Speakers sacrifice some (of the) precision in conveyed meaning to accommodate robust communication
The process of encoding an intended meaning into a linguistic
utterance is well-known to be affected by production pressures. For
example, there is mounting evidence that speakers prefer to distribute
information uniformly across the linguistic signal, which maximizes
information transmission (Uniform Information Density, UID) and
ensures robust communication. Support for UID comes from apparently
meaning-equivalent alternating forms: speakers prefer the form with
more linguistic signal when the encoded meaning is contextually
unexpected (e.g. different word pronunciations, object drop, auxiliary-
contraction, optional that-mentioning). However, it is an open
question whether and to what extent precision in encoding the intended
message trades off with the preference to distribute information
uniformly when two linguistic forms are similar, but arguably not
meaning-equivalent.
We addressed this question in a corpus study investigating the choice
between simple vs. partitive "some". These forms, while generally
assumed to differ in meaning, arguably also overlap in meaning. We
found that the preference to uniformly distribute information over the
linguistic signal affects the choice between these forms even after
controlling for meaning differences. This suggests that, rather than
production pressures applying only after restriction of permissible
forms by the semantics, the pressure to robustly communicate a core
meaning applies in parallel with the goal to find the most precise
form to encode an intended meaning.