like "schlager" and "dusch" and is typically spelt "sch". However, not only can such words be pronounced using [s_] (or even [%]) instead, but [S] is also a direct "alternative" to [%], where the latter is typically used. For example, "sjösjuk" can be pronounced either ['%ø:'%ü:k] or ['Sø:'Sü:k]. The use of [S] in such circumstances seems to dominate among women, and so [S] is sometimes referred to as "fruntimmers-sje" ("womens' /sj/"). [s_] is usually the result of 'r+s', but it's not uncommon (especially in the northern parts) to use this sound for other types of /sj/; this is, I reckon, usually done for "proper" [S], but may also replace [%] (['s_ø:'s_ü:k] is not at all alien). So, in summary, the various realizations of /sj/ are a quagmire indeed (and there are probably even more sides to it than I have tried to bring up here). :)

Next up, calling the /sj/ realization of 'ch' hard and the /tj/ one soft (I presume this is what is meant, but, as outlined above, neither transcription is satisfactory) is not consistent with the 'sk' case: both sounds should be considered soft. Furthermore, the variation between /sj/ and /tj/ for 'ch' is also very unpredictable; some pronounce "chans" /sjans/, others say /tjans/. Now, actually, there is a hard variety of 'ch', namely [k] (as in "och"), but both this and the soft variation are wholly disconnected to the soft/hard vowel environment. Hence, the inclusion of 'ch' in the table, as it is, is misleading.

Continuing with the soft/hard consonant combinations, I would also object to the "single-type" ones' being labeled "hard" -- it would be better to just have one IPA column and instead have multiple rows for 'sk' ("sk (hard)" and "sk (soft)", for example), and possibly 'ch'. Also, as already explained, 'kj'/'tj' as [S] is inaccurate while the /sj/ entries need to include [S].

Finally, I would add the sequence 'gn', which is pronounced [Nn].

English

I would also like to comment a bit upon the English "equivalents".

First of all, I'm not so sure that having them is such a good idea in the first place. This is partly due to the fact that English and Swedish differ much in several basic phonetic areas, such as the purity of vowels and the presence/absence of affricates; therefore, any attempt to approximate the pronunciation using only English data will never give an adequate picture of the actual Swedish conditions. And partly, as Dave pointed out, in trying to circumvent these limitations by using more or less specialized phonetic occurrences, all those who are not familiar with the case at hand will be unable to understand what is meant. Can't we trust people who are interested enough to look at a phoneme table to interpret (or at least be able to look up) IPA only?

Now, if they are to stay, I would suggest making them more (in some cases, much more) refined, so as to not only constitute the one phoneme in English which is perceived as being closest to the Swedish one in question, but rather function as a reasonable approximation, in terms understandable to non-phoneticians. The current "level" is much too inconsistent on this point.

For some concrete examples, consider that:
1) it should be clearly stated that [e], [o] and [ø] are not diphthongs, so that only the first part of the English sound should be used
2) the description for [ü] seems fairly useless to me; would not the sound explained actually be [Y]?
3) [â] is a very poor approximation of our short 'a' [a]
4) stating that short 'å' is [O] == 'au' in "caught" is ambiguous, since if I'm not mistaken that is only valid for BE -- in AE, this vowel sound is usually [A] (and there is therefore a direct contradiction with the column header)

Reactions?

--LRC 20:39, Feb 9, 2005 (UTC)

My chief reaction is that one of the most frustrating aspects of learning Swedish is the exaggerated sureness many Swedes possess, that their personal dialect is the one and only correct variant. Rather than changing IPA-transcription back and forth, it would be of great value if it could be denoted which dialects use the different realizations of the variant vowel- and consonant sounds.

As a sidenote, I discovered that you, LRC, changed the notion /sound/ to [sound], which I find astonishing, as I'm used to the convention that brackets are used for IPA-transcriptions of words but slashes for sounds. /Tuomas 00:33, 12 Feb 2005 (UTC)

Yes, I agree -- and that is precisely why I raised the issue; I find many of the current (or recent) claims to be too specific to be applicable to a general article about the Swedish language as an official entity. This is also the reason why I only made minor changes to the article itself, since if everyone changes stuff back and forth between "their" correct variants, things could get ugly (or at least confusing and misleading for people who actually want to discover something from reading the article).

