Age-related effects on speech production: A review

Abstract

In discourse, older adults tend to be more verbose and more disfluent than young adults, especially when the task is difficult and when it places few constraints on the content of the utterance. This may be due to (a) language-specific deficits in planning the content and syntactic structure of utterances or in selecting and retrieving words from the mental lexicon, (b) a general deficit in inhibiting irrelevant information, or (c) the selection of a specific speech style. The possibility that older adults have a deficit in lexical retrieval is supported by the results of picture naming studies, in which older adults have been found to name objects less accurately and more slowly than young adults, and by the results of definition naming studies, in which older adults have been found to experience more tip-of-the-tongue (TOT) states than young adults. The available evidence suggests that these age differences are largely due to weakening of the connections linking word lemmas to phonological word forms, though adults above 70 years of age may have an additional deficit in lemma selection.

Notes

¹Consistent with the TD hypothesis (MacKay & Burke, 1990), we use the term priming to refer to the excitation of a representational unit (i.e., a node) that prepares it for activation. The theoretical mechanism of node priming is different from the empirical process of priming which denotes an increase in the speed and accuracy of responding due to preactivation of the target response. Concerning the empirical process of priming, we distinguish between semantic priming, which refers to the preactivation of a target word's meaning, and phonological priming, which refers to the preactivation of its phonological form.

²Some studies have shown that there are biases in how specific phonemes influence voice onsets and in how reliably voice onsets are detected by a voice key (Kessler, Treiman, & Mullennix, 2002; Rastle & Davis, 2002). These differences might contribute to the presence or absence of age differences in tasks in which utterance onset is the dependent measure. However, as long as young and older speakers produce the same words (i.e., their utterances are matched for onset syllable), the problems related to measuring voice onset do not arise (unless, of course, voice keys react differently to young and older speaker's voices. This may be true, but we do not know whether anyone has studied this). When young and older speakers produce different words, the problems may arise, along with many other problems (e.g., the words may not be matched for frequency, imageability, etc., either).

³Rather than reflecting a failure to retrieve the complete phonological form of a target word, it has been suggested that TOTs occur when speakers activate words associated with the target and these non-target words block or inhibit retrieval of the target (Brown, 1991; Jones, 1989). According to the ID hypothesis, the age-related increase in TOTs reflects that older adults activate more blockers and are more susceptible than young adults to inhibition of target retrieval from these blockers (Brown & Nix, 1996; Zacks & Hasher, 1994). However, because of compelling evidence against this account of TOTs (Brown, 2000; Burke, 1997; Burke et al., 1991; Cross & Burke, 2004; Heine et al., 1999; Maylor, 1990; Rastle & Burke, 1996; Vitevitch & Sommers, 2003), it has largely been abandoned and is not considered here.

⁴Nicholas et al. (1985) did not specify if these increases refer to the number of occasions where only one cue was needed (semantic or phonemic), or if there was an increase also in the need for double-cues.

⁵Thomas et al. (1977) only included the incorrect-prime condition to ensure that in the correct-prime condition, participants did not produce the picture name without recognising the picture, but simply translated the written form of the name into the corresponding phonological form. Thus, they did not analyse the difference in naming latency between the unprimed and the incorrect-prime condition. However, they presented a figure displaying the mean naming latencies in the unprimed, correct-prime, and incorrect-prime conditions for the different age groups. This figure shows that there was a slight increase in the latency in the incorrect-prime condition relative to the unprimed condition for both younger and older adults. This suggests that reading the name of a non-target object interfered with naming of the target object and that this interference was similar for the different age groups. The similarity of the interference for younger and older adults contradicts the ID hypothesis of cognitive ageing.

⁶Rather than reflecting an effect of object name familiarity, the latency difference between unique aged and unique contemporary objects in Poon and Fozard's (1978) study might reflect a difference in the age of acquisition of the different object names. The older adults in this study will have learned the names of the aged objects earlier in life than the names of the contemporary objects, whereas the young adults will have learned the contemporary names earlier than the aged names (see Hodgson & Ellis, 1998).

⁷Conditions in which the relationship between distractor and picture was phonological, mediated phonological, and semantic/phonological were also included. We will return to these conditions in the next section.

⁸For the young participants, the homophone priming effect was absent only for participants who were unaware of the prime-target relationship. When data from aware participants were included in the analysis, a homophone priming effect was found. This suggests that effects of homophone priming in young adults arise because young adults are aware of the prime-target relationship and use this awareness to anticipate the target name.

Additional information

Notes on contributors

Linda Mortensen

Correspondence should be addressed to Linda Mortensen, School of Psychology, University of Birmingham, Birmingham B15 2TT, UK. [email protected]