Page 43 - 2017Spring

Figure 2. Changes in median fundamental frequency across the life span based on data collected during spot-the-difference tasks (diapix) carried out between pairs of talkers. These data have been accumu- lated across studies carried out with children (reported in Hazan et al., 2016) and with young and older adults (reported in Hazan and Tuomainen, 2016). F0, fundamental frequency. Circles and star de- note outliers.
Moving Toward Life Span Studies
The changes in speech production across the life span docu- mented above suggest that we should view speech commu- nication as a highly dynamic process. This process is dy- namic not only because of the ongoing adaptations that are made in communication to adapt to different environments and differing needs of our conversational partners but also because our speech undergoes ongoing adaptation through- out our life.
Currently, our understanding of these life span changes is limited by the lack of studies that span a large age range. Longitudinal studies spanning several decades would be fas- cinating but impractical, although a small number do ex- ist for “‘exceptional” individuals such as Queen Elizabeth II and British radio broadcaster Alistair Cooke (see Figure 3), for whom there are recordings over a 50-year period (Reu- bold et al., 2010), or for groups of individuals who have been recorded at regular interviews throughout their lifetime as in the British “Up” set of documentaries (Gahl et al., 2014). These longitudinal studies reflect changes that result not only from physical aging but also from sociophonetic fac- tors described above, as documented for Alistair Cooke, for example, who changed his accent several times throughout his lifetime (Reubold and Harrington, 2015).
Between-group life span studies, which should be more eas- ily achievable, still involve a number of challenges. First, few
Figure 3. Data showing longitudinal changes in mean fundamental frequency (F0) and mean first formant frequency (F1) for two speak- ers recorded over a 50-year period: Queen Elizabeth II and the British broadcaster Alistair Cooke. Reprinted from Speech Communication, Vol. 52, Reubold U., Harrington, J, and Kleber, F., Vocal aging effects on F0 and the first formant: A longitudinal analysis in adult speakers, 638-651, Copyright 2010, with permission from Elsevier.
tasks and speech materials are useable for both children and adults because factors such as lexical knowledge or working memory demands need to be taken into account. Even when such tasks are found, one needs to consider whether they involve widely differing degrees of cognitive load for par- ticipants of different ages because this could impact speech production. For example, if investigating changes in the ar- ticulation rate across the life span, a task imposing a greater cognitive load for children and older adults than for younger adults could lead to slower articulation rate that is task re- lated. Participant selection criteria, which are already diffi- cult to control within a specific target population, become even more of a challenge across a broad age range due to the wider range of external factors that could influence speech communication. The paucity of standardized cognitive and phonological assessments that are normed across a wide age range is a further limitation. Finally, it is still the case that a majority of researchers within the field of speech sciences who have an interest in the effect of age on speech commu- nication specialize in either development studies or studies into ageing, with few having the practical experience of run- ning studies with different age ranges, which each have spe- cific demands and challenges.
Despite these many obstacles, moving from the currently fairly compartmentalized fields of speech research into development and aging to a life span approach, as already done by a few pio- neers, could result in a greater understanding of both speech production and perception processes and their interaction and would also ultimately result in broader theoretical models of speech communication. Let’s embrace this challenge.
Spring 2017 | Acoustics Today | 41