Wednesday, 13 June 2012

casing the IPA

Michael Everson has created an IPA edition of Lewis Carroll’s Alice in Wonderland and has kindly let me see a copy.

It presents the entire text of Alice transcribed into IPA. (This complements his publication of Lewis Carroll works in a variety of lesser-known languages.)

I may have some comments to make on the transcription of particular words and phrases once I have finished reading it. Meanwhile, I thought I would comment straightaway on one striking visual characteristic of the transcription in general, namely Everson’s decision to use not only punctuation and italicization but also capitalization — and capitalization not just for proper names but also, as in ordinary English spelling, to mark the beginning of a sentence and even for the pronoun I.

In the days when the Journal of the IPA was called Le Maître Phonétique and written entirely in transcription, the rule was very clear: phonetic symbols are always lower-case. To indicate that a word was a proper name, the orthographic capitalization was sometimes replaced by a preposed asterisk, a usage specified in the old Principles of the IPA booklet (1949).

* may be prefixed when it is desired to show that a word is a proper name. (p. 17)
On the other hand the same booklet concludes with a brief note on Phonetic Spelling, which does use capitalization not only for proper names but also to mark the beginning of a sentence.
ƌi jutiliti ov ði Intərnaʃonal Fonetik Alfabet iz not konfaind tu ði aktjual reprezenteiʃon ov pronənsieiʃon. ƌi alphabet mei oolsou bi emploid (wið sjutabl konvenʃonz) az a ɟenəral system ov speliŋ for kərent pərposez and az a translitereiʃon system. (p. 51-52)

The decision to distinguish upper and lower case means that Everson also had to decide on a suitable upper-case shape for each IPA symbol required in the transcription. Since he is a typographer and font designer, this was right up his street. Here are his solutions.

55 comments:

  1. Now if he got the weak forms right…

    ReplyDelete
    Replies
    1. Yes -- I found reading the excerpts in my head very uncomfortable for this reason.

      Delete
    2. This comment has been removed by the author.

      Delete
  2. Another potential source of confusion would be the colon. In the sentence where he introduces the repertoire of italic and capital forms, he finishes in a colon. I presume that this is not a set of length marks, although it looks the same as the length marks used in the rest of the passage.

    ReplyDelete
    Replies
    1. The vowel length mark (ː) are formed with triangles while the colon (:) is formed with circles (in serif fonts) or squares (in sans-serif fonts).

      Delete
  3. A wonderful idea. I adore that book. And it would be absolutely marvellous if Mr Wells could contribute by making suggestions about the intonation marking.
    I would very gladly pay for something like that.

    ReplyDelete
  4. He may as well get rid of the stress marks.

    ReplyDelete
    Replies
    1. Or at least, I'd be interested to know on what basis the I in IPA gets one (ˈAɪ) as does eyes (ˈaɪz) but not the pronoun I ().

      Delete
    2. It seemed to me that the spelt-out letters and the second part of the phrase "his eyes" received more sentence stress than the first person pronoun.

      Delete
  5. I find captial sigma for ʃ particularly peculiar, as well as a large "g" instead of G.

    ReplyDelete
    Replies
    1. In Eugen Dieth (1938): Schwyzertütschi Dialäktschrift, there was a suggestion for a different design of a capital letter for ʃ. I immediately liked Dieth's design better than the Africanist Ʃ. Though it never caught on, I guess you could use it as an alternative glyph design of the same character. I am going to get copies of Dieth's design and post them. He included antiqua and fraktur variants both in printing style and in handwriting style.

      Delete
    2. I would be interested to see them, but I am really not put off by the Africanist Ʃ.

      Delete
    3. There is the scan: Vorschlag für ein sch-Zeichen (Dieth 1938, p. 28).

