Talk:Esperanto phonology

[Untitled]

I tried linking to eo:Egzismo, but the link won't display for some reason. Note that if you do make the link, the Esperanto article is factually incorrect. --kwami 21:05, 20 Apr 2005 (UTC)

"More common"?

Within compounds, an epenthetic vowel is added to break up unacceptable clusters, though what is "acceptable" varies from speaker to speaker, and also depends strongly on the language medium: Although you will find the compound vortprovizo "word stock" in Vikipedio (Esperanto Wikipedia) with an rtpr cluster, it is rather more common to see, and much more common to hear, the epenthetic form vortoprovizo.

"vortprovizo" gets 3800 Ghits to 80 for "vortoprovizo". I am going to change this paragraph to reflect that. Personally I think I've heard "vortprovizo" more often than "vortoprovizo", too, but I'm not going to go so far as assert that it is absolutely more common in speech, for want of a proper speech corpus analysis. --Jim Henry | Talk 15:19, 14 July 2005 (UTC)

Okay. Personally, the Usonians I know, and from what I remember Don Harlow, wouldn't normally try pronouncing a cluster like that. Well, a few would, but not many. But such things are going to be strongly influenced by one's native language. In West Africa and in Japan, inserting the o in such words in speech seemed nearly universal, though I admit I'm going on memory and don't have a corpus available either. kwami 22:55, 2005 July 14 (UTC)

V vs. Ŭ

Removed this from the intro:

(This is not strictly true, as although v may appear between a consonant and a vowel while ŭ only ever appears as the second part of a diphthong, there are minimal pairs such as naŭa (ninth) and nava (of naves, concerning naves) which are distinguished only by these two phonemes.)

This was discussed elsewhere, but looks like I forgot to cover it in this article. Could be a case of /nav.a/ vs. /na.va/. Anyway, it doesn't belong in the intro. kwami

There is no way that v and ŭ can be treated as allophones: as mentioned above the former may only oocur pre-vocallically whilst the latter is usually only the second part of a diphthong, but these positions are NOT mutually exclusive as both can occur intervocalically. And despite the possibility of the minimal pairs being analysed with different syllabic structures, this is not taken into account when determining the phonemic status of a speech sound in any language. There are good reasons for this. 1) Syllable boundaries are nto clearly defined in languages such as English and Esperanto where syllable structure can be CV or VC and are often a matter of personal opinion 2) taking the morphonological view, such as used by John Wells, the syllables boundaries can be analysed and coinciding with morpheme boundaries giving [nav.a] and [nau̯.a] 3) as suggested some speakers may not distinguish between v and ŭ, which if they were true allophones would mean they could be given the same phonetic realisation without a loss of contrast (such as the realistation as English /l/ which is always 'dark' in soem varieties). But if both Esperanto sounds are realised the same, the contrast between "naŭa" and "nava" will be lost, regardless of where one theoretically places the syllable boundary as this boundary is not explictally given in the pronunciation.

For similar reasons, the offglides in English closing diphthongs cannot be treated as allophones of the semivowels /j/ and /w/, which only appear in syllable initial position. Matt.

Ŭ may be a non-syllabic /u/ rather than /v/ or /w/.

What would you say about the Ŭ sound in English cooperate? Might be similar. kwami 09:32, 20 January 2006 (UTC)

The glide in 'cooperate' is part of a diphthong /@U/ which is treated as phoneme in itself, not a sequence of /@/ plus /U/. This kind of analysis could be applied to Esperanto, in which case 'aŭ' and 'eŭ' are phonemes. In this case whatever 'ŭ' represents it still cannot be an allophone of /v/. The point I am trying to make regarding 'v' and 'ŭ' is that if the utterance /naPa/ was produced (where /P/ represesnts a voiced labio-dental approximant) what is beign said? The sound could be interpreted as either [v] or non-syllabic[u](or [w]), but this then leads to the question whether 'nava' or 'naŭa' is the intended word. I understand that in many Slavic languages there is an alternation between a labio-dental fricative or approximant pre-vocalically and a vowel-like glide word-finally or before a consonant, but because an intervocalic position is by virtue also pre-vocalic, the realisation will always be the fricative/approximant and as such is NEVER used contrastively, regardless of syllable boundaries. therefore the system in these languages is different to that used in Esperanto.

Quite true. I don't think the article makes the claim any longer that Ŭ and V might be allophones. I'll double check. kwami 18:30, 20 January 2006 (UTC)

Well it mentions it in the introduction as a possibility whcih seems rather unnecessary to me.

Hx/K

Include the alternate forms like orhxestro/orkestro. --Error 03:00, 29 November 2005 (UTC)

Done. Orĥestro is obsolete, but the merger is found in other words. kwami 09:10, 29 November 2005 (UTC)

At the end of the section on Ĥ/K, it says that Esperanto roots cannot end in h. This is not true. Note the verb subtrahi. — Preceding unsigned comment added by Gary Rector (talk • contribs) 10:33, 2 January 2011 (UTC)

Assimilation

-- I have a few questions about assimilation. I see people saying that "abs" is pronounced "aps". In Portuguese (Brazilian), we have words starting with "abs" (absoluto, absurdo) and nobody pronounce them as "aps...". If there were words with "abz", "aps" and "apz", it would hard to distinguish among them, but possible. Maybe the reason is that we usually introduce a short "i" vowel between the consonants. Could i say that "absolute" in Esperanto should be pronounced as abisolute, then :-P ???? I think introducing a vowel it's no less valid than voicing or devoicing one of the two consonants (Why the first? Why not change the second?).

Now for "kz". To me, pronouncing it "gz" would sound very strange. I think my pronunciation of "kz" is more like "ks", but never "gz". But, still, when pronouncing "ks", i think i do it differently (and i don't introduce a vowel between the consonants). Maybe the best representation would be "ksz" and "kss"? How can i tell which sounds i'm really saying? And again: why favor one pronunciation change (k -> g) over the other (z -> s)? --Yuu en 23:38, 10 December 2005 (UTC)

Good questions. The answer lies in the fact that Esperanto is based on Slavic phonology, which relies on assimilation rather than epenthesis (like adding your i) for consonant clusters like these, and Slavic voicing assimilation works backwards (/kz/ to [gz] and /bs/ to [ps]) rather than forwards (to [ks] and [bz]).

Zamenhof specifically mentioned the pronuciations [gz] and [ps] as acceptable, although he did say that strict adherence to the 'one letter one sound' principal would result in no assimilation at all. (As a strict rule this would be nearly impossible, however, as all human language has assimilation to some degree.) These wouldn't appear so bizarre to a Spanish speaker, for example, so they're not confined to Slavic.

Also, the whole reason for the odd spelling ekz in the first place is to disambiguate these words from the large number that start in eks because of the prefixes ek- and eks-. If assimilation made them both [eks], then we would no longer be able to spell Esperanto words from their pronunciation, which would be a real problem. Since no word contains /gz/, the [gz] pronuciation is safe. Personally, I think that Zamenhof chose kz rather than gz so that the result wouldn't look overly bizarre, but a lot of people wish he'd simply gone for gz.

