Grosjean, F. (1983). How long is the sentence? Prediction and prosody in the on-line processing of language. Linguistics, 21, 501–529.
Presentation: Shelly
Summary: Sally
In this study, two questions were investigated: (1) whether listeners can tell when a sentence is over when listening to the potentially last word of a sentence; (2) if they can, whether they are able to indicate how much more the sentence would last. The author designed 32 sentences, which were further categorized into four types on the basis of their structures. Zero-word-ending sentences started with a sentential adverb and ended on an object noun, e.g. Yesterday my sister made a cake. Three-word-ending sentences were identical to zero-word-ending ones except that a three-word prepositional phrase was added to the end, e.g. Yesterday my sister made a cake for the fair. In a similar sense, six-word-ending sentences have another three-word prepositional phrase added to the end, e.g. Yesterday my sister made a cake for the fair at the school. The nine-word-ending ones have still another prepositional phrase added to the end, e.g. Yesterday my sister made a cake for the fair at the school on the hill.). The gating paradigm (Grosjean, 1980) was adopted for the experiments. In Experiment 1, a 4AFC identification task was conducted, in which 32 participants were asked to circle the correct answer from the four sentence types on their test sheet upon listening to different gates of the four types of sentences. They were also asked to indicate how confident they were in giving their answers. In Experiment Two, a different task was conducted: 24 out of the 32 who attended in Experiment 1 were asked to press a key at the point when they thought the sentence is over on 0% and 100% gates of different stimuli sets from the gate pool of Experiment 1. Results showed that regardless of task difference, participants were surprisingly accurate at predicting the length of upcoming endings in both experiments. The prediction was random at the earlier gates (0-50%), but became very good at the end (75-100%). Acoustic analyses were later done, showing a strong relationship among the F0, amplitude and durational measures. In addition, as compared to those of the entire sentences, the acoustic measures of the potentially last word were better predictors of the experimental data, which indicates that the only necessary information for prediction is carried by the word itself.
Grosjean, F. (1983). How long is the sentence? Prediction and prosody in the on-line processing of language. Linguistics, 21, 501–529.
Presentation: Shelly
Summary: Sally
In this study, two questions were investigated: (1) whether listeners can tell when a sentence is over when listening to the potentially last word of a sentence; (2) if they can, whether they are able to indicate how much more the sentence would last. The author designed 32 sentences, which were further categorized into four types on the basis of their structures. Zero-word-ending sentences started with a sentential adverb and ended on an object noun, e.g. Yesterday my sister made a cake. Three-word-ending sentences were identical to zero-word-ending ones except that a three-word prepositional phrase was added to the end, e.g. Yesterday my sister made a cake for the fair. In a similar sense, six-word-ending sentences have another three-word prepositional phrase added to the end, e.g. Yesterday my sister made a cake for the fair at the school. The nine-word-ending ones have still another prepositional phrase added to the end, e.g. Yesterday my sister made a cake for the fair at the school on the hill.). The gating paradigm (Grosjean, 1980) was adopted for the experiments. In Experiment 1, a 4AFC identification task was conducted, in which 32 participants were asked to circle the correct answer from the four sentence types on their test sheet upon listening to different gates of the four types of sentences. They were also asked to indicate how confident they were in giving their answers. In Experiment Two, a different task was conducted: 24 out of the 32 who attended in Experiment 1 were asked to press a key at the point when they thought the sentence is over on 0% and 100% gates of different stimuli sets from the gate pool of Experiment 1. Results showed that regardless of task difference, participants were surprisingly accurate at predicting the length of upcoming endings in both experiments. The prediction was random at the earlier gates (0-50%), but became very good at the end (75-100%). Acoustic analyses were later done, showing a strong relationship among the F0, amplitude and durational measures. In addition, as compared to those of the entire sentences, the acoustic measures of the potentially last word were better predictors of the experimental data, which indicates that the only necessary information for prediction is carried by the word itself.
Presentation: Shelly
Summary: Sally
In this study, two questions were investigated: (1) whether listeners can tell when a sentence is over when listening to the potentially last word of a sentence; (2) if they can, whether they are able to indicate how much more the sentence would last. The author designed 32 sentences, which were further categorized into four types on the basis of their structures. Zero-word-ending sentences started with a sentential adverb and ended on an object noun, e.g. Yesterday my sister made a cake. Three-word-ending sentences were identical to zero-word-ending ones except that a three-word prepositional phrase was added to the end, e.g. Yesterday my sister made a cake for the fair. In a similar sense, six-word-ending sentences have another three-word prepositional phrase added to the end, e.g. Yesterday my sister made a cake for the fair at the school. The nine-word-ending ones have still another prepositional phrase added to the end, e.g. Yesterday my sister made a cake for the fair at the school on the hill.). The gating paradigm (Grosjean, 1980) was adopted for the experiments. In Experiment 1, a 4AFC identification task was conducted, in which 32 participants were asked to circle the correct answer from the four sentence types on their test sheet upon listening to different gates of the four types of sentences. They were also asked to indicate how confident they were in giving their answers. In Experiment Two, a different task was conducted: 24 out of the 32 who attended in Experiment 1 were asked to press a key at the point when they thought the sentence is over on 0% and 100% gates of different stimuli sets from the gate pool of Experiment 1. Results showed that regardless of task difference, participants were surprisingly accurate at predicting the length of upcoming endings in both experiments. The prediction was random at the earlier gates (0-50%), but became very good at the end (75-100%). Acoustic analyses were later done, showing a strong relationship among the F0, amplitude and durational measures. In addition, as compared to those of the entire sentences, the acoustic measures of the potentially last word were better predictors of the experimental data, which indicates that the only necessary information for prediction is carried by the word itself.