      It is the phrase Si müeßti d Täʃʃen uf Ʃtans ʃicke, Swiss German 'she would have to send the bag to Stans', in 11 (!) different styles of Latin. The samples 8–11 really read Si müeßti ſi Täʃʃen uf Ʃtans ʃicke which is ungrammatical but doubtlessly serves to illustrate the difference between ſ and ʃ.

      Eugen Dieth's design of Ʃ looks like an inverted (turned and mirrored) digit 2. Now that I see it again, I am not so sure any more whether I like that design better than the Africanist Ʃ that looks like a Greek Sigma Σ. Dieth's design mainly is a curious fruit of the anti-German nationalism of 1930s Switzerland. I doubt that it has ever been employed, in spite of the nice drawings which even include a typewriter keyboard layout: Schreibmaschinentastatur für die Dialektschrift (Dieth 1938, p. 28).

      The reference is: Eugen Dieth (1938): Schwyzertütschi Dialäktschrift. Leitfaden einer einheitlichen Schreibweise für alle Dialekte. Nach den Beschlüssen der Schriftkommission der Neuen Helvetischen Gesellschaft (Gruppe Zürich). Zürich: Orell Füssli.

      Delete
    4. Aren't there dialects that have si for the more common sini = 'his' (pl.) or even 'his/her/their' (referring to the subject's)?

      A lot of Dieth's proposals are still in use, as far as I know, though many authors use the German convention of sch and plain s in front of p, t for ʃ. (The latter makes sense Swiss dialects won't tolerate sp, st even in loans, as opposed to Standard German.)

      Delete
    5. Dieth's proposal is certainly often used, but only in its milder "Übergangsbestimmungen" that (according to Dieth, only temporarily) use sch, not ʃ.

      You're right about the original possessive adjective form sy. I tend to forget about that form because I am among those who have generalized the independent possessive pronoun syni to be used as possessive adjective. In Dieth's sentence, however, the subject is feminine so the possessive adjective should be iri.

      Delete
    6. I understood "she" was to send "his" bags.

      Delete
    7. I stand corrected, of course you're right.

      Delete
  6. Ed: The colon and length mark are different characters; the former has round dots, and the latter has two triangles. These are distinguished in the font.

    Kilian: Many IPA characters have acquired casing forms in African orthographies, and Unicode reflects this. Ʃ and ʃ are a case-pair. Unicode distinguishes regular g from IPA ɡ; the former is paired G/g, and the latter will be paired /ɡ, though the former (U+A7AC) is under ballot and you will not find it in your fonts. Since the IPA's script-ɑ and script-ɡ are similar, and since the capital form of the former is /ɑ, I used the same logic for the capital script-ɡ.

    ReplyDelete
    Replies
    1. (Not sure what happened there; that was supposed to be a response to Beatrice Portinari's comment above.)

      Delete
    2. I think that I can tell the difference if I strain my eyes. I hope that the distinction between colons and length marks is clearer in the actual book.

      Delete
    3. I agree that while there is a difference, it might not be easy to keep them apart for non-pros. I suppose it would increase printing costs to use different colours for punctuation marks.

      Delete
    4. People distinguish . and , easily enough. : and ː can likewise be distinguished without too much trouble.

      In my Hawaiian Alice I did make the ʻokina a bit larger than the quotation marks. That practice is evidently fairly common in Hawaiʻi.

      Delete
    5. Asserting that something is so doesn't make it so. No, sorry, the : and ː (colon and length mark) are NOT distinguishable in the passage above at the top of the post. I'm with Ed on that.

      Font matters here. In the bold font in the posts on this blog, the length marks don't look like colons, but I did have to learn to distinguish them from ɪ (small capital i).

      Delete
    6. This comment has been removed by the author.

      Delete
    7. Dear Mr. Everson,

      Where can one get this book (or at least a sample thereof)? It doesn't appear to be on http://www.evertype.com/carrolliana.html ...