As for your vowels, a slight epenthetic schwa for [ab^əsolute], or a similarly obscure vowel like an [i] (and likewise perhaps a glottal stop in [ek’zemple]) would be acceptable, so long as people didn't hear it as a real vowel and it didn't noticeably add any syllables. But [abisolute] with five syllables and a full [i] would mark you as having a very strong accent. kwami 00:42, 11 December 2005 (UTC)

-- Wow, thanks. Very good answers! :) I thought i would start a little flame war, but your explanation is very well written! Now i see the point of "gz" and it's nice to read some "fresh" information that "short obscure vowels" or a glottal stop can be acceptable pronunciations (this article seemed a little one-sided to me because alternative pronunciations are not considered). I also see that a full [i] would mark me as having a strong Brazilian accent. I guess most beginners here pronunce "kvar", "kvin" as "ki var", "ki vin". :) And the Japanese would most likely say "ku var", "ku vin". :) Now i need to get rid of my habit of pronouncing "is still" as "is-i-still" :) --Yuu en 20:12, 11 December 2005 (UTC)

Yes, I would hear "is-i-still" as is he still !

There was resistance to adding any discussion of allophony to the article at all, except to say that it's wrong, and my additions were reverted for a while. Many Esperantists believe that 'one letter one sound' means there can be no variation. Impossible, of course. I only discussed the allophony that's covered in the Esperanto lit for this article; what I've written here about epenthesis is based on personal observation of good Esperanto speakers and would count as original research. Personally, I use full voicing assimilation (no epenthesis that I'm aware of), an slight Italianesque lilt, and a Slavic approximantic /v/, and people can't tell where I'm from from my accent. Rather fun!

kwami 21:21, 11 December 2005 (UTC)

-- Now i wonder why "kv" is not a case for assimilation into "gv" (or "kf")... Most likely, the answer will be "Slavic phonology" :-). Is "kz" phonologically any harder than "kv"? Now practicing: kza, kza, kza, gza, gza, gza, ksa, ksa, ksa, kza, kza, kza... They sound pretty different to me :-P. But i wouldn't bet my life on this. ;-) -- Yuu en 07:32, 3 June 2007 (UTC)

Kv is different from kz because for many speakers (certainly for many Slavs), v isn't a fricative as z is. It's "supposed" to be, according to most books, but you only need to go to a UK to hear that kv is often pronounced somewhere between [kv] and [kw]. Kz doesn't have that option. kwami 07:18, 17 September 2007 (UTC)

Orthography & pronunciation

I have changed the table in the "Orthography & pronunciation" section so as to remove the explicit (and, as it stood, absolutist) reference to a kz exception, and refer the reader to the (much more nuanced) discussion under "Assimilation"—for two reasons: (1) Previously, the table seemed to espouse egzismo unambiguously, as though it were the only option; (2) There was no reference to other possible assimilations, leading the unwary reader to suppose that kz–>gz was the only one.

Yes, good call. This table used to be an article of its own, with a very schematic outline of pronunciation. kwami

I was also sorely tempted to include an entry for consonantal ŭ, since no matter how rare (and arguably obsolescent) it is, it can occur in standard Esperanto—and not just in words like "ŭato" and proper nouns, but in the zamenhofa "ŭa!", and of course the name of the letter itself, "ŭo". I only refrained from doing so because it would have required more than just the insertion of one line for the article to be consistent, due to its assumption that "initial ŭ violates Esperanto phonotactics". I suggest that the table is incomplete without it (even if it needs to be flagged as "rare"), but it may involve changes elsewhere in the text. Vilcxjo 02:00, 15 December 2005 (UTC)

The name of the letter ŭo is really the only word in the basic vocab. Onamatopoeia exists outside a language's normal phonology - English would be a click language (tsk! tsk!) with syllabic sibilants (shhh!) if it were included, for example. We should probably add ŭ as a footnote to the table. kwami 06:57, 15 December 2005 (UTC)

General and Specific Comments

You have provided an interesting and useful description of Esperanto phonology. I use description in the sense of descriptive phonology as opposed to normative phonology. I want to argue that what Esperanto primarily needs is a normative phonology, i.e., a set of phonological norms.

First let me indicate what qualifications I have and don't have to speak on this matter. I have taken all the courswork for a PhD in linguistics (30 years ago), but never wrote a dissertation and have never worked as a linguist. Thus I know considerably more about linguistics and specifically phonology than the average layman, but I am by no means a professional phonologist.

Esperanto is fundamentally different from natural languages in the following sense. For every natural language there is a community of people who use the language every day for every purpose and whose children acquire the language by imitation of what they hear. Most speakers are unaware of any grammatical description of their language or of any set of rules or norms. (Even in the case of educated speakers of major languages, they become aware of the rules only after they already speak the language and it is doubtful that their knowledge of these rules affects their use of the language very much.) A linguist doing field work on a given language describes some aspect of that language either in the hope that his description will shed some light on the general phenomenon of language, or just for the value of the description. His approach is fundamentally non-normative.

In the case of Esperanto there is no analogous community of native speakers. Almost all speakers learn the language as adults on the basis of rules or norms, usually in non-ideal conditions. Lots of people all over the world study English and other foreign languages in the context of well-organized courses with competent instructors and often well-designed languages labs. Very few people learn Esperanto under such favorable conditions. Even the handful of "native speakers" of Esperanto learn the language from people (their parents) who learned the language as an adult. Whatever idiosyncracies are in the language of the parent will be acquired by the child.

Some phonological matters are routinely dealt with in Esperanto textbooks, and many speakers (certainly not all) succeed in conforming their pronunciation of the language to those norms. But many matters (aspiration of unvoiced stops, regressive assimilation of nasals, secondary stress, etc., etc.) are either never mentioned or dealt with only very cursorily. Relative to those matters, a linguist doing field work on Esperanto speakers will learn nothing about Esperanto; he may learn a lot about the phonology of the native language of those Esperanto speakers. Because of the way in which every speaker of Esperanto learns the language, many of the empirically observable characteristics of his pronunciation are characteristics of his native language, not of Esperanto.

I should clarify that relative to many matters, e.g., aspiration, secondary stress, prosody in general, bilabial fricative (Spanish) versus labio-dental fricative (English), apico-dental t and d (Spanish) versus apico-alveolar t and d (English), I don't propose any norms. Speakers of Esperanto will simply retain the speech patterns of their native languages.

