Coursera Flash Sale
40% Off Coursera Plus for 3 Months!
Grab it
Learn about computational methods for identifying and analyzing speech repairs in transcribed spoken language through this lecture by Mark Johnson from the Center for Language & Speech Processing at Johns Hopkins University. Explore the linguistic phenomena of disfluencies, false starts, and self-corrections that occur naturally in spontaneous speech, and discover algorithmic approaches for automatically detecting these repair patterns in speech transcripts. Examine the challenges of processing real conversational data where speakers frequently interrupt themselves, repeat words, or restart utterances mid-sentence. Gain insights into the computational linguistics techniques used to model and identify these repair structures, including pattern recognition methods and statistical approaches for distinguishing intentional repetitions from genuine repairs. Understand the practical applications of repair detection systems in speech processing pipelines, automatic speech recognition post-processing, and natural language understanding systems that must handle the messiness of authentic human speech rather than carefully planned written text.
