[ Contents | Search |Previous | Up ]
From: Lisa LaSalle
Date: 19 Apr 2010
Time: 07:52:56 -0500
Remote Name: 69.222.77.73
Hi Erin! Your student is correct in not counting singing, but as far as whether to count utterances containing interjections or not, that brings up the issue of which rate measure you are interested in. There are two different measures - Articulatory Speech Rate (AR) and Overall Speech Rate (OSR). OSR is also known as a disfluent speech rate, and OSR is the number of words produced per number of minutes, including pauses and disfluencies, in order to deliver a complete message (e.g., monolog) or to read a complete passage. It is a simpler measure to collect than is AR. When you ask if you add up to 60 seconds,, maybe this is what you are getting at. • To Calculate of OSR: Clock the exact time in seconds from the time the speaker begins his sample (reading, monolog) until they are complete. Stop and restart the clock during durations when the conversational partner talks, when 2+ seconds (s) of silence occur, or when the speaker is yawning, coughing, etc. • Interpretation of OSR: Disfluencies and formulaic pauses take up time. Disfluencies include sound, syllable, word and phrase repetitions, with variable numbers of iterations per repetitions, prolongations > 500 ms, revisions, and interjections. Therefore, people who stutter speak at a slower OSR, averaging 123 words per minute (wpm), than do normally fluent speakers, who produce on average, 170 wpm (Walker, 1988). I don't know if this is broken down for children between 4- and 8 yrs of age. But AR is. Articulatory rate (AR) is also known as fluent speech rate, and is the number of perceptibly fluent phones or syllables produced per second, excluding all disfluencies and pauses > 250 ms (Yaruss, 1997). Of course, as you indicate, syllables per second can be converted to syllables per minute (spm) by multiplying by 60. Whereas OSR might as well be measured online, AR can be measured online or offline, and is more accurate when measured offline. It sounds like your student has a recording. • Offline Calculation of AR: Yields syllables per second (sps) or phones per second (pps). How to: o From an audio recorded sample, select three perceptibly fluent utterances that meet the following criteria: The utterances should be between 5-15 syllables in length (Rationale: short utterances, 1-5 syllables are often slower than longer utterances > 15 syllables). Note: In conversation, the utterances should be spontaneous and not rote, over-learned or automatic samples (e.g., Pledge of Allegiance) (Rationale: Rote is often faster than spontaneous). o Count the number of spoken syllables in each utterance (e.g., “I needed to go to my next class at two.” = 11 syllables) o Use a cursor on a digital audio recording program like Audacity at http://audacity.sourceforge.net/ or Praat at http://www.fon.hum.uva.nl/praat/ to determine onset and offset of the first and last phoneme of the utterance. Record the time in milliseconds (ms). Divide by 1000 to convert to seconds (1000 ms = 1 sec). Report the number of syllables per second for each of the three utterances and then report the average. For example, a speaker’s AR would average 5.35 sps, with a range of 5.20 – 5.47 sps, taken from these three measured utterances: 11 syl / 2.01 s = 5.47 sps ; 9 syl / 1.73 s = 5.20 sps; 9 syl / 1.67 s = 5.38 sps As for norms for this child, Hall, Amir, and Yairi (1999) found that there was no significant difference in the mean articulatory rate in syllables per second (sps, which as you say, can be converted to syllables per minute by X 60) for preschool-aged children (i.e., 37 to 58 months of age) who were in a control/normally fluent group (n=8) and who stuttered (n=16), followed longitudinally who either persisted in stuttering at a 2 year follow-up visit (n=8) and who recovered from stuttering (n=8). All preschoolers ranged between 3.18 and 4.12 sps, and yet there was a trend for children who would persist in stuttering to be faster speakers. Hall et al. (1999) did find a significant difference when phones per second (pps) were collected and compared across the three groups, and yet this is a more tedious measure to collect clinically. I hope this helps; I'm developing a clinical handout on speaking rate for my students this semester, because while the info is available from various sources, it doesn't seem to be compiled anywhere. I'd recommend Carol Seery and Ehud Yairi's new 2011 Stuttering text - sections of that text have helped develop the AR analysis part in particular. Bottom line - I'd recommend your student to do the AR rather than the OSR analysis and compare AR to the 3.18 - 4.12 sps norms.