All of this having been said, let me turn to those matters which I am actually interested in, for example, regressive voicing assimilation. It is not the end of the world to introduce a voicing assimilation rule that caused 'absoluta' to be pronounced [apsoluta] or 'obtuza' [optuza], but I don't think we should do this unless it is necessary. There is a syllable boundary after the 'b'. English dictionaries say that 'b' is pronounced [b] in the corresponding English words, and that is what I hear in my own speech. (I do understand that native speakers often don't hear their language accurately, and if a phonetician were to say to me that I actually say [apsolut] I will accept his judgement.) For the time being I hear [absolut]. If we are going to have a voicing assimilation rule, we need to say how extensively it applies. It will be very difficult to achieve agreement, and the simplest solution is to say that there is none. (I am being normative not descriptive.)

In the case of 'ekzemple', there is a syllable boundary between 'k' and 'z', and it isn't hard to say [ekzemple] if one wants to. Nor is it hard to say [okdek] (your example), or [ekdecidi] ('ekdecidi') or [ekzomi] ('ekzomi') or [ekbani] ('ekbani') or [sepdek] (70). So far as I know, morpheme boundaries don't have to have phonological consequences, although they sometimes do. If we start changing the language to accommodate everyone who has some pronunciation difficulty, where do we stop?

Relative to place assimilation of nasals, I think it would be reasonable to say that there are two nasal phonemes /n/ and /m/, and that /n/ assimilates in position to the following consonant and does not occur before labials. I am not trying to word this rule with precision. The assimilation of /n/ to velars and palatals is different from the above cases, because there is no collision with another phoneme. I agree with the spelling 'Vashintono'.

From a normative point of view, I don't think we should mention vowel length (duration). In the abstract all vowels have the same duration. Actual speech will depend on what the speaker brings with him from his native language.

Again from a normative point of view, we need rules to say when we insert a linking vowel (usually 'o') in compound words. It isn't logical to use 'vortprovizo', as most people do according to Google, but 'partopreni'. Since a syllable boundary occurs between 'rt' and 'pr', it probably isn't helpful to think of 'rtpr' as a cluster. The relevant question is whether 'rt' is an accepable coda, and according to the sonority principle that you mention, it is. For consistency 'partopreni' should become 'partpreni'.

I have said enough. Your article was interesting and helpful.

Not yet. One more comment. Relative to what Vilcxjo said about the ŭ before vowels, I agree that it is rare but not that it is obsolescent. I think it will become increasingly common as proper names are adapted to Esperanto. In NPIV several Chinese place names have it. I regard the v in Nikaragvo, Gvatemalo, etc. as very contrived and undesirable, but these words are established. If we decide to Esperantize the states of Mexico (those of the US are already Esperantized), how will we represent "Chihuahua" [ĉi-wa-wa] where I use w for the back on-glide? I suggest "Ĉiŭaŭo". --Neal Parker

Which I would read as ĉiu aǔ-o ! Chihuahua could also be expected to be "Ĉivavo". There's nothing undesirable about that except that it isn't Anglo-Romance. (IL, anyone?) As for the v in the other country names being "contrived", I don't agree. It's normal Central-Eastern European orthography, which of course is what Esperanto is. You find "Gvatemala" and "Nikaragva" in Croatian, Icelandic, Slovenian, Lithuanian, and Latvian, and really in Polish too, once you allow that Polish <w> equals <v> in those other languages. Yes, extra-European influence may produce a true /v, w/ contrast eventually, but we're not there yet.

You're very welcome (I hope I don't offend people by taking too much credit!)

All pedagogical material is going to be normative, but that is specifically not the purpose of Wikipedia. That's a Wikibooks thing.

Thanks for the partopreni example. That paragraph needed fixing up, and you've just given me an idea on where to go with it. kwami 06:04, 6 January 2006 (UTC)

I did not intend to imply that [gv] and [kv] are intrinsically unnatural or undesirable, and I have no quarrel with general words like 'kvartalo' or 'akvo'. When we borrow proper nouns from a specific language, we try to stay as close as possible to the original form, and it seems obvious to me that the only reason we use 'Gvatemalo' instead of 'Gŭatemalo' is the taboo against the use of /ŭ/ as an on-glide.

/j/ is both an off-glide and an on-glide. The phonological system would be more symmetric if /ŭ/ were also both an off-glide and an on-glide, and having been trained as an abstract mathematician, I like symmetry. The /ŭ/ phoneme already exists. Not to permit /ŭ/ + vowel seems very artifical, since the combination is easily pronounced and since there are very few restraints on other pronounceable combinations (only the 3 diphthongs ij, iŭ, and uŭ come to mind). We accept intrasyllabic consonant clusters like [ks] ('kseneno') and [pn] ('pneŭo') that many people find hard to pronounce, not to mention [sc] ('scii'), and if someone were to introduce a previously non-existent cluster like [zv] in, let's say, 'zvotuto', no one would object.

Since 'Chihuahua' and 'Guanajuato' are very specifically Mexican words, I object to making the Esperanto version conform to a Slavic or Baltic model when we have available what we need to do a "natural" conversion.

In cases where the vowel /u/ is not accented, there is very little difference between [uV] and [ŭV] as in the words 'trotuaro', 'puerpera', 'tualeto', 'duodeno', and 'Kuala-Lampuro'. It seems to me that the difference is mostly a matter of relative duration of the two vowels, about which we have no rules in Esperanto. Perhaps the taboo against the back on-glide is less real than we think.

I recently sent an analysis of compound words like 'partopreni' and 'vortprovizo' and whether a linking 'o' is present or not to akademia-diskuto in Yahoogroups. I would like to forward it to you for your opinion. You can contact me at [email protected]. Neal Parker Jan 7, 2006

If you want a naturalistic conlang, you should check out Interlingua. Esperanto generally does not contort itself to fit an alien phonology. The reason Esperanto does not allow ŭ as an onglide is because it is not possible in Polish, German, Russian, or Yiddish, Zamenhof's fluent languages. It's not a 'taboo', any more than English has a 'taboo' against tones, it's an integral part of the phonology. Check how a German or Russian pronounces it. If Ĉiŭaŭa is actually pronounced Ĉivava, then that is how it should be spelled. By your logic, Esperanto should be a tonal language in order to properly accommodate Chinese and Vietnamese names, and should have rounded front vowels to accommodate French and German names, and should have /θ/, /ð/, and /ŋ/ to accommodate English. No one would be able to pronounce it! Either that, or we abandon a phonetic alphabet. (Tones are much more important to Chinese than [w]/[v] is to Spanish. Besides, Esperanoto v is often somewhere between a [w] and a [v]; an Esperanto Ĉivava wouldn't distort the Spanish nearly as much as English [mεksikəʊ] does.)

As for mathematical symmetry, my background is in physics, so I can sympathise. But it's completely impractical. Mathematically and logically designed conlangs like Logban are nearly impossible to learn. (In fact, I don't think anyone ever has learned Logban, not even its creators!)

Zamenhof intentionally avoided a /v/-/w/ contrast. He dropped or omitted a lot of contrasts that various European nationalities have difficulty with in the interests of making the language more universal: palatalization, rounded front vowels, central vowels, dental fricatives, /ŋ/, some of the diphthongs, etc. You may feel that /w/ is important enough to change Eo for; personally, I would like to add /ŋ/. Many French would probably like to add /y/ or /ø/. And of course many Chinese think it ridiculous to not have something as basic as tone. (More languages distinguish tone than distinguish /v/ from /w/!) We have to draw the line somewhere, and there isn't any real difficulty with leaving it where Zamenhof drew it.

