Template talk:IPA hover

A couple of points

• Consider the following. How about the former as opposed to the latter?

• We're likely to run into dialect problems. For example, I pronounce roses & Rosa's the same. Care must be taken to keep in mind that the symbol chosen to transcribe a pronunciation might not be based soly on that pronunciation.

Jimp 17:22, 27 January 2007 (UTC)

[edit] Content for English version

For consonants we might use the following from the IPA for English article:

For vowels we could start from content from wikipedia talk:Manual of Style (pronunciation) (after removal of duplicate symbols):

−Woodstone 20:10, 25 January 2007 (UTC)

This is a good list. But CLOTH does not necessarily need to be included (the symbol under CLOTH says only: depending on your accent, use either THOUGHT or LOT). If CLOTH is included, we should also include BATH (depending on your accent, use either TRAP or PALM). The symbol for BATH might be "a" print a followed by the long mark. The list also includes only uneven coverage of vowels before "r" (Sirius and merry but not marry, sorry, hurry, NURSE, LETTER, SQUARE, NEAR, CURE, or FORCE/NORTH). See below for my 'bare-bones' proposal, which would deliberately include only the bare minimum of syllables.--Gheuf 21:29, 30 January 2007 (UTC)

[edit] Proposal to do it ALL

I have tried some work on using subtemplates and {{#ifexist:IPA hover/{{{1}}}}} to keep the gigantic case statement out of the code. Here's an example: {{IPA hover/2.1|ʒ}}: ʒ This allows the possibility of multiple options per character, and default behavior. I suggest that instead of one helptext per character we have the following possibilities:

• The explanation for a loose English transcription ("r as in run"). This might go at {{IPA hover//r}}. This could be the default if it exists.
• A more technically correct explanation ("trilled r as in Spanish rubio"). This might go at {{IPA hover/r}} and could be the default if the above is missing.
• The name of the technical article ("alveolar trill"). This might go at {{IPA hover/art r}}, and could be the default if both of the above are missing.
• Just a default text, possibly linking to the IPA article, over the given character, if all of the above are missing. {{IPA hover/default}}.

There remains the question of how/whether to do linking to the technical articles. There are clearly cases where such linking is undesirable (primarily, when a link already exists, for instance to an audio file). This could be done internally with parameters (to avoid duplicating the whole set of subtemplates) but externally with two templates, IPA hover and IPA linked hover. Finally, this proposal would create a large number of small subtemplates, hard to deal with on a watchlist; I tentatively propose that a master table be created at {{IPA hover/master table}}, and a simple bot be created to propagate the changes to the individual templates. I could help write such a bot (which would be really dumb, complaining to its talk page whenever anyone broke format or edited the subtemplates directly) but I could not host it. --Homunq 14:21, 29 January 2007 (UTC)

This could work. The big case statement is a limitation. Using lots of subtemplates would still prevent the capitalization problem from previous discussions, and allow for a near-infinite amount of descriptions. Links would have to be in the form: [[Mac OS X|<span title="This is a hovered title">Mac OS X</span>]]. I'll work on a template set that will cover the possibilities. —W. Flake (talk) 22:10, 29 January 2007 (UTC)

You know, there's actually no technical reason why this template couldn't put the entire, comma separated, phonetical help for a given word in a single popup text. That would increase accessibility by avoiding the persnickety micro-movements of the mouse. The question is, is lots of popup text like that readable? Old browsers put it down at the bottom of the window, where space is limited. What does your browser do? Put your mouse here. --Homunq 13:46, 30 January 2007 (UTC)

I like this idea (see my proposal below). It would only be necessary for the pop-up to show whatever "difficult" the word may contain. We do not need "/f/ as in "fast"."--Gheuf 04:34, 1 February 2007 (UTC)

[edit] Simplified IPA Pronunciation Guide for English

[This is reposted from Template_talk:Ʒ.]

I think it is awonderful idea to have IPA pronunciations link to a pronunciation key. Most "real" dictionaries have, at the bottom of EVERY PAGE, a short table showing how every symbol is pronounced: it is an abbreviated version of the "Guide to Pronunciation" which is usually put at the beginning of the dictionary. Wikipedia doesn't have pages: but why not use hypertext to cause the pronunciation guide to pop up?

The problem I see with the current proposal is that it seems that we would have separate page for EVERY SOUND. So to find out, for example, how "xenon" was pronounced, we would have to go to four separate pages: one for z, one for ee, one for n, one for short o..... This is extremely cumbersome.

