How to match binomial expressions in R?

user_8039978 0
Viewed 50

I want to match binomials, that is, bisyllabic words, sometimes hyphenated, with slightly varied syllable reduplication; the variation always concerns the first (and, possibly, second) letter in the reduplicated syllable:

x <- c("pow-wow", "pickwick", "easy-peasy", "nitty-gritty", "bzzzzzzz", "mmmmmm", "shish", "wedged", "yaaaaaa")

Here, we have said syllable reduplication in pow-wow, pickwick, easy-peasy, and nitty-gritty (which are then the expected output) but not in bzzzzzzz, mmmmmm, shish, wedged and yaaaaa.

This regex does at least manage to get rid of wedged(which is pronounced as one syllable) as well as monosyllabic words by requiring the presence of a vowel in the capturing group:

grep("\\b\\w?((?!ed)(?=[aeiou])\\w{2,})-?\\w\\w?\\1\\b$", x, value = T, perl = T)
[1] "pow-wow"      "pickwick"     "easy-peasy"   "nitty-gritty" "yaaaaa"

However, yaaaaa is getting matched too. To not match it my feeling is that the capturing group should be disallowed to contain two identical vowels in immediate succession but I don't know how to implement that restriction. Any ideas?

regex-lookarounds r regex
1 Answers

It looks as though you want to match words that cannot contain ed after the initial chars and 2 or more repeated chars if the same chunk is not found farther in the string. Also, the allowed "difference" window at the start and middle is 0 to 2 characters.

You may use

\b\w{0,2}(?!((.)\2+)(?!.*\1)|ed)([aeiou]\w+)-?\w\w?\3\b

See the regex demo

Details

  • \b - a word boundary (you may use ^ if your "words" are equal to whole string)
  • \w{0,2} - two or more word chars (replace with \p{L} to only match letters)
  • (?!((.)\2+)(?!.*\1)|ed) - no ed or two or more identical chars that do not repeat later in the string are allowed immediately to the right of the current location
  • ([aeiou]\w+) - a vowel (captured in Group 3) and 1+ word chars (replace with \p{L} to only match letters)
  • -? - an optional hyphen
  • \w\w? - 1 or 2 word charsd
  • \3 - same value as captured in Group 3
  • \b - a word boundary (you may use $ if your "words" are equal to whole string)
user_38329700
Related