Applying Phonetics. Murray J. Munro
of perhaps a million words. What's more, it is perfectly possible to use new combinations of phones to invent new words. As far as I know, splenk (/splɛŋk/) is not an English word, but there is nothing to stop an inventor from developing a new household tool and calling it a splenk. And splenk users would realize, without being told, that the plural form of the word is likely splenks!
Linguists use an asterisk (*) to denote a non‐existent word or impossible sound sequence in a particular language.
It is quite easy to pinpoint some of the ways in which non‐human communication systems lack the properties of human language. For instance, the bee dance mentioned earlier expresses the angle of the sun relative to a food source through the angle at which the dance is performed. In that case, there is a connection between the communicative symbol and the thing it stands for, so in that respect bee communication lacks arbitrariness. In a similar vein, there is no evidence of discreteness or productivity in the meowing of a cat; it isn't possible to break a cat vocalization down into smaller pieces and rearrange them to generate new messages.
However, we must not make the mistake of being too categorical in our assessments of human versus non‐human communication systems. In the first place, animal vocalizations are not entirely devoid of speech‐like elements. Vervet monkeys, for example, use different vocal alarm calls to alert other members of their group to imminent dangers. Seyfarth, Cheney, and Marler (1980) studied these vocalizations by recording them, analyzing their acoustic composition, and playing the sounds back through loudspeakers while observing the vervets' behavior. The monkeys produced a low‐pitched, grunt‐like call, for instance, on the approach of an eagle, and a higher‐frequency call at the sight of a snake. When other vervets were exposed to only the recorded calls with no visual stimuli, they responded as if an eagle or snake were present. On the one hand, the alarms apparently aren't divisible into smaller units, and there is no indication that vervets can rearrange the sounds of one call to create another call with a different meaning. Consequently, the vervet vocalizations can be said to lack discreteness and productivity. But, on the other hand, the calls do have the speech‐like property of arbitrariness: there appears to be no connection between the sounds and the things they represent. In that respect, they share something with human speech.
A second observation is that human speech does not fully conform to the design features we've mentioned. For one thing, not all aspects of speech are arbitrary. English, and presumably all other languages, has scores of onomatopoeic words like bang, burp, chirp, and clap, which sound to some degree like the things they refer to. Research has also uncovered intriguing examples of SOUND SYMBOLISM, in which particular speech sounds are associated with certain meanings (Westbury, Hollis, Sidhu, & Pexman, 2018). For instance, in linguistic judgment tasks, people tend to link the sounds /k/ and /t/ to the concept of sharpness, while /m/ and /l/ suggest roundness. These and other non‐arbitrary mappings may be much more than trivial matters. Some evidence indicates that they may facilitate child language acquisition. We will return to this topic when we discuss its applicability in the complexities of product naming in Chapter 14.
1.1.2 technology and our changing understanding of “speech”
Several decades ago, we would have stopped with the four‐way classification of communication types that we have developed so far. But contemporary technology is changing our understanding of the nature of communication. Suppose that an animated character in a movie or a video game displays facial expressions and gestures indicating anger or a threat of violence. Of course, the character itself has no feelings or desire to communicate, but the animator has created a representation of non‐linguistic, non‐vocal communication that is readily grasped by viewers. In a similar vein, computers do not volitionally use speech with an intent to communicate (at least, not at present!). However, we have no problem calling artificially‐created utterances speech, even though they can be generated entirely without a vocal tract. Such utterances have a communicative function, whether the purpose is to give voice to a human user who cannot speak, to “read” a text aloud to a blind person, or to convey an account balance to a bank customer over the phone. We can capture these recent developments by adding a third category to the vertical dimension of the grid to cover synthetic communication types, both non‐linguistic and linguistic.
To sum things up so far, communication refers to the transmission of a message from one organism or entity to another; language is a means of communication that uses arbitrary symbols; and speech consists of communicative sounds produced in the vocal tract or synthetically.
1.2 the sound structure of speech
Figure 1.2 is a visual representation of an English sentence (“The museum hires musicians every evening”), which was generated from speech using an application called Praat (Boersma & Weenink, 2019). Depictions like this are of great use in phonetics research, and we will discuss them in more detail in later chapters. For now, it is enough to know that the top portion is an ACOUSTIC WAVEFORM capturing the oscillations of air particles when a speaker utters something into a microphone. The lower panel is a SPECTROGRAM illustrating the sound frequency components of speech, with lower frequencies at the bottom of the display. Dark regions in the spectrogram indicate concentrations of acoustic energy. Notice that the utterance appears as a variable acoustic pattern with occasional abrupt changes in darkness and shape. However, finding discreteness (as described earlier) in this representation turns out to be quite a challenge. In some instances, it is a straightforward matter to locate the beginning and end of a word, but in others it is much more difficult. Often, it is not possible to find clear demarcations between individual vowels and consonants within the words because these units overlap one another to varying degrees. What this means is that the phenomenon of discreteness is actually not an aspect of the acoustic signal itself. Rather, it is something that we humans partially impose upon the speech stream we hear. Put another way, discreteness is the result of the way we interpret vocally‐produced sound. This apparent lack of a one‐to‐one relationship between sound and perception adds an interesting layer of complexity to our understanding of the nature of speech—one that we will revisit throughout this book.
The APSSEL website provides a link to where you can download Praat and some instructions on getting started.
Figure 1.2 Acoustic waveform (top) and spectrogram (bottom) of “The museum hires musicians every evening,” as produced by an adult female speaker
TRY THIS
☛ Download and install the Praat software on your computer, and record yourself saying the sentence in Figure 1.2. Use the software to display a waveform and spectrogram as in the figure. Compare your own production with the one in the figure.
1.3 phonetics as a field of study
Phonetics focuses on the sounds of language rather than on written forms. Moreover, phoneticians generally accept the primacy of speech over the written modality. Historically, it must have preceded writing because many world languages have a spoken form but no written one, yet we know of no natural language that can be written but not spoken. Another reason for assuming the primacy of speech is that children become highly proficient in oral language well before they are capable of reading and writing. In fact, many people never learn to read or write at all. No one seriously doubts that literacy is an important aspect of human culture, but it is a mistake to regard written language as more important or more “correct” than speech. That simply isn't true. Writing was invented by humans as a way of representing