Now, your suggestion about identifying dialectal variations is, from a factual point of view, very good. However, there is also the question about how much of this information should be included in the main Swedish article, and what could/should be lifted to a separate article about Swedish dialects -- one which may even fit better in the Swedish version of Wikipedia. I myself have no objection to including such data in this article, but I am not familiar with the current view on how extensively this kind of "narrow" information is to be included in general-purpose articles -- for if we do start down this road, it will get extensive and would require the participation of a good number of Swedes. The alternative way would be to keep the article about "official Swedish" (rikssvenska, if you will) and just note that there are variants regarding certain things (such as my attempt with the open allophones of 'ä' and 'ö').

And it is against this background that my post should be seen -- I reckon that this article should primarily (as in, where not stated otherwise) deal with rikssvenska and I tried to mark the portions where the text should be called into question on this point. I do believe that I tried to present a neutral case based on the Swedish that is taught in schools, and I apologize if I still came across as, as you put it, possessing an exaggerated sureness that my personal dialect (emphasis mine) holds sway, for that was not my intent. It should, however, be noted that my emphasis on the /sj/ vs. /tj/ issue is due to the fact that Swedish does possess both phonemes and they do form minimal pairs -- while it is true that not all speakers of Swedish may recognize the difference, you would never know about their co-existence from reading the current article. And this, I feel, is a serious fallacy, but I am uncertain as to how to go about integrating the information, as I tried to lay out earlier. In fact, there is probably material here for a whole separate article about sje and tje.

Finally, regarding // vs. []. As far as I know, the convention is that phonemes are put between slashes, whereas the actual phonetic transcriptions (of phonemes, morphemes or words) are put within brackets. Example:

• in American English, /t/ can be realized as [t], [t^h] and [2]

Seeing as how the schwa sound is not a contrasting phoneme in Swedish, but rather (as stated in the article) an allophone of 'e', I decided to change it. I must say that I find nothing particularly astonishing about it.

--LRC 13:26, 12 Feb 2005 (UTC)

Maybe I should confirm that I totally agree on / phoneme / vs [ phone ]./Tuomas 13:27, 22 Mar 2005 (UTC)

My view, as someone who lived for four years in Sweden, and had started studying the language almost ten years before, is that the different high status varieties is a phenomenon extremely frustrating to cope with for students of Swedish, once the problem is recognized, and that these therefore better are given a prominent and non-biased coverage in this article. Details, peculiarities and other dialects are (imho) better to cover in specialized articles. If there somewhere should be a coverage specificly or exclusively for "rikssvenska", then that is preferably in the Sveamål article, which I guess rikssvenska in this context should be understood as.

In my opinion, this English language article probably best is (primarily) geared towards readers knowledgable and interested in general linguistics, but it can't be forgotten that foreign students of Swedish are likely to edit the article from time to time. Teachers of Swedish, and textbooks on Swedish, may have rather conflicting views on what's a "correct" realization of the initial sounds in skjära, tjära, stjärna, gärna, kär and skjär — to mention one of the most problematic issues. The article should not invite to changes back and forth by "forgetting" to mention these differences. Bättre att stämma i bäcken än i ån!
--Ruhrjung 14:19, Feb 12, 2005 (UTC)

My use of the term rikssvenska was unfortunate, since it as the article (correctly) states is both controversial and vague. In my circles, so to speak, it refers to some kind of (more or less imaginary) neutral and standardized dialect of Swedish; that is, a common template (or compromise) for the Swedish language as a singular entity. The news anchors of the national television channels are probably those who most would agree speak this "dialect". A better term for what I mean would perhaps be Swedish as it is taught in schools and is given in domestic dictionaries, which imho would be the best starting-point for a foreigner who wishes to learn (about) the language -- s/he can specialize hirself in a variety of choice later.

