LING 1P92 Lecture Notes - Lecture 4: Sound, Parallel Communication, Virtual Representation

• School: Brock University
• Department: Linguistics
• Course: LING 1P92
• Professor: Lynn Dempsey

Document Summary

Speech perception: our misperceptions tends to perform in grammatical rules, slips from ear//inattentiveness, we can identify 25 speech sounds per second. Identifying the individual phonemes (smallest unit of speech that have meaning) in the speech signal. Attaching correct meaning to the phonemes, matching them to representations stored in memory. Spectrogram: virtual representation of a sound wave, frequency on vertical axis, time on horizontal axis, loudness represented by darkness, dark bands represent formants. Cues for identifying speech sounds: position of formants, differences in shapes of formants, clarity of the formant structure. Formants: high tongue position= low f1, low tongue position=high f1, forward tongue position= high f2, back tongue position= low f2. Cues for identifying vowels: cvc combinations, 3 sound segments. Why speech processing is difficult: lack of invariance. Role of context in speech perception: bottom-up processing. Speech perception is triggered automatically based only on the acoustic signal. When this is efficient, top down is unnecessary: top down processing.