Abstract
French liaison gives evidence of the relationship between phonology and other levels of description. Various hypotheses attempting to explain this phenomenon in French: such as the insertion of some linking consonant or other; the inefficient truncation of citation forms, are shown to be inadequate.
Similarly, the idea of the liaised syntagm being a “phonological word” cannot begin to offer an explanation, because linking is determined by syntax. But syntax alone will not explain the barriers to liaison and enchaînement.
The word-final consonant, though sometimes lacking phonetic representation, is a mental reality. This is shown by the mannerisms of political speakers and by the best efforts of aphasies.
It is government alone which adequately defines the group for external sandhi in French.