As for your examples: I reckon anyone would be hard pressed to find any textbook which does not state that skära, stjärna and skär use "sje-ljudet" whereas tjära, kärna and kär use "tje-ljudet"; they may disagree upon how to pronounce them (and I know that some people are incapable of distinguishing some /sj/ relizations from a /tj/ one and vice versa), but the classification is certainly there. So, although there is considerable disagreement, if you will, regarding the realizations of /sj/ and /tj/, which I also tried to bring forth before, I will repeat that they are contrasting phonemes and should be treated as such, which I think the current article fails to do.

--LRC 18:30, 12 Feb 2005 (UTC)

I started to comment on this, but felt my old frustration shining through too obviously. Jumping over the comments, I would only like to stress that natives and non-natives most probably view this differently.

What about three columns for pronounciation in the tables, one for high-status Götalandsmål (like Ljungby-variety or Lund-variety of Rikssvenska), one for "utjämnad" Finland-Swedish (that according to my perception phonetically would cover also the high-status Norrland-variety rather well) and one for high-status Svealand pronounciation (like Uppsala-variety of Rikssvenska)? Wouldn't that make these things less confusing?

--Ruhrjung 20:56, Feb 12, 2005 (UTC)

That might be a good idea (although I would emotionally protest against the notion of any Småland-dialect to be denoted as high-status variety. ...But I think I understand your thoughts. Educated, maybe even rural[1], dialects from the Ljungby area may be perceived as rather easy to comprehend for foreigners, and closer to the high-status variety people sometimes adopt in Lund than for instance the speech of Carl Bildt).

--Johan Magnus 11:22, 15 Feb 2005 (UTC)

Pronunciation files and phonetics

I've added a bunch of pronounciations recorded by myself for almost all the vowels. Do you think we still need the English approximations?

Also, I definetly think we should remodel the phonetics tables. It doesn't make any sense to sort a guide on phonemes according to spelling. That should somehow be explained under Orthography. You can see how I structured it at svenska over att Swedish wikipedia. That seems more logical to me. - karmosin 07:36, Mar 5, 2005 (UTC)

Well, I went on and got bold and redid the tables. I've always been skeptic to approximations of phonemes, and now that I've recorded pronunciation files, it feels quite unnecessary. I know the .ogg-format isn't all that popular, but it's what we've got and it's probably better that we try to push it as much can than to ignore sounds altogether.

I'm going to add comments about dialect variations on all the phonemes later, as well as a proper source. - karmosin 23:44, Mar 8, 2005 (UTC)

I guess you will reinsert the contents of the remarks that got lost?

• Swedish pronunciation of consonants is similar to that of most other Germanic languages, including English.
• Some consonants and consonant combinations have both "hard" and "soft" sounds. These consonants are hard when preceding a back vowel (a o u å), and soft when preceding a front vowel (e i y ä ö).
• T, N, L, and to a lesser extent D are pronounced dentally (with the tongue touching the teeth instead of the hard palate).
• Following a long vowel, the combination R + an alveolar consonant (T, D, N, L, S) may be pronounced as a single retroflex consonant (ʈ, ɖ, ɳ, ɭ, ʂ). This change occurs in northern and middle Swedish dialects, and is not hindered by word-boundaries.
• The sound "sj" or soft "sk", (IPA ʂ ~ ɧ ~ ɸ) is listed here as "sh" as in "marsh", although there is considerable variation inbetween speakers of Swedish. See Voiceless dorso-palatal velar fricative for more details.

/Tuomas 16:50, 9 Mar 2005 (UTC)

Yup, but more detailed. Don't worry, they'll be up in a week or so, and more specific to boot.

The old tables weren't bad, but they were too concerned with orthography. I don't think orthography and phonetics shouldbe should be mixed up like that. I could use some help with the IPA. How do I write the two different kind of accents properly? You know, to separate "stegen" from "stegen" ("ladder" from "the steps"). - karmosin 16:56, Mar 9, 2005 (UTC)