Go ahead and email me your analysis. (From my talk page.) Might be something to post here, though. kwami 02:17, 8 January 2006 (UTC)

I would be completely opposed to the introduction into Esperanto of front rounded vowels or interdental fricatives or tones. Since I am not opposed to the use of the back on-glide ŭ in new borrowings I need to construct a logical argument that there is an essential difference between the two cases.

In the first case we would introduce new phonemes into the phoneme inventory, and I perceive that to be a major change. I think it is very important that Esperanto be stable, and thus I am opposed to unnecessary change. Some people would like to spell 'Barato' and other words of Indian origin with 'Bh'. I have argued against that on the grounds that we should not introduce 1) aspirated voiced stops (if that is what 'Bh' is) or 2) new difficult to pronounce consonant clusters (in the case that the sound is a sequence of 2 consonants). There seems to be disagreement among phonologists about the best representation.

In the second case, /j/ already has 2 allophones: the front off-glide, which occurs after vowels to form diphthongs, and the front on-glide, which appears before vowels. /j/ represents tongue movement either from (off-glide) or to (on-glide) the position of the high front vowel /i/. /ŭ/ is a closely analogous phoneme involving the high back vowel /u/. I am merely proposing that we allow /ŭ/ to behave like /j/ already behaves. To me that is very minor change. It improves symmetry without introducing any practical difficulty.

About a year ago I reported the following facts about German, Polish, and Russian in a message to the group akademia-diskuto. No one disputed them, and a message from the Norwegian phonologist Otto Prytz tended to confirm them. In German [kw] and [gw] do not occur in native words. [kv] occur in a few native words ('Quelle' = source, 'Qual' = pain) and in some words of Latin origin ('Quartal' = quarter). I have found in my dictionary several loan words pronounced with [kw]: 'Kuomintang' from Chinese and 'Guanako' = guanaco, 'Guano' = guano, 'Guajave' = guayaba, 'Guarani' = Guaraní from Spanish. I don't find [gv] in any word. The combination [tsv] is fairly common, as in 'zwei' = two.

In Polish [w] before vowels is common, and in native words the letter for that sound is 'l' with an oblique hypen through it. Here I will use an ordinary 'l'. For example, 'bialy' = white, 'cialo' = body, 'Bialistok', 'glodny' = hungry. [kv] and [gv] also exist, and they appear to me to be common. The letter 'w' is pronounced [v]. For example, 'kwiat' = flower, 'kwiecień' = April, 'gwiazda' = star, 'gwizdać' = whistle. I don't know Polish, and these definitions may not be syntactically correct.

In Russian [kv] and [gv] occur, and I believe that [kw] and [gw] do not.

I am having difficulty understanding the meanings of your statements "Zamenhof intentionally avoided a /v/-/w/ contrast." and "More languages distinguish tone than /v/ from /w/." To me /v/ is an ordinary fricative (labio-dental phoneme in English, bilabial allophone of the bilabial stop in Spanish) which I believe to be very common (primarily in the labio-dental form) in the languages of the world. The tongue is stationary as in vowels and in other fricatives, and thus the sound can be prolonged. /w/ is a glide which represents tongue movement. Glides cannot be prolonged. To me they are fundamentally different sounds, and they can be adjacent as in the Spanish word 'abuelo' [avwelo] = grandfather. Like other fricatives /v/ participates in consonant clusters, some of which are more common than others.

In English you could say that /v/ and /w/ contrast because of pairs like 'west' and 'vest'. Are you saying that in the languages of the world, that is an unusual situation, that generally if /v/ occurs before vowels then /w/ does not, and vice versa?

'Chihuahua' is pronounced [ĉiwawa] in Spanish. It won't be the end of the world if the Esperanto word is 'Ĉivavo', but as yet I am not convinced of the need for that.

There is a saying, "Everything is easy if you know how." I don't contribute to Wikipedia on a regular basis, although I did write the article on 'Konsonantoj' in the Esperanto section. Thus I am not familiar with all the procedures and possibilities. When I click on your name, I go to a page which I assume to your 'talk page'. I had been there before, but I have never been able to discover how I would send e-mail to you. Please tell me what to do. Thanks. Neal Parker January 8, 2006

On everyone's talk & user pages, there is a 'toolbox' of links under the Wiki logo, navigation, and search toolboxes. One of the links is 'email this user'. Not everyone has registered their address, but I have.

Okay, /w/ vs. /v/. There are few languages with both. Many have only [w]; relatively few have [v]. Often a language reported to have one or the other actually has neither, but rather has something like a [ʋ] - for instance, Hindi and Mongol are like this. The Indians, Chinese, Japanese, Russians, Germans, and Turks, among many others, have great difficulty distinguishing /w/ from /v/. (You're right about Polish. I'd forgotten about that. But that's a recent development that is unusual for a Slavic language.) You're proposing to introduce a new phonemic distinction, I would guess one that you think is natural only because it has separate letters in the Latin alphabet. Esperanto ŭ is allophonic, or at best a marginal phoneme. (Whether it's allophonic with u or with v is not important here.) When Z adopted French words and thought a v would be overly distorting, he used u (vualo, trotuaro, etc.) That's what Russian does. Many languages preserve a w-v distinction in foreign words in writing, but they don't pronounce it. For example, Belarusian (which also has the letter ŭ) allows ŭ at the beginning of foreign words like tweed and Wales, but they pronounce it as a /v/. That is, this is a non-phonemic spelling.

It's not that, say, in Russian [kv] occurs and [kw] does not. It's that the monolingual Russian speaker does not distinguish [kv] from [kw], and cannot hear the difference any more easily than you or I can hear the aspiration of English /p t k/ or the diphthongs in English /e o i u/.

You say "I am merely proposing that we allow /ŭ/ to behave like /j/ already behaves. To me that is very minor change. It improves symmetry without introducing any practical difficulty." But there is a practical difficulty: For the majority of the world's population, this would make the language significantly more difficult. And what does symmetry have to do with it? The point of a language is to communicate, and a phonemic /ŭ/ would hinder communication.

Esperanto <j> represents two sounds: a consonant /j/, as in jes, and a non-syllabic /i/, as in kaj. Eo also has a non-syllabic /u/, spelled <ŭ>. But it doesn't have a consonant /w/, for the reasons just mentioned. (You could argue that it shouldn't have a few other consonantal distinctions either, like ĝ vs. ĵ, but it's too late to change that, and in any case it doesn't carry a very heavy load.) I'm not sure than Eo /j/ has two allophones. It may be that the letter <j> represents two phonemes, /j/ and non-syllabic /i/. (The Plena analiza gramatiko says that if we're interested in symmetry, the latter should be written <ĭ>.) Look cross-linguistically, and you'll find that /j/ is much more common in the world's languages than /w/. Also, falling diphthongs like /aĭ/ and /aŭ/ are much more common than rising diphthongs like /ĭa/ and /ŭa/. The "symmetry breaking" in Eo merely reflects the reality on the ground. kwami 22:14, 8 January 2006 (UTC)

I am contemplating your arguments. I have realized that the conclusion one comes to depends on the model he is using. Your quantitative information is helpful.

