How can we measure, represent, analyze, and compare instances of sonic eloquence? Thanks to Eddie Lee at the Wisconsin Institute for Discovery's Center for Complexity and Collective Computation, we have one possibility. With a tool called Sound Arguments we can represent audio clips in multiple ways in an attempt to understand the persuasive dimensions of sound. This tool is still in its very early (alpha) stages, and Eddie and I plan to continue work on it (and to add functionality). However, even this early version of Sound Arguments helps us think about analyzing and deploying sonic eloquence.
Sound Arguments is written in the programming language R, and it offers multiple ways of representing sound. Currently, Sound Arguments can provide a visual representation of a clip and can also translate a clip into a series of tones, in effect translating any audio clip into music.
For instance, here is a clip from a speech delivered that Huey Long delivered in 1935:
This clip comes from Americanrhetoric.com, and it is used to demonstrate the rhetorical figure of hypophora: "[A] figure of reasoning in which one or more questions is/are asked and then answered, often at length, by one and the same speaker; raising and responding to one's own question(s)." The site has an entire section called "figures in sound" that provides a nice starting point for analyzing sonic eloquence. In addition to listening to these clips and attuning ourselves to how sound is participating in various rhetorical events, we can use Sound Arguments to carry out computational analysis of American Rhetoric's MP3 files.
Sound Arguments can generate a spectrogram of the Huey Long clip:
The lightest portions of this spectrogram (the white sections) indicate the dominant frequency at a given moment in time. Obviously, the image on its own might not tell us much (and it is perhaps only useful to rhetoricians if we have experts alongside us, helping us read and interpret). But by comparing spectrograms, we can begin to track differences among different sound samples. Spectrograms offer us another way of examining sonic eloquence in action.
The Sound Arguments tool can also translate the Huey Long clip into a series of tones:
This clip is a bit noisy, so it's difficult to discern the "song" that might be at work here, and this indicates one of the limits of the Sound Arguments tool. But what it we tried this on some cleaner clips? In the "Musical Language" section of this chapter, I discussed how looping audio can reveal the musicality of language. That discussion included two examples, a clip from Diana Deutsch, who studies the psychology of sound, and another clip of Johnny Cochran from the O.J. Simpson Trial. These are shorter clips, and they provide better samples for Sound Arguments.
Here is Diana Deutsch saying "sometimes behaves so strangely":
"Sometimes behaves so strangely"
And here is what Sound Arguments does with that clip:
"Sometimes behaves so strangely" translated into tones
Finally, here are the two clips over overlaid:
"Sometimes behaves so strangely." Sound Arguments' tones overlaid with the original clip.
You can here a song emerge here, indicating the sonic eloquence of Duetch's phrase. But what about a phrase that's more aimed at persuasion? What might Sound Arguments reveal? Here is Johnny Cochran's famous sentence from the O.J. Simpson trial: "It it doesn't fit, you must acquit."
"If it doesn't fit, you must acquit."
And here is what Sound Arguments does with that clip:
"If it doesn't fit, you must acquit" translated into tones.
Here are the two clips over overlaid:
"If it doesn't fit, you must acquit." Sound Arguments' tones overlaid with the original clip.
Like the Huey Long clip, this one has a good bit of noise, and Sound Arguments is picking up some of that noise. This once again suggests that sonic eloquence is not confined to the use of human voice. What other sounds in the courtroom are part of the sonic-rhetorical landscape?
Even in its current "alpha" stage, Sound Arguments shows us the sonic eloquence of this clip. There is a song inside of Cochran's words. While it is not nearly as evident as the song that Deutsch is "singing," it still helps us think about how sonic eloquence might have been at work during the closing days of the Simpson trial. These audio clips translate spoken words (or any other sound clip) into a series of notes, and they present us with the opportunity to hear music in unexpected places. What songs or refrains emerge from the spoken word when translated in this way?
One way of using Sound Arguments to conduct rhetorical analysis would be to compare the spectrograms and tones it generates for different speeches by the same orator. Alternately, Sound Arguments could compare orators to one another. In addition, we might conduct this kind of analysis across a larger speech corpus. For instance, we might analyze speeches made during presidential debates, paying attention to how frequencies change when different topics or audiences are addressed. But this tool need not only be used for rhetorical analysis. Sound Arguments could also be used as a tool for generating other digital objects. One could use this tool to remix other audio files, translating sounds into a series of tones and then using the output in an audio composition.
Instructions for downloading and running Sound Arguments
Sound Arguments version 0.1 currently requires the user to download R or RStudio. However, a future version will use an R package called Shiny that allows programmers to use R to create Web applications.
[Download Sound Arguments 0.1] (README.txt contains instructions for installing and running)