Page 28 - Summer 2021

LANGUAGE ENDANGERMENT
spectrographs to a magnifying glass that bring aspects of speech sounds into sharp focus. They make it possible for speech sounds to be examined in three dimensions: fre- quency, intensity, and duration.
Africa and the Preservation of the World’s Phonetic Diversity
Now that the UN has declared IDIL 22-32, greater efforts and monetary investments should be made for the docu- mentation of all critically endangered languages. African languages should receive considerable attention for two reasons. First, critically endangered languages in Africa are underdocumented. Kandybowicz and Torrence (2017) mention 308 such languages. Sadly, they report that 201 African languages have gone extinct without ever being documented. This is a huge loss for the sci- entific understanding of phonetic diversity because, as Ladefoged (2007) notes, “endangered languages may con- tain even more unusual phenomena.” He opines that they may have as many as 600 consonants and 200 different vowels that are not found in safe languages. The second reason for focusing on African languages is eloquently stated by Clements (2000, p. 123).
“The African continent offers a generous sample of the great diversity of phonological systems to be found in the world’s languages, as well as some original features of its own. African phonological systems range from the relatively simple to the staggeringly complex. Those on the more complex end of the spectrum contain pho- nemic contrasts little known elsewhere in the world, rich patterns of morphophonemic alternations, and intricate tonal and accentual systems, all offering stim- ulating grounds for phonetic and phonological study.”
Brenzinger (2007, p.198) concurs and adds, “With about the third of the world’s languages, the African continent is among the linguistically richest areas on the planet. Thus, a great deal of the future of linguistic diversity in general depends on what is going to happen to African languages.” In other words, if one wants to really document phonetic diversity in the world’s languages, one need not look further than the African languages. For example, the African lan- guages are unrivaled by the quantity and diversity of their stop consonants. Stops are sounds that are made with two articulators coming together to block, albeit momentarily, the free flow of air molecules. There are 22 different types of stops in African languages. Plain stop sounds such as [p, ph, b, t, th, d, k, kh, g] are found in many languages outside of the
African continent, including American English. However, the implosives [ɓ, ɗ, ɠ] (see bit.ly/2L74AhG) and the ejec- tives [pʼ, kʼ, tʼ] (see bit.ly/2YoxtJ4) occur more frequently in
African languages than in languages elsewhere. As for the labiovelars [k͡ p] and [g͡ b] (see bit.ly/36pm8gj) and the clicks [ʘ, ǀ, ǃ, ǁ, ǂ] (see bit.ly/3t6useB), they are almost exclusively found in African languages, except for an occasional sight- ing in the indigenous languages in Papua New Guinea (Hale, 1992). The labiovelars [k͡ p] and [g͡ b] are ubiquitous in most West African languages, whereas clicks are found mostly in a cluster of languages in southern Africa.
A Closer Look at Postulate 6
As noted earlier, most of the documentary efforts of the past three decades or so can be likened to hospice care. The focus is/was clearly on collecting artifacts about dying languages so that when they eventually die, future gen-
erations will still have some information about them. As commendable as these efforts are, documented linguistic artifacts cannot be used to generate novel utterances in the dead or dying language. A paradigm shift in document- ing endangered languages is therefore needed during IDIL 22-32. For this to happen, documentation must undergo a change of mindset from the hospice care mentality to a palliative care mentality. This means that Postulate 6 must undergird future documentation efforts and goals, an idea echoed by Grenoble and Whaley (2006). This approach is currently being experimented on Jeju, (see bit.ly/3j1xPyS), a moribund language spoken in South Korea.
The new mindset under Postulate 6 calls for documenting endangered languages for use in speech-enabled technolo- gies. The very first step in this process is speech digitalization. Rabiner and Schafer (1978) note that when speech is fully digitalized, it can be used for automatic speech recognition; Text-to-Speech; Speech-to-Text; voiced-enabled assistive technologies; digital transmission and storage of speech; speech synthesis; speaker verification and identification; and enhancement of signal quality. The speech digitaliza- tion process that makes these technological applications possible is summarized pictorially by Rabiner and Juang (1993). The methodology consists of extracting, measur- ing, and collecting all relevant features that are to be used to build voice-enabled intelligent systems. Koffi (2020) has amply explained and described how this can be applied to
African languages. An oversimplified demonstration of how one might go about digitalizing a critically endangered lan- guage for speech synthesis is now discussed.
28 Acoustics Today • Summer 2021