Why not have, instead of a popup for each sound, one general popup for the whole word which would correspond to the pronunciation chart of most dictionaries. It would have every symbol followed by a keyword, like the "zh as in beige" example given before. The keysound of the keyword could be written in bold (the "ge" of "beige"). Hopefully this could be incorporated into the code for squiggly bracket-squiggly bracket-IPA-close squiggly bracket-close squiggly bracket. It would be enough, I think, at first to make just a pronunciation chart for English, including the normal English sounds as well as the marginal sounds like the voiceless velar fricativea, the nasalized vowels, or the glottal stop. But, in principle, a separate chart might be made for every language, each showing the 'standard transcription' for that language. This would necessitate using different templates, of course, eg IPA_french, IPA_english, etc.

Someone might object that as it is the "international" phonetic alphabet, separate charts should not be needed for separate languages. But this is not really a valid objection. IPA is international in that it provides a standard body of symbols to all the world in which to notate the sounds of speech. But nevertheless it remains true that, as each language has its own sound-system, each language will choose (and has in fact chosen) different elements from the vast set of IPA characters which will be taken as their "standard" transcription.

Of course, the standard transcription varies from dictionary to dictionary: it is a typographical convention chosen by the editors like any other. Since we all are the editors of Wikipedia, we would have to choose which transcription would be standard for us.

Alternatively (and I think this very simple step would go a long way to making IPA easier to read), the template could just like to IPA_chart_for_English: this is already a pronunciation guide of sorts. This would be a great improvement over what is done currently, which is (in many cases) to link to the main IPA page, which is really not what is wanted.

Here is how such a pronunciation table might look. (It would only include unfamiliar symbols, or symbols for which confusion seemed likely.) The sound would be presented in quasi-alphabetical order:

/aɪ/ my /aʊ/ now /ɑ/ father /ɒ/ not /æ/ bad /ɔ/ law /ɔɪ/ boy /dʒ/ judɡe /ð/ these /ə/ about /ɚ/ winner /ɛ/ bed /ɝ/ bird /eɪ/ day /ɡ/ ɡoat /i/ fleece /j/ yes /ŋ/ brinɡ /oʊ/ no /ɹ/ red /ʃ/ sheep /tʃ/ choose /θ/ think /ʊ/ book /u/ blue /ʌ/ run /ʒ/ beiɡe See IPA chart for English

This chart presents the "bare minimum" for notating English pronunciation. Obviously such a chart can be quite compact and useful. It also preserves as many contrasts as possible in order to work across accents. For example "father" and "not" are written with different vowels, even though in GenAm they are pronounced with the same vowel -- speakers who do not make the distinction can still see the symbol, read the keyword, and know how to pronounce the word in their accent. (If they are intrigued by the apparently unmotivated distinction, they need merely follow the link at the end for more info.)

Of course the exact symbols chosen would be debated. In particular, it might be thought that the transcription has a rhotic bias (see the transcriptions for "bird" and "winner"). To a certain extent, this objection would be answered with what I said before: speakers who "drop their r's" will still be able to do so, since the transcription relates to key-words and not to phonetic absolutes. The rhotic bias (like the extra distinctions in vowels) is necessary so as not to give a false impression to someone who will read the key-words with an accent that distinguishes these sounds.--Gheuf 04:04, 30 January 2007 (UTC)

--Gheuf 17:05, 30 January 2007 (UTC)

At a second glance, I seem to have misunderstood the proposed template, which would not link up every IPA sound with its page, but would rather put a pop-up with a keyword for every IPA sound. This is of course much better but still a bit cumbersome. As another person has pointed out on this page, it would be much easier if merely one window popped up for the whole word. That window might contain something like the list I offered above, or else a smaller list with only the "difficult" symbols contained by the word.

I also think we need to clarify whether we want this to be language-specific or IPA-wide. As I wrote above, I think it should be language-specific (starting with English), because, firstly, the symbols are used differently in different languages, and, secondly, a keyword system is nonsensical in a cross-linguistic context. French does not have "f" as in "fast", it has "f" as in "frites". In order to have the system work for all transcriptions regardless of language, we would have to retreat to a fundamentally unhelpful designation such as "voiceless labio-dental fricative."

To see the truth of this, consider that the French /e/ and the German /e:/ are both transcribed with the same symbol, but one is /e/ as in "café", the other is /e:/ as in "Kaffee." This is a real difference. The French vowel sounds rather like the English /e/ as in "bear", the second rather like the English /i/ as in "fear".