      Delete
    8. Alexander Kayumov

      John Wells has stopped posting, so many people no longer visit this blog. To be sure of contacting Michael Everson
      1. Go to this home page
      2. Search for the details of the Hawaiian Alice.
      3. Follow the link (top left hand corner) to About Michael Everson
      4. At the bottom of the page, click on (HTML) Michael Everson. This should set up an email to Michael.

      Delete
    9. David Crosbie,

      Thank you very much for your advice, I have written to Mr. Everson directly!

      You see, my wife teaches English as a foreign language (to Russian students) - which, naturally, involves teaching phonetic symbols, which comes rather hard to most students. When she saw the picture that appears at the top of this blog entry, she became very excited and said that this could be an ideal tool to teach the phonetic alphabet to her students! With dictionaries, students have the original word alongside its transcription and so can sort of guess what the transcription means, so one needs a text that is completely in a phonetic alphabet where students would have no alternative to trying to decipher the phonetic symbols themselves, thereby learning them. Now, I have had a very hard time finding phonetic texts (perhaps you could direct me to some?) - which is to say I have found almost none so far... I believe my wife is right in that this book would be ideal for this purpose: the underlying text is known to most students and the language is rather accessible (I mean, at least compared to extracts from Le Maître Phonétique :) )...

      Delete
    10. Alexander Kayumov

      Phonetic texts have become unpopular with book publishers, but a surprising number of books have been uploaded the internet and can be downloaded for free. This could men that you and your wife can delete or paste white space over any standard orthography, leaving only the transcription.

      There are problems. Different works use different symbols for British pronunciation — let alone American. And some texts are pretty dull. My favourites are English Pronunciation Illustrated by John Trim and Peter Kneebone, and anything by Colin Mortimer such as Elements of pronunciation. If you google those titles you'll find several sources, which in turn will offer other titles for you to consider. The John Trim book is has funny captions — usually a word or phrase — to attractive pictures. If you type John Trim in the search box at the top left hand corner of this blog, you'll find a few threads where the book is discussed. The Colin Mortimer book is suitable for more advanced students, with (mostly) dialogues giving practice in stress, linking, short forms etc.

      Delete
  7. The Turkish pairing I used for İ/i and I/ɪ may be revised, as a “CAPITAL SMALL CAPITAL I” is used in Unifon, and if that character is added to Unicode, it will be the case-partner for ɪ. It would, in this font look like a T with a bottom horizontal mirroring the top.

    ReplyDelete
    Replies
    1. If it was used, the capital for i would be I.

      Delete
    2. I’m afraid this isn’t a good choice, Brian, I is capital ɪ. Only capitalizing i as İ or something else other than I is unambiguous.

      Delete
    3. This comment has been removed by the author.

      Delete
    4. He said that another character (apparently from here: http://std.dkuug.dk/jtc1/sc2/wg2/docs/n4262.pdf) will be used if it is accepted into Unicode.
      Oh, and the "Latin theta" could also be used. (It looks like [ϴ 03F4 GREEK CAPITAL THETA SYMBOL] and [θ 03B8 GREEK SMALL LETTER THETA], only using a Latin form.)

      Delete
  8. Michael, didn't you consider to go for a descending form for the capital EZH, which has parallels in blackletter (where the lowercase EZH comes from)?

    Even if the capital EZH comes from a practical orthography (as your Sigma form for CAPITAL ESH), this would clearly qualify as allographic variation.

    Sz

    ReplyDelete
    Replies
    1. Ezh does not come from blackletter. It derives from Pitman's Phonotypic Alphabet. It was originally a reversed Sigma (in Alphabets No. 8 and 9, and in the 1847 Phonotypic Alphabet).

      Of the capitals with descenders in that body font, Eng and Script-G alone have them. To me a capital Ezh with descender looks like it is taking a very big yawn.

      Delete
  9. Yeah, I'd lose the stress marks too. (He seems to use it in all polysyllabic words and no monosyllabic ones, no matter what.) This works best if two different sets of symbols are consistently used for strong vowels and for weak vowels, so that e.g. dɪskəs is discus and dᵻskʌs is discuss.

