A relative of mine recently went on a rant regarding the pronunciation of 'jewelry' (as joo-la-ree) and 'realtor' (as ree-la-ter). It reminded me of the oft criticized pronunciation of 'nuclear' and I wondered if this represents a pattern of native speakers avoiding putting /l/ in consonant clusters mid word. Are there more example like this? Is this a pattern?

  • 4
    It's specific consonant clusters. In the case of jewelry, it's an /lr/ cluster. In the case of realtor, it's an /lt/ right before an /r/. In the case of nuclear, it's an unstressed ambisyllable /kliyər/. There are several ways to remedy the situation for a challenged speaker. One can insert a vowel (normally /ə/) to break up the cluster (this is called "epenthesis", and the vowel is an "epenthetic vowel"); this is what happened with the first two. Or one can swap in bisyllabic /kyələr/ for /kliyər/, which is called "metathesis". Commented Aug 5, 2014 at 23:54
  • @JohnLawler: 'already' has an /lr/ cluster. 'alter' has /lt/ before /r/.
    – Mitch
    Commented Aug 6, 2014 at 2:45
  • The vowel before the cluster may make a difference, because of the different lip positions to form different vowels.
    – Barmar
    Commented Aug 7, 2014 at 6:00
  • 1
    There's also 'February' and 'library' which have difficult 'r' clusters for some people. Febewry and li-berry. Not forgetting seckerterry.
    – Mynamite
    Commented Aug 11, 2014 at 21:39
  • 2
    Bear in mind that, outside the US, jewellery is spelt thus. Perhaps the pronunciation you describe is related? Commented Nov 14, 2014 at 23:36

1 Answer 1


Short answer, Yes.

An interesting way of looking at this pattern might be from a perspective of Latin vs. Germanic origins.

Nuclear, Realtor, and Jewelry are all Latin borrowings through French. For modern English a more Germanic syllable structure has been imposed onto this Latin vocabulary. This has created new "proper" pronunciations for these as English words. So from New-clee-air to Nuke-lee-ur, from Re-al-tor to Reel-tor, and from Joule to Jew-el or Jule.

The effect of breaking and rejoining these syllables creates consonant clusters that are "unexpected" in English. A more "native" syllable has transitional vowels surrounding the 'l'; a word like melon can survive 3 different syllable interpretations and still sound very much the same (stress on the 1st mel-on, on the 2nd meh-lon, or unstressed mehl-lin).

Another more native usage of the 'l' is as a transition from and into a fricative. But here we still see that Latin imports may subjected to the Germanic structure. The 'alt' cluster hardening so that Altar becomes All-ter lining up more closely with Germanic compounds like All-Ready rather than softening like the French Auter.

There's another interesting transformation in the tension that 'l' creates between Latin and Germanic words. Words like Principle and Participle are consistently pronounced in a way that obviously contradicts their spelling. Again we can attribute this to the syllabic breaks, the appropriate French breaks being more like 'prin-see-plai-' and the English making a 'prin-sip-l". The creation of the break between the p and l creates a very awkward scenario in English, which is uniformly resolved by making the 'ple' ending into 'pel'.

In this case we see that the strength Latin exerted over writing was enough to overrule the German imports, Apfel becoming Apple and Stapul becoming Staple, in spite of the actual pronunciation.

  • Resonants tend to fade into one another, and /r/ and /l/ are very similar resonants. In Latin, they were actively dissimilated in suffixes, for instance; and words like colonel remind us of the resemblance. So, whenever you see a strangely placed L in a word, or hear an /l/ in a spoken word when you don't expect it, always check to see whether there's also an /r/ somewhere in the woodpile, as there is in all of jewelry, realtor, and nuclear. Commented Oct 15, 2014 at 19:36
  • @John Additionally, PIE itself had a similar active dissimilatory rule in certain suffixes (the nomen instrumentis suffix *-tlo- in particular); cf. Latin stabulum vs. latrina, both with the same suffix. Commented Oct 15, 2014 at 19:58
  • Yes, and the tuli and latus forms of fero had some peekaboo laterals accelerating things, too, iirc. But resonants are always marginal consonants, one way or another. Commented Oct 15, 2014 at 20:23
  • Good supplemental notes! Commented Oct 16, 2014 at 3:14
  • It particularly makes me ask whether the presence of the r in nuclear has a cognitive "trigger" that might explain why Nuclear is a more common trap to people than Nucleus. Or a relationship to Heracleos morphing into Hercules. Does the mere proximity of the resonants generate a processing fault even when they don't directly interact? Commented Oct 16, 2014 at 3:22

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.