subwords.R

# Find words (in a dictionary file) that are made up of other words run together.
#  e.g.: "deliberating" = "deli" + "berating"
#  e.g.: "categorising" = "cat" + "ego" + "rising"
# By Nick Brown (nicholasjlbrown@gmail.com), June 2025.

# Some parameters. I hope their meaning is reasonably obvious.
min.split.words <- 2
max.split.words <- 3
shortest.single.word <- 3
longest.single.word <- 8
shortest.compound.word <- 12

# My data file. Plain text, one word per line.
filename <- "corncob_lowercase.txt"
words <- read.table(filename)$V1

if (shortest.single.word <= 1) {
  words <- c(words, "a", "i")     # these one-letter words aren't in my dictionary
}

words.bylen <- list(rep(NA, longest.single.word))
for (i in 1:longest.single.word) {
  words.bylen[[i]] <- words[nchar(words) == i]
}

# We are only interested in compound words with a certain minimum length.
words <- words[(nchar(words) >= shortest.compound.word)]

subwords <- function (word, level=1, start=1) {
  if (level > max.split.words) {
    return(NULL)
  }
  word.len <- nchar(word)

  for (len in shortest.single.word:min(longest.single.word, word.len - start + 1)) {
    end <- start + len - 1
    part <- substr(word, start, end)
    if (nchar(part) < shortest.single.word) {
      return(NULL)
    }

    if (part %in% words.bylen[[len]]) {
      if (end == word.len) {
        if (level >= min.split.words) {
          return(list(part))
        }
      }
      else {
        rest <- subwords(word, level + 1, end + 1)
        if (!is.null(rest)) {
          return(c(list(part), rest))
        }
      }
    }
  }

  return(NULL)
}

for (i in 1:length(words)) {
  word <- words[i]
  sub <- subwords(word)
  if (!is.null(sub)) {
    cat("Found ", word, " (", sep="")
    for (i in 1:length(sub)) {
      cat(sub[[i]][1])
      if (i < length(sub)) {
        cat(".")
      }
      else {
        cat(")\n")
      }
    }
  }
}