1.   A probability was assigned to any word string that could be formed from the spoken input.

2.   The syntactic analyser identifies the most syntactically acceptable word strings, whilst the semantic analyser identifies the most semantically plausible.

3.   There are also equivalence classes of word strings created by lack of word boundary information, e.g.

4.   A typical measure based on dynamic programming techniques aligns the recognised word string with the correct word string and reports correct, missing, substituted and inserted words.

5.   However, there were many other words and word strings covering the region of potato, so why should LA apply special procedures to rescue this particular word?

6.   However, we found indications in the last chapter that this could lead to a very large number of possible word strings.