Assimilatory /r/ insertion

In an earlier post (7/20/2006), I asked for examples of liquid dissimilation in English, such as omitting /r/ in the(r)mometer, Feb(r)uary, su(r)prise, etc.

There also seem to be cases where an /r/ is inserted into words that already contain an /r/. Some examples I’ve heard or had reported to me include:

  1. farmiliar
  2. contractural
  3. ardurous (the OED gives this as a ‘poetic variant’)
  4. verneer (would you trust this dentist?)
  5. fruneral (African American English)
  6. borogroves (this has entered the epigraphic record, conveniently for future philologists)
  7. perservere
  8. sherbert
  9. phertographer
  10. catergorize
  11. lavartory
  12. Kervorkian
  13. intergral
  14. perjorative

There are also historical examples like cartridge from cartouche, and treasury from thesauria.
This process is interesting because it is the reverse of long-distance dissimilatory loss of /r/. The existence of such a reverse process is predicted by Ohala’s theory that dissimilation results from hypercorrection on the part of the listener. According to this theory, the long-domain acoustic cues of /r/ can cause the listener to be uncertain whether there is one source of rhoticity in the word or two. Errors are possible in either direction.

Has anyone noticed other cases of this?

8 thoughts on “Assimilatory /r/ insertion

  1. Eric Bakovic

    Ed, don’t they also say Fair Lawrn for that town in NJ?

    (When Ed first told me about the Hawrthorne case, I had thought it was just hypercorrective r-insertion in codas, but now I see that it might happen more/only when there’s another nearby coda r.)

  2. trevor@kalebeul

    If I could get into this, I think I would find JC Maxwell either claiming or disputing that bastard / barstard is or was used to denote class. The extra r certainly does in rural southern England. This may all tie in somehow to talking like a pirate.

  3. Michael Covarrubias

    There certainly is one word in which I hear this more than any other. My last name. People have a hard enough time figuring out the pronunciation–my American accent says itɒ [koʷ.və.’ɻu.bjəs]. Once people know my name and can say it without stuttering, the most common mispronunciation is [koɻ.və.’ɻu.bjəs].

    Interesting that the [r] when regressively assimilated typically becomes a coda before a consonant. Is there some general consonant assimilation going on as well? The following onset consonant adds a +cons feature in the form of a sonorant to the preceding vowel. (?)

  4. Peter Kirk

    Trevor’s comment may help to explain the origin of this, for he mentions speech in rural southern England. Now for most of us in more urban south east England, both RP speakers and those whose accent is closer to Cockney, this whole post is rather hard to understand – because for many of the word pairs you mention we make no difference in pronunciation, or a very subtle one of vowel quality only. For us /r/ is silent in many positions, although it can affect neighbouring vowel qualities. But when more rural or western speakers hear words like this from more speakers like myself, or on radio or TV, they adjust the pronunciation according to their own accent, but sometimes incorrectly disambiguating the /r/. And a similar effect may work in America between accents which pronounce or drop /r/. The presence of another /r/ in the word may affect this, as may similarly with other words (e.g. “sherbert” to match “Herbert”, “perjorative” to match “perjury” and many other words with a “per-” prefix); this last point alone can explain “borogroves”, to match “groves”. Buy I’m not sure how “Febuary” can be explained like this, for we pronounce both r’s here.

  5. Nancy Hall

    Thanks for the examples and suggestions, everyone. I’ll add those to my list.

    I had a look at the Maxwell article that Trevor suggested, which is a reply to a reply to a reply to an earlier article that I couldn’t track down. It appears that he’s discussing a difference in vowel quality (b[æ]stard vs. b[ɑ]stard), rather than actual presence / absence of an [r]. The argument is apparently in relation to an instance of ‘barstard’ in a Shakespeare Folio.

    Peter’s suggestion is interesting. I’m not sure, though, that the dialect contact hypothesis is supported by what’s known of the history and geography of the process. Hempl describes dissimilation as extensive in his dialect of Eastern Southern Michigan in 1893. I doubt that there were many [r]-droppers around that area, and obviously the broadcast media were not a factor yet.

  6. Ed

    I asked a coworker who lives in Fair Lawn and he had never heard Fair Lawrn. But he;s not a native.
    On a related note, Novartis is launching a new drug called Tekturna, which is causing all sorts of pronunciation problems, most people say Tekturner.

Leave a Reply

Your email address will not be published. Required fields are marked *