Below the logo I see the navigation box, then the search box, and then the toolbox box with 7 entries in it: What links here, Related changes, User contributions, Upload files, Special pages, Printable version, and Permanent link. I can't find "e-mail this user" anywhere. I have just discovered the 4 tildes. H. Neal Parker 02:05, 10 January 2006 (UTC)

That's okay, I didn't have my email switched on for the longest time because I didn't know how. (It's in your preferences.) The email user link only appears on user pages. kwami 07:44, 10 January 2006 (UTC)

-- Kwami said that the Japanese, among many others, have great difficulty distinguishing /w/ from /v/. I think the Japanese have difficulty distinguishing /v/ from /b/ (not from /w/!). At least, when they write foreign words in Katakana, V becomes B. A similar phenomenon occurs in Spanish: the phoneme /w/ is clearly defined, but the distinction between /b/, /v/ and intermediate phonemes is blurred. I think /w/ is much closer to /u/ than to /v/ (I agree with H. Neal Parker that Chihuahua should retain its "U"s in Esperanto), altho it might be possible to find a language where even /k/ and /t/ are allophones, so it's no surprise that many parts of the world think /w/ and /v/ sound similar. Every language has its own ideas about what variation in pronunciation is permissible... -- Yuu en 07:50, 3 June 2007 (UTC)

Yes, the Japanese and Spanish (and Chinese, and many others) pronounce v as [b]. Others, like the Germans and Russians, pronounce w as [v], or as [ʋ], which is between [w] and [v]. Zamenhof decided that Espo should only have a two-way distinction, as in all of those languages, not a three-way distinction as in English. Unfortunately, most Japanese and Spanish speakers eliminate the [w] they can pronounce, and then fail to master the [v]-[b] distinction that is foreign to them. Such people have more success when they pronounce v as closer to w (that is, further from b), for then the difference with b is clear. Plus they no longer feel a need to introduce their [w] into Espo in order to make it "authentic". (I mean, if Washington can be spelled with a v, why not Chihuahua or Wakayama?) On paper, people insist pronouncing v as anything but [v] is wrong, but it actually works quite well. I take care to pronounce v as [ʋ], and people from various countries have complimented me on my "authentic" pronunciation. (It also makes it impossible for them to guess my nationality!) kwami 07:12, 17 September 2007 (UTC)

Sometimes these pedantic phonology discussions go in such tight circles they get detached from the real world. Did Zamenhof actually say ŭ may never occur before a vowel, or are linguists reading this into the language? In any case, most Esperantists have never heard of these rules, so they de facto don't exist for them. People spell Guatemala Gvatemalo on analogy with kvar, because they see that Esperanto traditionally prefers kv over kŭ. But since Chiuaua has no official spelling, and nobody since Zamenhov's death can claim official authority on how to spell new place names (or what Esperanto's phonology is, for that matter), it really comes down to how the people who most frequently say Ĉivavo/Ĉiǔaǔo choose to spell it, and whether one particular spelling later catches on among most of the Esperantistaro. Such as is my own home state of Washington/Vaŝintono/Vaŝingtono/Vaŝintonio. I think most people prefer spelling Vaŝintono but they pronounce it alternatingly with /n/ or /ŋ/. Of course, those who have never heard the English pronunciation will pronounce it /n/, and maybe one day their influence will cause anglophone Esperantists to universally pronounce it that way too, but that'll be a while in coming. In the meantime, there's the disagreement over whether Koreo is a place or a people, whether the capital of Brazil is Brazilio, and whether the capital of Mexico is Meksiko or Meksikio or Meksikurbo, to say nothing of Mexico State. This is a non-phonological version of the same issue. Sluggoster (talk) 03:15, 3 July 2009 (UTC)

Likewise, the jida issue seems to be overblown. So "Yiddish" was assimilated in its most natural way. This is supposed to be rejected just because somebody decided that since there are no other ji- words in the language, there can't be any? That's like how English words supposedly can't start with /ŋ/ or /ʒ/ -- or at least they couldn't until Nguyen became a common name and people wanted to pronounce Georges and du jour correctly. Sluggoster (talk) 03:15, 3 July 2009 (UTC)

Zamenhof was not a linguist, and hardly said anything about such things. Generally we have to go on what he did, not what he said.

The problem with a consonant /w/, and the reason Z didn't include it in Eo, is that many languages do not distinguish it from /v/, including Z's two native tongues. Now Idist types want to "fix" Eo by adding /w/ to it, but in practice it will be (mis)pronounced by many. IMO, best to write the way we speak, which is with /v/.

Nguyen? Seriously? What percentage of the American population do you think can pronounce that? The most difficult thing for the actors of Avatar was getting that ng at the beginnings of words--they had an easier time with ejectives. — kwami (talk) 01:21, 25 June 2010 (UTC)

To Yuu-en:

it might be possible to find a language where even /k/ and /t/ are allophones

There is a language like that indeed, Hawaiian. It's even better because there are many more (but less common) possible allophones of the Hawaiian /k ~ t/ phoneme. --Florian Blaschke (talk) 21:43, 17 June 2011 (UTC)

Pf affricate

Would this count as a marginal loan phoneme considering it exist in some loanwords...?Cameron Nedland 15:03, 8 February 2007 (UTC)

More likely a consonant cluster. Covered under phonotactics. kwami 06:49, 17 September 2007 (UTC)

Sources

The article mentions that r, v, ĉ, ŝ, ĝ and ĵ can also be pronounced as [ɾ], [ʋ], [tʂ], [ʂ], [dʐ] and [ʐ] respectively, instead of their standard pronounciations as [r], [v], [tʃ], [ʃ], [dʒ] and [ʒ] respectively. I can't find a source that supports this claim. The most reliable source about Esperanto grammar is the Plena Manlibro de Esperanta Gramatiko, and it mentions only the standard pronounciations, not the pronounciations as [ɾ], [ʋ], [tʂ], [ʂ], [dʐ] and [ʐ]. If no one provides a source that supports the current claims, I will remove them, so that the article represents the phonetic inventory in the same way as does the Plena Manlibro de Esperanta Gramatiko. Marcoscramer 20:15, 23 June 2007 (UTC)

Since no one presented any sources, I have now removed these alternative pronounciations. Marcoscramer 14:43, 28 June 2007 (UTC)

They reflected variants based on speakers' native languages. There's no end to that, of course, but considering they were Slavic, I thought them significant. No problem deleting them, though. kwami 06:51, 17 September 2007 (UTC)

