Angela Weech and Dr. J. Halvor Clegg, Spanish and Portuguese
Even among linguists, there exists some disagreement as to the contribution that various acoustic parameters make or don’t make to the perception of Spanish intonation patterns. The three parameters commonly in question are tone, intensity, and duration. While it is common to only include tone in the definition of intonation, this project adopts the idea that “the multitude of functions intonation actually has in communication requires more than just pitch variation for its fulfilment.” I designed this project to focus on the parameters of tone and intensity in order to test the hypothesis that the perception of intonation patterns in Spanish is not based on tone or intensity exclusively, but rather can be determined by either of these two acoustic parameters, depending on the utterance.
The first phase of this project was to digitize and clean speech samples from a collection of extemporaneous interviews conducted with educated informants from Santiago, Chile, in the early 1970s, as part of the Habla Culta project. Then I isolated a set of utterances and derived pitch and energy contours for each utterance using Kay Elemetrics instrumentation. The pitch contours show how the frequency or tone of the speech varies in time, and the energy contours show how the amplitude or volume varies in time. Often times, there is a positive correlation between the pitch contour and the energy contour (see Figure 1), which can be explained by the functional relationship between the subglottal air pressure and the vibration frequency of the vocal chords; 2 however, this is not always the case. After reviewing the contours, I specifically identified utterances whose tone and intensity patterns differed (see Figure 2). For example, the intensity may fall, indicating a cadence, at the same time that the tone rises, indicating an anticadence, or remains unchanged, indicating a suspension.
In the second phase of my project, I met individually with eight native Spanish speakers in order to determine which parameter would act as the primary indicator of the perceived intonation patterns for the utterances whose tone and intensity patterns didn’t correlate. The native speakers were asked to listen to WAV files of the Chilean utterances and indicate whether they perceived a cadence, an anticadence, or a suspension by circling their response on a questionnaire. There were thirteen utterances in all. The first three utterances were an anticadence, a suspension, and a cadence, respectively, in which the tone and intensity patterns positively correlated. These were used to train each informant to understand the notation used to represent each intonation pattern on the questionnaire. The remaining ten utterances were used to collect data. In nine of the ten utterances, the tone and intensity patterns behaved differently. One utterance with a cadence in both tone and intensity was included to act as somewhat of a control. I wanted to be able to identity those informants who did not perceive a cadence, even when both parameters in question indicated such, so I could better determine the validity of the results.
I was surprised to find that out of eight informants, only four perceived a cadence in the control utterance. Two perceived a suspension, and two, an anticadence. Instead of completely discounting the data from the latter four, I analyzed the data using two different approaches, one considering the responses of all eight informants, and one considering only the responses of the four people who perceived the control utterance to be a cadence. Interestingly, the results didn’t differ much depending on the analysis approach. When considering all of the responses, tone acted as the indicator of the perceived intonation pattern 39 percent of the time, and intensity, 28 percent of the time. In the remaining 33 percent of the cases, the response did not correlate with either parameter. When only considering the responses of the informants who perceived a cadence for the control utterance, tone again acted as the indicator 39 percent of the time, intensity, 22 percent of the time, and the response didn’t correlate with either 39 percent of the time.
I don’t place much weight on the exact percentages, considering they come from a maximum of eighty responses, but I do find them to demonstrate an important point. The perception of Spanish intonation patterns was not based solely on tone or intensity, but rather depended on the utterance, as my hypothesis stated. However, this is not a conclusive study. I view this project as a pilot study for something that I would like to investigate more in depth for my Master’s thesis.