Wednesday, 5 May 2010

scoprar il nuovo dì

Pronunciation is not determined by spelling. Contrary to popular belief among non-phoneticians, when we speak (or sing) we do not say letters aloud. I have always thought that one of the best pieces of evidence for this fact is the intrusive r that we non-rhotic speakers use — an r-sound where there is no letter r. This sound can only result from neurological (phonological) processes in the brain, not from marks on paper.

We use intrusive r even, and perhaps most strikingly, when attempting to speak or sing in foreign languages.
Here’s a YouTube clip of a pubful of English people singing Que/Y Viva España. You can just about detect the r in ˈvi:vɑːr eˈspænjɑː. Faced with a final open vowel at the end of viva, followed by another vowel at the beginning of the next word España, it would require a rather sophisticated effort for us not to link them with an r-sound.
(Since the karaokeers are nonrhotic, they naturally also sing por favor as ˈpɔː ˈfæˈvɔː. Please don’t anyone write to tell me that in proper Spanish it’s i ˈβiβa esˈpaɲa, poɾ faˈβoɾ.)

That was untrained karaoke. But even trained singers do the same thing. The choir I sing in is currently rehearsing the drinking song from La Traviata.
The song ends with everyone promising to enjoy drinking and singing until dawn.
Godiamo, godiamo, godiamo,
la tazza e il cantico
la notte abbella e il riso,
in questo paradiso,
ne scopra il nuovo dì,
si, ne scopra il nuovo dì.

For each song our choir plans to perform, the music team very nobly prepare rehearsal tracks for us to listen to as we learn the words and music (blog, 19 January). Here’s a clip of the very last phrase, from the tenor 2 rehearsal track. It contains a very clear intrusive r between scopra and il.
This also serves as a counterexample to the claim sometimes made, that a preceding r (scopra) blocks the operation of the r-insertion rule.


  1. I think that if 'scoprar' in the title is meant to be an infinitive (with apocope of the final vowel), then it should be 'scoprir' (i.e. from 'scoprire'). ('Scopra' in the text is a subjunctive form, so adding an 'r' won't give the infinitive.)


  2. Vincent Grousset5 May 2010 at 11:23

    Unfortunately "viva España" reminds Franco's era and its intolerant patriotism to many Spaniards. For example it would be a very very bad idea shouting "viva España" in Euskadi or Catalonia...

  3. Oh, please. It is not meant to be an infinitive. It is meant to show the English treatment of an Italian word.

  4. The odd thing here is how the music sets the Italian phrase. Normally Italian doesn't have hiatus except in very careful speech, and the line would be rendered ne scopr' il nuovo di (iambic trimeter, septasyllabic) in spoken verse. Similarly, the third line would be la nott' abell' ejl riso, also a septasyllabic, where ejl is ad hoc notation for the contraction of e il into a diphthong. So I was quite surprised to hear -pra and il in separate syllables when sung.

    The most striking example of this I know is actually Latin, and is one of my favorite lines of Vergil, describing Polyphemus. It's written monstrum horrendum, informe, ingens, cui lu:men ademptum (16 syllables) but scanned as monstr' 'orrend', inform', ingens, cui lu:men ademptum (only 13 syllables).

  5. I guess di should be ("day", though nowadays one would normally say giorno). Di means "of".

    @Cowan: I'd rather say "except if there's a major prosodic break between the two vowels or if (sometimes) if either vowel is stressed" than "except in very careful speech". Even in the most formal situation I can think of, consecutive unstressed vowels in the same intonation group would at best be pronounced as a diphthong [skoː.pra̯il.nwɔː.vo.di] (unless one after saying scopra stopped to think about how to continue the sentence). (In poetry you can also get three consecutive stressed vowels with intervening punctuation to go in the same syllable...)

  6. It looks in my font as though there are two different "a"s in "esˈpaɲa". Is that a real thing, or a font issue?

  7. Conversely, Spaniards tend to introduce an "e" before liquid "s" in English. Not just at the beginning of a word ("Espanish"): the brand name "movistar" (mobile telephone) is commonly pronounced like "moviestar".

  8. Remember that in proper Spanish that would actually be [i ˈβiβa esˈpaɲa, poɾ faˈβoɾ].

    /s/ is not a liquid consonant. It's a voiceless alveolar fricative. /l/ and /r/ are examples of liquids in English. But yes, hispanohablantes do have a tendency to say "esnake" and things like that when speaking English. I knew a foreign exchange student from Chile who did that.

  9. Tonyo, that's because there's a morpheme border there, otherwise /-ist-/ would be fine.

    "Anonymous", yes, a Chilean would typically even have /ehnejk/ or /enejk/ for snake.

  10. Lipman, I assume that the now-famous Chilean was speaking English and pronouncing the word "esnake" in English complete with a perfect [s] and the imperfect addition of [e].

