Hide all but Video Show rest of page.

Note: Mac Safari users may have problems loading some videos. If so, use either Google Chrome or Mozilla Firefox to continue.

Jump to

Sample 40
Sample 41
Sample 42
Sample 43
Sample 44
Sample 45
Sample 46
Sample 47
Sample 48
Sample 49
Sample 50
Sample 51
Sample 52
Sample 53
Sample 54
Sample 55

Part 4 Manual

Show Manual

Please note, this is the text from the manual placed here for your convenience. The manual is also a work book and you will want to refer to it for other useful resources.

PART FOUR: RATING NATURALNESS

Rating Naturalness is arranged somewhat similarly to the practice format with which you're now familiar. Its purpose is to teach you to incorporate ratings of speech naturalness in your data collection procedures. When you have completed this section, you will be able to count syllables and stutters and rate the naturalness of a sample simultaneously. Then you will have accomplished the goal set forth at the beginning of this program: to be able to assess the three major dependent variables that are fundamental to describing, empirically, vital characteristics of the speech of people who stutter.

Although speech rate and stuttering frequency are important characteristics of speech production, it has become clear that they do not provide a complete picture of a person's speech - and especially the posttreatment speech of people who stutter. For example, speech following treatment may be stutter free and within normal speech rate ranges, yet still sound stilted or artificial or cautious- i.e., unnatural (Runyan & Adams, 1978; 1979). Or, a child learning to change stuttered speech to fluent speech may attempt to rely on sing-song speech or unusual animation to produce fluent utterances. These aspects of speech naturalness are not necessarily captured by SPM or %SS data.

The concept of speech naturalness encompasses many aspects of speech beyond rate and fluency, including prosody, vocal quality, placement and length of pauses, expressiveness, articulation, syntax and semantics, discourse conventions, and other less well-defined features. Because of this multiplicity of characteristics, only some of which may be relevant to a particular person's speech, we have selected as our measure of naturalness a perceptual rating scale developed by Martin, Haroldson, & Triden (1984). It is simple to use and has been shown to be highly reliable (cf, Ingham, 1985). The clinician/judge merely listens to a sample of speech and then rates its naturalness according to a 1-9-point scale where 1 means the speech sounded highly natural and 9 indicates the speech sounded highly unnatural. Ratings between 1 and 9 indicate perceived degrees of naturalness of between those extremes. Martin et al. (1984) found that speech naturalness ratings for normal-speaking adult talkers averaged 2.3. A rating of 3 is typically considered the uppermost boundary of normal naturalness (Ingham, 1985; Ingham, Gow, & Costello, 1985; Ingham & Onslow, 1985; Schivetti & Metz, 1997).

The next section of this training program will present speech samples that you have previously heard of nonstutterers and stutterers. The samples are one and two minutes in length and present speech that illustrates various components and degrees of naturalness. Following is a list of directions explaining how to get started rating naturalness. Read through the list before performing any of the operations. Then return to #1 and carry out the directions.