    ReplyDelete
  10. What's wrong with the stress marks?

    ReplyDelete
  11. A few things that struck me:

    Weak forms absent where I would expect them (e.g. "was", "are", "and", "of"). This makes it awkward -- almost painful -- to read rapidly.

    Inconsistent use of linking R (absent in "hatter opened", present in "familiar italic").

    Inconsistent use of stress marks in monosyllables (present in "eyes"; absent in "wide").

    First vowel of "phonetic" seems odd for RP.

    Two primary stress marks in "italicization".

    Secondary stress in "capitalization" in an unexpected location.

    "The" before a vowel has the FLEECE vowel -- surely it should be the happY vowel.

    Syllabic /n/ (from underlying /ən/) is given as /n/ in "raven" but as /ən/ in "punctuation".

    ReplyDelete
    Replies
    1. Some forms like "of" will certainly be reduced to [əv].

      Try to imagine that this is not as easy as it seems, hm?

      Delete
    2. I've no doubt that it's difficult -- you have taken on a project of huge (possibly unprecendented?) scope, and I take my hat off to you.

      If you're not looking for critical feedback, then please disregard my remarks. As to the weak forms, I think the issue is consistency. At the beginning of the first passage we have /tə/, which primes the reader to expect more weak forms where appropriate. So when we later see /wɒz/ it's a bit of a shock (at least for me -- perhaps in other dialects the strong form of "was" is more common than the strong form of "to").

      Delete
  12. Aɪm sərˈpraɪzd tu si ˈMaɪkəl ˈjuzɪŋ ə trænˈskrɪpʃən əv ɑrˈpi ivən wɛn ˈspikɪŋ ɪn ˈproʊpriə pərˈsoʊnə (hɪz oʊn ˈæksɛnt ɪz əntaɪrli əmɛrəkən). Aɪ spoʊz kənˈsɪstənsi ɪz moʊr əmˈpɔrtənt ðən ɔθənˈtɪsɪti ɪn ðɪs ˈkantɛkst.

    ReplyDelete
    Replies
    1. Surely you mean "Ɋr Pi" :)

      Delete
    2. Dear John Cowan!

      I see from your transcription that you are a rhotic speaker who has unmerged HORSE-FORCE + unmerged HORSE-START. This greatly excites me. Where are you from?

      Delete
    3. ˈsɑɹrɪ, aj ˌmɛnt:

      ˈdɪɹ ˈdʒɑn ˈkawən!

      aj ˈsij frəm jʊɹ ˌtrænˈskrɪpʃən ðət ˌjʊɹ ə ˈrowtɪk ˈspijkɚ huw hæz ˈʌnˌmɜɹdʒd ˈhɔɹs-ˈfors + ˈʌnˌmɜɹdʒd ˈhɔɹs-ˈstɑɹt. ˌðɪs ˈgrejtlɪ ɪkˈsajts mij. ˈhwɛɹ ɚ jʊ ˈfrɑm?

      Delete
    4. Aɪm səˈpraɪzd ðət ju prəˈnaʊns ðə fɝst ɑr ɪn ðə wɝd səˈpraɪzd. Aɪm ˈɔlsoʊ səˈpraɪzd baɪ ðə fɝst vaʊlz (ˈvaʊəlz? vælz?) ɪn "əntaɪrli" ənd "əmˈpɔrtənt". ˈSiɪŋ ðə ˈdɪfθɔŋ "oʊ" ɪn "moʊr", bʌt ðɛn "ɔr" ɪn "əmˈpɔrtənt" wʌz ˈɔlsoʊ ə bɪt səˈpraɪzɪŋ tə mi. Bʌt aɪm moʊst səˈpraɪzd baɪ "ˈkantɛkst". Wʌz ðæt ə mɪˈsteɪk? ˈƩʊdənt ðæt bi 'kɑntɛkst? Aɪm ʤəst ˈkjɝiəs əˈbaʊt jɚ ˈæksɛnt.