--Gheuf 04:39, 1 February 2007 (UTC)

I have made such a template.{{IPAGuide|pə'tu:njə}} gives:

pə'tu:njə

I think this would be better if the pop up could some how be limited to the sounds in the word, and yet be only one window..... Is that possible?--Gheuf 20:29, 5 February 2007 (UTC)

It would require some clever javascript, but it definitely is possible. I think this is the best solution, because it doesn't require users to change the way they enter IPA. I am working on a prototype that should be ready in a couple days. The basic idea is to construct a lookup table for each IPA symbol, and use a cleverly designed regular expression to match all the relevant characters in the span. Then, for each matched IPA character, lookup the relevant help text in the lookup table and construct the help popup by concatenating only the relevant entries from the table.

This method has the side benefit of obviating the problem in Firefox wherein it truncates long tooltips, because in most caes the relevant help text will be fairly short. Nohat 19:51, 9 February 2007 (UTC)

I don't understand. You can concatenate the relevant text together all in one tooltip, without any javascript at all, if you use the "pipe" syntax that this template currently uses, and standard template tricks (look at the variables twice, one for the popup and again for the visible content). If you know how to use javascript to make more sophisticated, non-tooltip popups, a la "popups", that would be great - but even then, we can construct the text to go into the popup with no need to burden the client computer with regular expressions and lookup tables. Homunq 06:15, 11 February 2007 (UTC)

Unfortunately the templ--Gheuf 15:35, 12 February 2007 (UTC)--Gheuf 15:35, 12 February 2007 (UTC)ate technique puts a big burden on editors to put pipes between every single IPA symbol in a transcription, making the wikitext of such articles hard to read, edit, and maintain. The Javscript solution I have devised, which you can sample by copying User:Nohat/monobook.js into your own User:Username/monobook.js requires no changes to any article wikitext, and the IPA text in articles such as List of words of disputed pronunciation have automagically-generated tooltips, and if you look at the Javascript you will see the client overhead to do this is quite marginal.Nohat 01:25, 12 February 2007 (UTC)

The style with pipes allows the editor to indicate combination symbols like aspirated plosives, diphthongs and dʒ. Does the auto-select script support that? −Woodstone 13:24, 12 February 2007 (UTC)

Yes, it handles diphthongs and affricates correctly. It does not handle aspirated plosives because it only works with a (somewhat) standardized phonemic transcription system for English. Nohat 21:49, 12 February 2007 (UTC)

The distinction between using JavaScript and doing without is beyond my ken; but I think any scheme is preferable that does not require alterations to the way IPA is entered. Ideally, I think, you shoud just type the IPA in the normal way, and get the pop-up.

Nohat, I like the way your guide works, but I am afraid the display I suggested is too hard to read. I forget who proposed that the /'s be taken away from the IPA symbols, but I think it might be clearer if they were restored. They identify the text contained in them as a special pronunciation symbol even for those not acquainted with IPA -- they are used this way in, for example, the pronunciation guide at the bottom of every page in Merriam-Webster's.

The text might also be clearer if the hyphens were removed, and a comma (or an extra space?) put after every keyword, to separate it from the preceding entry.

Homunq suggested to me that "easy" symbols should be LISTED (but not DESCRIBED) in the text as well. They might serve as a place holder to help people keep track of where they are in the transcription. On the other hand, they take up extra space. We could maybe try it both ways and see which looks clearer.

So far the script seems to know that "oU" is to be treated as one symbol; I would suggested that vowel-plus-r combinations be so treated too. This is tricky, since the vowels have a different keyword depending on whether a vowel or consonant follows the "r". (Aspiration signs are not used in this particular system of transcription, which limits itself to English. See Wikipedia: Pronunciation.)--Gheuf 15:35, 12 February 2007 (UTC)

[edit] IPA/"spelled" pronunciation javascript toggle suggestion

What about implementing something like dictionary.com has done with a toggle between the two different ways to notate pronunciations that they have chosen. e.g. look at http://dictionary.reference.com/browse/pronunciation and notice that you can click a javascript link (Show IPA Pronunciation) that toggles the IPA and "spelled" pronunciation guides. I think this should be implemented in the Mediawiki software somehow. I'm not sure if this is the best place to post this, so feel free to crosspost this suggestion wherever. Thanks. --Rajah (talk) 07:56, 7 April 2008 (UTC)