    Up to around 40 years ago there was a push by the Academy of the Hebrew Language to eradicate the consonant clusters begining with [s] (e.g. [st], [sp] etc.) from the language which exist in loan-words such as sport (nowadays pronounced [ˈspoʁt]). All radio broadcasters were made to pronounce such word-initial clusters with an [a] sound in the same manner that Spanish adds [e], so [ˈaspoɾt] was the "correct" norm (with the alveolar tap also enforced from above).
    Obviously this didn't succeed in transmitting over to the regular spoken language because Hebrew speakers already were able to pronounce [sp] without a problem.
    Can monoglot Spanish speakers not pronounce such clusters, or is it just a norm that is firmly entrenched within the language? these clusters do occur in words where the [s] is part of a syllable coda (such as "raspar", "estar", "escapar" etc.)

  11. It's interesting to see how this phenonemon developes in affected languages, eg in Italian, where it went back and consistently took along harmless words with an etymological vowel in front of the -st- cluster, una storia straordinaria, or in some South American Spanish varieties, where the e- stays, and the -s- vanishes as all esses at the end of a syllable.

    Israeli Hebrew superstitiously believes in Biblical grammar, where no word-initial clusters or even affricates existed. Some rare loans there and lots more, mostly Greek, in Mishnaic Hebrew and Talmudic Aramaic, where the phonotactic rules were still similar, get a prosthetic vowel. Of course, the phonetics and phonotactics of Israeli Hebrew are entirely those of South-Western Yiddish.

    Monoglot Spanish speakers in fact "can't". I take it the difference is that in raspar &c. it's mentally ras- and -par, not ra- and -spar, hence the problem with movistar, where the brain knows it's movi- and star.

    Cluster surrounded by vowels are always (?) easier in a language. Take the German for "he's using it quite a lot": Er nutzt's ziemlich oft, with the cluster of [tststs]. Can't think of a German word starting with that.

  12. Israeli Hebrew are entirely those of South-Eastern Yiddish, I meant, commonly called Ukrainer.

  13. Italian used to have an epenthetic I when /sC/ came to be after a consonant (e.g. "Spain" = Spagna, but "in Spain" = in Ispagna), but this is very old-fashioned now.

  14. My (native Hindi-speaking) father-in-law rather charmingly inserts prothetic /ɪ/ before English loanwords beginning with clusters such as "station", which becomes [ɪsteːʃən]

  15. But even trained singers do the same thing...

    Perhaps these singers could benefit from some extra "training" from a distinguished phonetician such as yourself :) I would expect a classically trained singer to know enough about Italian to avoid intrusive R.

  16. A strornery story indeed, Lipman, and a depressing one about SE Yiddish and bungled attempts to patch it up from on high. But what chance could there ever have been of imposing a Latin-type "reconstructed pronunciation" on all those people who actually knew a usable version of the language?

    I note you say "it's mentally ras- and -par, &c." Are you sure about the &c? Do you agree with Gadi about the s of the verb estar also being mentally in the coda? I find it hard to imagine that monoglot Spanish speakers think of it any differently than the movie version, whether they are etymologically aware or not.

    In Ispagna can never die in Leporello's catalogue of Don Giovanni's conquests!

  17. Don't get me wrong, Israeli Hebrew is a fascinating language I quite like, the frustrating part is mainly that it's generally perceived to be something else. Then again, try and tell a Greek that the language of Homer isn't exactly the same he's using.

    Not sure what you mean concerning estar.

  18. Lipman is right: "raspar", "estar", "escapar" are thought as "ras-par", "es-tar", "es-ca-par", and so there is no "e" in "raspar", whereas "movistar" is generally thought as "movi-(e)star".

    @Gadi: "Can monoglot Spanish speakers not pronounce such clusters, or is it just a norm that is firmly entrenched within the language?"
    When the cluster appears at the beginning of the word, it is almost impossible for a native Spanish-speaker. Even when we sing lyrics in other languages, a new syllabe miraculously appears and spoils the song. In adopted foreign words, the "e" may appear even in the ortography, and is often officialized by the Academy (e.g. "espín")

  19. I can't resist adding my favorite instance of Spanish prosthetic "e": I once heard a Hispanophone tram driver in Boston announce the Boylston Street stop as [bojlestonestrit].

    Lipman, I very much enjoy your erudition.

  20. "This also serves as a counterexample to the claim sometimes made, that a preceding r (scopra) blocks the operation of the r-insertion rule."

    I was just reminded of this on the most recent episode of Lost, when Naveen Andrews (a Londoner playing an Iraqi) refers to Hydra Island (a setting on the show) as "Hydra-r-Island".