      (ˈSɑri fɚ maɪ oʊvɚˈjus ʌv kwoʊˈteɪʃən mɑrks æn ðə wɝd səˈpraɪzd. Maɪ əˈpɑlədʒiz ˈɔlsoʊ fɚ ɛni trænˈskrɪpʃənəl ˈɛrərz ænd/ɔr ˌɪnkənˈsɪstənsiz.)

      Delete
    5. Jason, Levente:

      Aɪm sɑri tu əv meɪd mɪsˈlidəŋ məˈsteɪks. Maɪ nɔrθ ən fɔrs wɚdz ɑr kəmˈplitli mɚdʒd, ən moʊr wʌz ə ˈsɪmpəl ɛrɚ fɚ mɔr; ˈlaɪkwaɪz ˈkantɛkstˈkɑntɛkst ən sərˈpraɪzdsəˈpraɪzd.

      Aɪ mAɪt seɪ ɪnˈtaɪrli ən ɪmˈpɔrtənt ən aɪsəˈleɪʃən, bʌt ən ˈrʌnəŋ spitʃ ðə fɚst ˈvaʊəlz ˈwikən tə /ə/, wɪtʃ əz maɪ ˈoʊnli "nɔrməl" wik ˈvaʊəl.

      Aɪ hoʊp Aɪv meɪd noʊ ˈfɚðɚ blʌndɚz.

      Delete
  13. Maɪ ˈæksənt kʊd nɒt ˈˈpɒsəblɪ biː ɪnˈtaɪəlɪ əˈmerɪkən æz Aɪ æm ˈlɪvɪŋ ˈnɪəlɪ ˈtwentɪ θriː jɪəz ɪn ˈAɪələnd. Maɪ ˈæksənt ɪz ˈsɜːtnlɪ nɒt ˈⱭː ˈPiː, bʌt ðæts ðə ˈlæŋɡwɪdʒ ðə bʊk eɪmz fɔː.

    ReplyDelete
  14. Well, AmE is essentially Hiberno-English as spoken by a German anyway, but the last time we met I didn't hear any Irishisms in your accent (as opposed to your vocabulary).

    ReplyDelete
  15. What I don't understand is why 'alphabet' and 'system' occur in normal spelling in this passage:

    ƌi alphabet mei oolsou bi emploid (wið sjutabl konvenʃonz) az a ɟenəral system ov speliŋ for kərent pərposez and az a translitereiʃon system

    Are they emphasized or distinguihed in any way in the original?

    ReplyDelete
  16. The World Orthography, basically a superset of the Africa Alphabet with other IPA letters, uses capitals. It would be interesting to see what they used.

    ReplyDelete
  17. The World Orthography is covered in some editions of the Principles of the International Phonetic Association (like 1949). In Daniel Jones’s 1948 Difference between Spoken and Written language, an adaptation of WO for English is given with the letters a b c d ð e ə f g h i j k l m n ŋ o p r s ʃ t θ u v w x y z ʒ. The capitals of ŋ, θ, ð, ʃ, ʒ, ə are given in a note and look like the African Ŋ (large ŋ), Θ, Ƌ (instead of Ð), Ʃ (like Sigma Σ), Ʒ (like turned Σ) and Ə.

    Dieth’s capital ʃ is interesting; in the Alphabet scientifique des langues du Gabon the capital looks like a large ʃ, but it's an exception.

    ReplyDelete
    Replies
    1. You mean like this:
      https://www.facebook.com/photo.php?fbid=438287212926951&set=a.104253596330316.14164.100002370286913&type=1&theater

      Delete
    2. Yes that's pretty accurate.

      Delete
  18. Apparently there is a mistake in the table where between and Kk should be Jj.

    ReplyDelete