Diphthongs

Given the way the diphthongs are presented in the section "Inventory", a reader might think that the diphthongs are seperate phonemes in Esperanto. However, Esperantologists generally agree that the diphthongs are just realisations of phoneme-pairs. Maybe we should make this clear in the article. Marcoscramer 14:45, 28 June 2007 (UTC)

Sources? Is this because of plural j, with the idea that plurals are agglutinative rather than synthetic, as a phonemic diphthong analysis would suggest? kwami 06:48, 17 September 2007 (UTC)

Actually, the question of whether they're phonemes or constrained by phonotactics is academic. kwami 02:05, 2 December 2007 (UTC)

The comment in an English text about 'buoy' 'as written' is seriously misleading. 'As written' would probably sound like Esperanto 'buoj', and in any case written English is seldom a reliable guide to pronunciation. What is especially confusing is that 'buoy' is pronounced as Esperanto 'buj' in American English, but as Esperanto 'boj' in British English - something few English-speakers seem to be aware of. Perhaps a more useful example for the pronunciation of 'uj' would be 'phooey', pronounced as Esperanto 'fuj' in all varieties of English. Or else 'screwy' (more or less Esperanto 'skruj')62.194.211.96 (talk) 21:18, 21 September 2010 (UTC)

Yeah, that's better. — kwami (talk) 02:08, 22 September 2010 (UTC)

Non-Esperanto sounds

How are non-Esperanto sounds like [θ] and [ð] represented in Esperanto? Are Esperantized letters like t̂ and d̂ used? Or the Icelandic letters þ and ð? Or the Sami letters ŧ and đ? Or English digraphs th and dh? Or modified digraphs like tħ and dħ (with a bar through the h)? Did Zamenhof deal with this explicitly? -- Evertype·✆ 12:02, 14 May 2012 (UTC)

They aren't. The original orthography can be used, of course, or IPA, but there is no Esperanto phonetic alphabet. I suppose individual authors might come up with ad hoc systems, but the only addition to the Eo alphabet I've seen regularly is a dot subdiacritic for irregular stress in proper names, and that's not universal. (Final stress is commonly indicated by an apostrophe, by analogy to poetic elision, which leaves final stress.) — kwami (talk) 22:07, 14 May 2012 (UTC)

English approximation

Kwamikagami reverted my edit regarding my edit to English approximation for pronunciation of Esperanto letters. I wonder how possibly would it be for a normal reader who is not familiar with phonetic jargon to get an idea of how would phooey be pronounced as one syllable, approximating the Esperanto diphthong, uj !? They won't understand it in either ways, whether it is approximated to something misleading or leaving it with no approximations. Leaving it without approximations is much better because at least it won't confuse anyone with something else [u.i]. The word is informal and not in plain English, so I doubt that using such examples is helpful at all! Also, how can an diphthong, as in boat, in major English dialects be used as an approximation for an Esperanto vowel o !? The o in more is the closest since the Esperanto vowel is the mid vowel, [o̞]. The vowel in more varies between [ɔ] to [o] across English dialects, which makes [o̞] in the middle and much better approximated than approximating it to a diphthong [əʊ̯~oʊ̯]. I also made an edit to the style to be clearer for the foot notes. No point at all of reverting my edit. --Mahmudmasri (talk) 22:04, 2 November 2012 (UTC)

"Phooey" is listed in the OED and MW, so that's hardly a problem. If a reader doesn't know what a syllable is, /ui/ is still better than /udʒ/.

"More" is not a good approximation because it is also a diphthong in many non-rhotic dialects. "War" or "dinosaur" would be a closer fit. — kwami (talk) 22:12, 2 November 2012 (UTC)

Now you reverted another edit which I corrected a style in, closing the italic with two apostrophes '' and you came up with a claim that the ewy part in chewy is pronounced as /jui/. A simple check in any dictionary would tell you that you mispronounce that word, it does not have the yod: Cambridge: chewy and Cambridge: chew. Your reverts are disruptive. I wonder if you really check what was edited before you revert? --Mahmudmasri (talk) 19:06, 4 November 2012 (UTC)

The OED has a yod in one of two pronunciations. It is perhaps old-fashioned in RP, but is found in other dialects. — kwami (talk) 20:31, 4 November 2012 (UTC)

Ma(t)ĉo

Really? It seems to me matĉo is almost as common, if not more common, than maĉo for “[sporting] match”. Some people criticize maĉo as seeming to imply chewing. (Sorry, kids, ludo kaj konkuro just don’t cut it.) Wiki Wikardo 21:19, 23 March 2013 (UTC)

Vowel pronunciation

The use of "bet" for the e vowel sound in esperanto is not universal across english. North American english speakers would pronounce bet as /bɛt/, not /bet/. In addition, many speakers of esperanto pronounce "e" as /ɛ/ in medial positions, but as /e/ in final positions (whether wrongly or correctly). Is there another word that can be used? It may simply be better to use a as in "gate" or "fate". Though these are not pronounced identically across english, and are actually pronounced as /eɪ/, they are better than "bet", in my opinion.

"War" may not be the best word for the o sound. Perhaps "bone" would be better, though this is also usually /oʊ/ in english.

The note regarding the pronunciation of "eŭ" is ridiculous, and nonsensical to many people. Many pronunciation guides use words equivalent to /ɛu̯/ rather than /eu̯/. About the best I've seen approximating the sound is "ayw" in "wayward." — ★Parsa ☞ talk 18:05, 3 July 2013 (UTC)

Esperanto vowels are not diphthongs, so ā and ō are not appropriate. "Wayward" would be a difficult guide to "Eŭropa", but it might be helpful. "Bone" would work for a lot of people, actually. — kwami (talk) 20:20, 3 July 2013 (UTC)

Doubtful assertions

I found several doubtful assertions on this page.

• /ɲ/: it's true that PAG considers it a phoneme, but I have never seen a textbook saying that there's a nj sound, and PMEG says nothing about nj. So is /ɲ/ really worth mentioning?
• Same thing about /d͡z/: the article seems to imply that it's definitely a phoneme, but PMEG says "Kontraste la malofta sinsekvo DZ (ekz. edzo), kiu estas kvazaŭ voĉa C, estas en Esperanto sinsekvo de du konsonantoj, kaj oni ĉiam povas elparoli D kaj Z aparte". It should at least be mentioned that not everyone agrees with the status of /d͡z/ as a phoneme.
• "It has been argued that the Esperanto sequences kv, gv are also phonemes": the linked page says nothing like this, it just says that some people may have a tendency to pronounce them as kw and gw but that it's a mistake. (Personally, I have never heard this pronunciation.)
• The article seems to imply that everyone agrees that there are six diphthongs. However, this article analyses aj, ej, etc. as combinations of a vowel + the consonant j. It cites Wennergren which agrees with this view, as well as Wells which even considers aŭ and eŭ as vowel + w.
• "The Esperanto alphabet is nearly phonemic": why "nearly"?

