Thus, when miscommunication does take place. a whole range of factors
may be responsible. It is probably the indirectness of the relationship
between .. linguistic knowledge" (in Chomsky's sense) and language use
which accounts for the fact that speakers with partly different grammars
understand each other much of the time in natural settings; they rely on a
wide and varied range of comprehension strategies. However, when comprehension
between such speakers is tested experimentally (i.e. when
factors I to IV above are controlled), they appear to be unable to associate
sentences not generated by their own grammar with an appropriate semantic
structure (see Labov, 1972a; Trudgill, 1981). Generally speaking. arguments
for a ··polylectal grammar" as a reflection of a speaker's pan-dialectal
competence have not been supported by experiments such as Labov's or
Trudgilrs which require subjects either to select a paraphrase for sentences
generated by other dialect grammars. or to judge whether such sentences
are ··possible" English sentences (see also Ross, 1979 for a different, but
related, approach). The gap between these experimental results and results
of the observational research reported in this paper is sometimes quite large,
and will be discussed below. Meantime. it is perhaps worth noting Plutchik 's
comments on the value of observational (as opposed to experimental)
research in a field where knowledge is limited (Plutchik, 1976, p. 23).
While then it is certainly true that speakers with markedly different
grammars are frequently able to understand each other in context, it does
not seem quite justifiable to adopt uncritically the common view cited by (for
example) Smith and Wilson that differences between the grammars of
dialects are essentially trivial, and that potential misunderstandings can be