givenness: Givenness illusion data illustrating the "Williams Effect"
In prosodylab/prosodylabDatasets: Data from prosodylab

When we speak, some words have more emphasis than others: this is an aspect of prosody that linguists call *prosodic prominence*. Words that encode new or contrastive are likely to be prominent (e.g., they are likely to carry a pitch accent), words that encode old or repeated information are likely to be reduced and hence not prominent (e.g, they are unlikely to carry an accent)—unless they are contrastive. This experiment is an illustration of the 'Williams Effect': In English, a phrase-final word cannot be accented if a previous phrase ended in a homophonous accented word, even if it encodes contrastive information (e.g. the following sentence sounds odd if the final 'Betsy' is accented: 'Daniel followed Betsy, then Daniel was followed by Betsy.'). This suggests that speakers treat them as if they were merely contextually given and not contrastive, a 'givenness illusion'. For more information about contrast in general and the experiment in particular, please take a look at the references.

givenness

A data frame with 13 variables:

condition: experimental condition
item: item set number
text: transcription of soundfiles
participant: participant identifier
conditionLabel: label of experimental condition
npType: pronoun or full NP?
voice: active or passive in second sentence?
order: trial number in experiment
acoustics: acoustic measures of final word
stressshift: perceptual annotation whether final word was deaccented (and hence prominence shifted to earlier word)