Primacy and recency effects in serial recall of synthesized vowel sequences

Abstract

Recognizing running speech involves not only audibility of speech signals but also cognitive processes that support the encoding and retrieving of speech information to and from memory. The current study aims to separate the effects of audibility and cognitive load on speech recognition performance. Synthesized vowel sequences were presented in noise, and the listeners were instructed to recall all vowels in the sequences in the correct order. In separate conditions, the sequence lengths were 2, 5, and 8 vowels, the presentation rates were 1, 2, 4, and 6 Hz, and the vowel durations were between 12.5 and 100 ms. The signal-to-noise ratios used in the experiment were adjusted for each listener and each of the vowel durations so that the expected recognition performance for isolated vowels was always 80%. A primacy effect was observed and it was more prominent at higher presentation rates and longer sequence lengths. A recency effect was also observed in most listeners, but no significant effect of presentation rate was found.