Mutichou (talk) 11:32, 22 December 2014 (UTC)

Italian model

The introduction of the article has the following sentence, which has had a "citation needed" label since May 2013: "Given that the comparison languages were not completely identical, he later advised that the pronunciation of Italian could be considered a model for Esperanto." In February 2014 I deleted the sentence with the following comment: "remove unreferenced claim: I tried to find a reference for it, and also asked my colleagues in the pronunciation section of the Esperanto Academy, but no reference could be found; hence the claim seems very dubious." Three days later, Kwamikagami re-entered the sentence with the comment: "can't find ref offhand, but still relevant". I suggest that if no reference for this claim is found within the next few days, this dubious claim should get finally deleted from the article. Marcos (talk) 09:50, 19 August 2016 (UTC)

As no reference has been provided, I have now deleted this sentence. Marcos (talk) 08:53, 2 September 2016 (UTC)

I found a reference which *seems* to hint that Esperanto was indeed modelled after Italian pronunciation. It is from: J. Cresswell and J. Hartley (revised by J. H. Sullivan) Esperanto - A complete course for beginners, 1987 (first edition:1957). He says: a good pronunciation of Esperanto is quite easy to acquire, and (b) it is one of the most beautiful-sounding languages on earth. If this seems a rash claim, consider what is the most beautiful language. Tastes differ, of course, but if a vote were taken, perhaps Italian might be the winner: and the sound of Esperanto very closely resembles that of Italian. This, of course, does not prove that Zamenhof *purposely* designed Esperanto to sound like Italian but it's a suggestion that such consideration may have been in his mind. — Preceding unsigned comment added by L0rents (talk • contribs) 19:12, 8 December 2017 (UTC)

"Slavic origins" - irrelevant?

"This inventory is rather similar to that of Polish, but is especially close to Belarusian"

Is it really? Most Esperanto sounds are found in most languages, especially Indo-European. The exceptions are:

• /t͡s/, which is indeed found in most Slavic languages, but also in German, Italian and Romanian
• /x/, beside Slavic languages found in Spanish and German

And other sounds are not even the same as in Belariusian and Polish: coronal consonants are alveolar, not dental; it's /ʃ ʒ t͡ʃ d͡ʒ/ (sounds found in English, French, Spanish, German, Czech etc.) rather than /ʂ ʐ ʈ͡ʂ ɖ͡ʐ/; mid vowels are /e o/, not /ɛ ɔ/; there is /x/, but there's also /h/ not found in Slavic languages (and for Slavic languages speakers it's probably more difficult to differentiate between the two than for an English speaker).

What I'm trying to say is that I don't see any reason why this inventory is "rather similar" to that of Polish or Belarusian. If I'd seen this inventory without reading about "Slavic origins" I would've thought that it's a generic Indo-European phonology.

Basically, if you take Belarusian phonology, and take away "minor differences" given in the article, i.e. 1) palatalization, 2) [ɣ], 3) absence of /h/, you get the following: /m n p b t d k g t͡s d͡z ʈ͡ʂ ɖ͡ʐ f v s z ʂ ʐ x w l j r/, which is an inventory (ʂ ~ ʃ, ʐ ~ ʒ, x ~ h) found basically in any Indo-European language (which is the point, as it's an auxiliary language).

Therefore, I think that that section should be removed, as it's really irrelevant and not necessarily true, unless there are sources that'd directly say Zamenhof was somehow inspired by Belarusian phonology. — Preceding unsigned comment added by 188.146.129.46 (talk) 16:48, 15 September 2016 (UTC)

Few non-Slavic Indo-European European languages distinguish between the full set of /ĉ ĝ ĵ ŝ/. Portuguese doesn't; Spanish doesn't; French doesn't; Dutch doesn't; German doesn't; Danish doesn't; Norwegian doesn't; Swedish doesn't; Icelandic doesn't; Welsh doesn't; Irish doesn't; Scottish Gaelic doesn't; Greek doesn't; Latin didn't; and PIE probably didn't either. Thus you are clearly mistaken in your claim that the consonants of Esperanto are "found basically in any Indo-European language", and I would suggest that you not make any edits based on that idea. — Preceding unsigned comment added by 80.216.7.157 (talk) 23:53, 15 September 2016 (UTC)

German does distinguish that set of sounds. The following four German words contain these four sounds: "Deutsch", "Dschungel", "Journalismus", "Schule". I think there is some point to the comments of 188.146.129.46. Unless the claim about Polish and Belarusian can be sourced, it should be removed. Marcos (talk) 17:18, 13 December 2016 (UTC)

I fully agree with the original comment: the sentence "This inventory is rather similar to that of Polish, but is especially close to Belarusian" does not really stand up to close scrutiny. In particular, between Polish and Belorussian, Esperanto is closer to Polish than to Belorussian (although Esperanto phonotactics is much simpler than Polish one). I've drawn up a spreadsheet comparing the sound inventory of Esperanto to the ones of Italian, Polish, English, French, Russian, Belarusian, Ukrainian, Czech, Croatian and Mandarin Chinese; I've listed these languages in descending order of my personal level of knowledge of their phonology (I'm an Italian native, don't know Chinese at all). There's a bit of ambiguity in such a comparison (e.g., /ʒ/ is officially absent in Italian but in practice it's used regularly in, e.g., foreign words of French origin, so it's completely unproblematic. Similarly, Russian words never use /d͡ʒ/ but the sound is used in foreign names etc.). By most reasonable measures Polish comes out victor (=most Esperanto sounds are present in the language) by a small measure, as the only feature of Esperanto not present in Polish is opposition /x/ vs /h/ (which is anyway obsolete in Esperanto). At second place I have an ex-aequo of Italian, Belarussian, Russian, Ukrainian, Croatian and English. Poor Chinese is very far down the list (Polish gets a score of 97%, the languages at second place 91%, Czech 88%, French 86%, Chinese 50%). In actual practice all Belorussians and a good deal of Ukrainians are Russian bi-lingual so adding the inventories Ukrainian+Russian would come second after Polish (95%) and Belarussian+Russian would come third (93%). I could throw in more Romance and Germanic languages and I believe they would score in the 80-90% range. In any case in terms of phonotactics Esperanto resembles much more Italian and Slavic languages and, all in all, I don't think it makes sense to say it is particularly close to Belarussian. L0rents (talk) 19:00, 8 December 2017 (UTC)

You don't show how you got those %ages. Apart from palatalization, which Z removed from his project before publication, Eo has the vowel system as Belarusian, not Polish. (Though missing the Belarusian diphthongs oŭ & iŭ.) The consonants are also closer; Polish has [w] which Esperanto doesn't have, whereas in Belarusian orthographic g is usually (but not always) [ɣ]. Eo does have the Polish-Russian pronunciation of g, as well as /h/ which doesn't occur in any of them. Actually, apart (as always) from palatalization, Eo has exactly the consonants of E.Yiddish. The vowels aren't as close, though that presumably depends on dialect, as Yiddish has a schwa and doesn't have all the diphthongs. Still, closer than Polish is. — kwami (talk) 23:27, 1 August 2021 (UTC)

Cleanup lang tag revert

@Kwamikagami: Greetings! Regarding this revert...I thought it was pretty clear from the HTML comment what to do? An automated spell check has detected in this article the following words which are not English and which are not inside a template:

oŭ, ŝl, ŝr, ŝp, ŝpr, ŝt, ŝtr, ŝm, ŝn, ŝv, uŭ, ogdek, egzisti, egzemple, supteni, lonktempe, glafsonoro, apsolute, optuza

Each instance of these words should be enclosed in {{lang}}, or {{respell}} or {{IPA}} or whatever template is most appropriate to the circumstance. Does that make sense? -- Beland (talk) 20:42, 1 August 2021 (UTC)

@Beland: Hi Beland.

