1.   And accordingly, listeners process acoustic information less carefully when dealing with predictable words.

2.   Furthermore, the same acoustic information can be perceived differently depending on its higher-level context.

3.   The semantic constraint may take precedence over acoustic information.

4.   The ability of speech recognition systems to discriminate words from acoustic information alone is not encouraging.

5.   The syntactic and semantic components must select from word hypotheses that are constantly changing their status or score as more acoustic information becomes available.

6.   We increased the acoustic information by incorporating lexical stress into the representation.

7.   Increasing the specificity of the acoustic information.