static int ai_tok_combinePhrases_maxPhraseLength = 3; // 3 words, see #1011207

static L<S> ai_tok_combinePhrases(L<S> tok) {
  ret ai_tok_combinePhrases(tok, f isKnownPhrase);
}

static L<S> ai_tok_combinePhrases(L<S> tok, O isKnownPhrase) {
  new L<S> l;
  int n = l(tok);
  bigloop: for i over tok: {
    S t = tok.get(i);
    if (odd(i)) {
      S phrase = t;
      int j = i+2;
      while (j < n && j < i+ai_tok_combinePhrases_maxPhraseLength*2) {
        phrase += " " + tok.get(j);
        if (isTrue(callF(isKnownPhrase, phrase))) {
          l.add(joinSubList(tok, i, j+1));
          i = j;
          continue bigloop;
        }
        j += 2;
      }
    }
    l.add(t);
  }
  ret l;
}