  • 1. As above, be sure the computer has been set up properly, including adjusting the volume level for the audio tone. Add each sample as a new subject according to the sample number. On the Run Configuration screen, input the proper settings, as follows.
    • Steps 1-4 (Samples 40-51) are one minute in duration. Speech samples in Step 5 (Samples 52-55) are two minutes long.
    • You will not be counting syllables or stutters during the first several samples. You will use your number keys on the keyboard to make your 1-9 Naturalness rating.
    • Reset the Run Configuration screen Click on Display Naturalness Rating, Continuously. This will make a box for naturalness ratings on the screen while data are being collected.
    • Add the components required for rating naturalness. First, click in the box beside Rate Naturalness (top, far right column) to turn on this function.
    • Then, in the box beside Rating Period, type in 60. This means that the computer will ask you to make a naturalness rating for every 60 seconds of speech, which will be at the end of each of the one-minute samples and at the middle and end of the two-minute samples.
    • Next, click Tone Prompt, so that you will receive an auditory signal from the computer when it is time to input your naturalness rating.
    • All previous settings on the Run Configuration screen should remain as previously set.
  • 2. Prepare now to listen to Sample 40.
  • 3. Before rating, play a brief bit of the sample so that you are familiar with the speaker's style of speech and have a general idea of its naturalness.
  • 4. Now you're ready to rate the naturalness of the samples that follow in Steps 1-3. All samples are one minute in duration. Step 1 will present three samples of nonstuttering speakers; Step 2 contains three samples of stuttering speakers; Step 3 contains samples with varying degrees of naturalness. Your job is to listen to each sample and concentrate on its naturalness, having in mind a rating between 1 and 9. When the naturalness tone sounds and the NA signal appears in the bottom left corner of the screen, use the number row on the computer keyboard to indicate your rating. You must press a number key within 5 seconds of the signal. At the completion of the sample the computer will automatically display your naturalness rating on the data summary screen. For samples that require multiple naturalness ratings, the cumulative averaged rating will be displayed in the NAT box during data collection, and the final averaged naturalness score will appear on the data summary screen at the end of the sample.
  • 5. When you have completed the naturalness rating(s) for a given sample, turn to the data summary page in your workbook (which follows these instructions for Part Four) and compare your ratings to the ones shown there. To determine the standard for these naturalness ratings, 27 judges (naïve undergraduate students) observed all of the speech samples contained in Part Four, each divided into 60-second segments. Each judge independently rated each segment on the 1-9-point rating scale following the same instructions given to judges by Martin et al. (1984). Without previous knowledge that they would be recalled, one to three weeks later all judges re-rated all of the samples. Only ratings from judges whose first and second ratings were reliable (within ± 1 rating unit) were included in the final calculations (20 of the 27 judges; 10 males and 10 females). The Target Ranges noted on the summary data page represent the ± 1 range around the averaged ratings of this group of judges for each speech sample.
  • 6. If your rating is not within the acceptable range, listen to the sample a second time, keeping in mind the naturalness rating assigned to that sample on the data sheet. This should serve as an exemplar of that particular rating scale value.
  • 7. When your count is within the Target Range, continue to the next sample. Stop this process when you have completed Step 3 and read the following paragraph before you begin Steps 4 and 5.

Instructions Regarding Steps 4 and 5, Naturalness Rating

In the three, one-minute samples of Step 4 (Samples 49-51) and the three, two-minute samples of Step 5 (Samples 52-55) you will integrate all of your new skills and simultaneously count syllables, stutters, and make naturalness ratings at one-minute intervals. You can do it (although it may require a bit of extra practice)! After you add a new subject for Sample 49, reset the Run Configuration screen as follows:

  • Run Length: 60 seconds for Samples 49-51 (120 seconds for Samples 52-55.)
  • Display Timer:
  • Start/Stop Tones:
  • Display Syllable Counts: Continuously
  • Display Stutter Event Counts: Continuously
  • Rate Naturalness:
  • Rating Period (sec): type in 60
  • Tone Prompt:
  • Display Naturalness Rating: Continuously
  • Click OK.

Now you're ready to put it all together! As soon as the sample speaker appears on the monitor, depress the space bar to start the SMS and immediately begin counting syllables and stutters. Concurrently, keep in mind your judgment regarding how natural the speech is sounding so that when the naturalness signal is presented you can quickly touch the number that describes the previous minute of speech. NOTE: The clock and the speaker continue during this time, so it is imperative to be prepared to depress the appropriate number key immediately and then return to counting syllables and stutters. Sometimes it's necessary to catch up on any syllables or stutters that were missed while you were making your naturalness rating.

At the completion of each sample, check your ratings with those of the summary data sheet (located at the end of these instructions for Part Four). If your data are not within the acceptable range in any category (SPM, %SS, or Naturalness), repeat the sample until your data are satisfactory.

Continue practicing the simultaneous recording of syllables, stutters, and naturalness in this fashion for the remainder of Part Four. When you're ready, begin Part Five, the Criterion Test.