The two-letter sequences are just that, letter sequences, not words. Like saying that English, Russian and French can all have words that begin with str-, the "str" isn't any particular language.

The longer sequences, like 'ogdek', are also difficult to classify. They aren't Esperanto per se, but pronunciation respellings. Like respelling 'New York' as "Noo Yawk", I don't know how we'd classify them with the language template. How would we instruct a screen-reader that 'ogdek' is how 'okdek' is pronounced in Esperanto? Especially as the screen-reader will likely pronounce them differently? — kwami (talk) 21:01, 1 August 2021 (UTC)

@Kwamikagami: According to the documentation on Template:Respell, it seems the right thing to do for the pronunciation respellings is to rewrite them in IPA notation (and use an IPA template), which would also cause screen readers to render them properly. Otherwise, it's unclear which language's pronunciation rules that readers should apply, especially since it's clearly not the Esperanto rules. The short strings like "ŝtr" appear to be from Esperanto orthography? If that's the case, they can be tagged as Esperanto with {{lang}}. In linguistic notation, they would conventionally be marked with angle brackets (in wikitext, {{angbr}}) to indicate they are orthographic. -- Beland (talk) 22:14, 1 August 2021 (UTC)

Angle brackets are typically used where italics would be ambiguous, which isn't the case here. And the documentation for Respell is specifically for English, where respelling is such a mess. It doesn't apply to languages where unambiguous respelling is practical.

I tagged the respelling as Esperanto. Not sure what a screen-reader will make of it, but at least it won't generate errors. — kwami (talk) 22:21, 1 August 2021 (UTC)

Well, that will generate future errors, if it's not a valid spelling in Esperanto, once we start spell-checking the non-English text in Wikipedia. I wasn't suggesting using ; I was suggesting using {{IPA}} or {{IPA-eo}} or {{IPAlink}} etc. -- Beland (talk) 01:57, 3 August 2021 (UTC)

It's not IPA, so that wouldn't work.

I doubt we'll ever have an Esperanto spell-checker to worry about. It would be impractical to list all the possible words. No dictionary does, so I don't know where we'd even start.

Could we tag it lang-xx or something? Or would that cause other problems? — kwami (talk) 08:23, 3 August 2021 (UTC)

Yes; why don't we simply convert it to IPA? That notation was designed to unambiguously express how things are pronounced.

No dictionary currently lists all the possible English words either, and yet we have English spell checkers. I'd start the Esperanto spell checker the same way I started the English Wikipedia spell checker - look up each word to see if it appears either in Wikitionary or an article titles in the Esperanto Wikipedia. Any word not found either needs to be added to the dictionary or dealt with as an exception. -- Beland (talk) 17:04, 3 August 2021 (UTC)

Why use IPA? It's a needless complication. Esperanto orthography is unambiguous, and our sources use it. I'm a big fan of the IPA, personally, but there's no need for it here. Screen-readers should accommodate the material they're reading.

As for a spell-checker, we can cross that bridge when we come to it. But Esperanto is fundamentally different from English in this regard. English dictionaries at least attempt to list all words, Esperanto dictionaries don't. That's because we don't just make up words as we go along in English, at least not in formal text, but in Esperanto doing so is normal. The reader usually won't even notice, but a spell-checker would be stumped. Unless maybe you don't count word spaces? Any sequence of joined recognized morphemes would be accepted by the checker? In that case, these respellings could be added as exceptions, or tagged with a 'do not spell-check' template. — kwami (talk) 21:00, 3 August 2021 (UTC)

What's included in the dictionary is a function of Wiktionary policy; if there are situations where pieces of words are in the dictionary and there are clear rules about how to combine them into valid words, certainly the spell checker can handle that and needs to handle that anyway for natural agglutinative languages. There is already code to handle compound words in English, and systematic words that Wiktionary excludes, like higher numbers and names of chemicals where we do make up words on the fly in formal settings.

In what notation do you write these words, if they are not IPA and they are not Esperanto? Where would readers go to decode that notation? As an English speaker reading the English Wikipedia, it is unclear if for example "egzisti" is pronounced like egg-zih-st-ee or egg-zee-st-eye, or what, and it's unclear which syllable the emphasis is on. Using IPA notation clarifies that. If you want to keep it freestyle and leave readers guessing, or if you want to keep this and also add IPA for clarity, the catchall template for excluding something from spell check is {{not a typo}}. -- Beland (talk) 02:49, 4 August 2021 (UTC)

It's not "leaving the readers guessing". It's transparent how these are pronounced. IPA would make things more difficult for our readers, and there's no point.

If you ever work out an Eo spell-checker, we can tag them as not-a-typo. — kwami (talk) 02:59, 4 August 2021 (UTC)

It may be transparent to you, but it's certainly not transparent to me and other non-Esperanto speakers. -- Beland (talk) 16:27, 4 August 2021 (UTC)

What is oĝalan?

The article seems to suggest that there are words in Esperanto which are stressed on the last syllable and not the result of elision, and cites the example oĝalan. This is also found on the Esperanto article "Akcento"… and basically only in these two sources. A Quoran told me he looked it up in «il vocabolario più strutturato in Esperanto con circa 25.000 lemmi» (the most structured Esperanto monolingual dictionary with about 25.000 lemmas», and didn't find it. So what is it? Where is it from? Are there other words with stress on the last syllable that aren't the result of elision? What is the situation of anstat = anstataŭ, which I've found on that Esperanto article? Is it an alternate pronunciation happening in fast speech that isn't standard, or perhaps a slang variant? MGorrone (talk) 21:57, 3 February 2022 (UTC)

These would just be proper names where the author wishes to retain the original accent. "Oĝalan" is presumably someone's name -- looks like I added that and failed to spell it out -- maybe I just took it from the WP-eo article. We should probably replace with a confirmable example. Any Esperantized name has the stress on the 2nd-last vowel, with the exception of poetic elision of -o. "Anstat", if it were ever used, would be stressed on the first syllable. — kwami (talk) 00:50, 6 February 2022 (UTC)