  21. Lipman, the reason you're not sure what I mean about estar is that I was talking rubbish. But the rubbish was just a very badly expressed attempt at common sense. My excuse is that it's a bit difficult to make any sense of all this at all: I was trying to say I don't actually think you can be mentalistic about any of the codas in "raspar", "estar", "escapar" etc, as I see such concepts as "coda" as descriptive models in linguistic theory, not in psychology! And if Tonyo is right about Spanish speakers thinking of the verb "estar" as "es-tar" and the "star" in "movistar" as "(e)star", then monoglot or no, then that psychology is linguistically all over the place! But if I am to change my ideas about monoglot Spanish speakers not thinking of them any differently, I would have to come to terms with the idea that there is something for them to think differently about. You certainly encourage this idea when you say you "take it the difference is that in raspar &c. it's mentally ras- and -par, not ra- and -spar, hence the problem with movistar, where the brain knows it's movi- and star", as you seem to be basing your argument on the idea that the s of the verb estar is in the coda mentally as well as in terms of linguistic analysis, but that equally mentally it's somehow part of an initial cluster in movistar. What, because of the spelling? You say monoglot Spanish speakers in fact "can't" (pronounce such initial clusters without the e). In the pronunciation the e is either there or it's not. It doesn't depend on Academy-sponsored spellings like "espín". Or because of awareness of its English origin? When it comes to etymological awareness, I was thinking even if it were credible that monoglot Spanish speakers go so far as to think panchronically rather than panlinguistically, the e is no less parasitic in estar for Latin stare than in estar for English star.

    Thanks for the wonderful Er nutzt's ziemlich oft! Now I'm finding it hard to imagine that you don't get [tsts] or something for [tststs], as with E [nekstɒp].

  22. Aha. I think. Anyway, here's what I think:

    A typical speaker of Spanish will add a prosthetic /e/ to a foreign word starting with /st/. The same is true if there's a morpheme boundary, juncture or what you like to call it, because the part starting with /st/ is perceived to be initial. So, if there were a Spanish verb *movistar from Latin *mobistare, it wouldn't get the extra vowel, but if there's a word movistar consisting of movi and star, which parts are still understood to be such by the speaker, it will get the vowel.

    The breakdown of /st/ to /-Vs#/ and /#tV-/ is a linguistic description that explains the vowel, it's nothing the average speaker would know of. In fact, they may not even be aware of the vowel at all when it's about new words, as opposed to established words right down from Latin (eg estar). That simply shows that the phonotactic rule is still active.

  23. I'm finding it hard to imagine that you don't get [tsts] or something for [tststs]

    I asked a native speaker and she confirms that:

    - People do say [tststs], it doesn't sound awkward and in most cases, it wouldn't occur to the speaker to be strange.

    - If they don't pronounce the full [tststs], they'd rather have [tstəsts], ie "nutzt es ziemlich".

    - Simply reducing it to [tsts] wouldn't work. People would understand it as "nutz ziemlich", actually lacking the object.

    In Northern dialects, the -t of the third person sg. is dropped anyway.

  24. I'm confused. Wouldn't ‹favor› end in [r] (rather than [ɾ]) in proper Spanish? According to Wikipedia, “[r] is found after /l/, /n/, and /s/, before consonants, and utterance finally; [ɾ] is found elsewhere” – as far as I recall, the Spanish textbook we used in school said the same.
    But Wiktionary says it's [po̞ɾfaˈβo̞ɾ].
    But then again this sounds like [-r] to me.

    Wikipedia further says “in more casual speech, a preconsonantal rhotic is the tap rather than the trill” – does this mean that in “por favor”, you'd use [-ɾf-] when speaking casually and [-rf-] in careful or formal speech?

  25. I never doubted that people do say [tststs]. I just guessed you might get [tsts] or something as well, as with E [nekstɒp]. E [nekststɒp] wouldn't occur to the speaker to be strange either.

    And it appears you do get the something. Thanks for telling me that that something is [tstəsts], ie the elision in "nutzt es ziemlich" is blocked.

    I did consider askinɡ about [tsːts] rather than [tsts], but that wouldnʲt have been so interestinɡ. It sounds as if it might be a possible option for the Northern dialects in which the -t of the third person sg. is dropped, though.

  26. [nʊtsːːimlɪç] is definitely a possibility for me, with two intermittent crescendos inside the overlong [s] to mourn the lost affricates ([ts.s̩.s]?)

  27. LOL. As interesting as it gets! And it does seem to suggest a limbo inhabited by my less interesting [tsːts].

  28. @John Wells
    Just noticed you have changed the title of this to "scoprar il nuovo dì". Must have missed your post "OK, have corrected dì". Or I would have expressed my regrets that you had not paid homage to the use of the acute and grave accents to differentiate between open and close vowels, and written "dí".

  29. Turkish has epenthetic "i".

    French also had epenthetic "e", as in être < STARE, but the rule is now dead: there are now plenty of borrowings in st-, sp-.