J Psycholinguist Res DOI 10.1007/s10936-015-9352-8
Semantic Richness and Aging: The Effect of Number of Features in the Lexical Decision Task Christelle Robert · Liliana Rico Duarte
© Springer Science+Business Media New York 2015
Abstract The aim of this study was to examine whether the effect of semantic richness in visual word recognition (i.e., words with a rich semantic representation are faster to recognize than words with a poorer semantic representation), is changed with aging. Semantic richness was investigated by manipulating the number of features of words (NOF), i.e., the number of characteristics that describe the meaning of words. Half of the words had a high NOF and the other half had a low NOF. Young adults (19.6 years) and older adults (66.3 years) performed a lexical decision task. An interaction was found between age group and NOF on word latencies. More precisely, a facilitatory effect of NOF was observed for the young adults, but not for the older ones. These data are consistent with the assumption of an age-related decline in feedback activation from semantics to orthography. Keywords Semantic richness · Visual word recognition · Feedback activation · Lexical decision task
Introduction Many theories assume that word meaning is represented by semantic features (e.g., McRae et al. 1999). Semantic features refer to attributes or characteristics that describe the meaning of a word. For example, the word “apple” has several features such as “is a fruit”, “has a core”, “has pips”, and so on. To generate features, a first method is to ask people to list what they know about the things to which words refer (McRae et al. 2005). Another method is to use linguistically based definitions of the things and to break down these definitions into individual feature information (Farah and McClelland 1991). Considering semantic C. Robert (B) Laboratoire de Psychologie, Santé et Qualité de Vie EA4139, University of Bordeaux, 3 place de la Victoire, 33076 Bordeaux, France e-mail: [email protected] L. Rico Duarte University of Toulouse, Toulouse, France
123
J Psycholinguist Res
features, the notion of semantic richness refers to the number of features (NOF) a concept has (e.g., Grondin et al. 2009). In fact, some words have a rich semantic representation (e.g., dolphin has 23 features) whereas others have a poorer one (e.g., nightingale has only 6 features). The effects of semantic richness have been investigated in visual word recognition studies conducted with young adults. These studies have shown that the richness of a word’s semantic representation affects its processing. Specifically, words are recognized faster when they have a richer semantic representation. For instance, Pexman et al. (2003) found that semantic richness was a significant predictor of decision latencies for concrete objects when participants had to decide whether each word referred to a concrete object or to something abstract. In the same vein, the facilitatory effect of concreteness reported in naming tasks (e.g., Strain et al. 1995) and lexical decision tasks (e.g., Binder et al. 2005) has been ascribed to richer semantic representations for concrete words. A facilitatory effect of semantic richness was also reported for concrete words when the NOF was manipulated (Grondin et al. 2009). Concrete words with a high NOF were recognized faster than those with a lower NOF. Pexman et al. (2002) argued that the effect of semantic richness can be explained within an interactive-activation framework of word recognition, i.e., the effect of semantic richness may be ascribed to feedback activation from semantic units to orthographic units. In fact, many-toone mappings from semantics to orthography for words with a high NOF lead to a more rapid accumulation of activation at the orthographic level for these words. Since responses in visual word recognition tasks are mainly based on the activation of orthographic units, words with a high NOF should be recognized faster than those with a low NOF. Also, by manipulating the polysemy of words, Pecher (2001) found that visual word processing is influenced by feedback semantics. The author further showed that the top-down semantic effect was stronger in lexical decision than in naming. Lexical decision may require more complete recognition of a word than the naming task. The present research therefore addressed the issue of an eventual age-related change in the efficiency of semantic-to-orthography feedback activation by examining the magnitude of the effect of NOF in a lexical decision task performed by young and older adults. In the aging literature, some authors have suggested that age-related changes in language performance are caused by a failure of activation from semantic units to phonological units on the part of older adults (e.g., Burke et al. 2000). Empirical evidence for a transmission deficit from semantics to phonology and/or orthography has been provided when investigating the tip-of-the-tongue (TOT) phenomenon (Burke et al. 2000). In fact, a TOT is experienced when the person knows the definition, grammatical characteristics (Miozzo and Caramazza 1997) and sometimes some sounds of the target word (Burke et al. 1991) but is not able to produce this word. According to the transmission deficit theory, a TOT is due to the fact that semantic units corresponding to the word are activated (giving access to semantic activation about the word) but activation of its phonological information is incomplete (preventing its production). Activation of phonological units might fail because the connections from semantics to phonology have become weak, thereby reducing the transmission of activation. Aging is a factor that contributes to the weakening of such connections and might explain the increase in TOTs that older people experience. The transmission deficit theory predicts an age-related decline in orthographic production (i.e., spelling), parallel to the decline in phonological production (i.e., TOTs). Like phonological nodes, orthographic nodes are integrated in a hierarchically organized network with a semantic system. Aging weakens connections, and orthographic nodes like phonological nodes are especially vulnerable to transmission deficits. To test these assumptions, MacKay and Abrams (1998) presented a tape-recorded series of difficult-to-spell English words to young and older adults who had to write the words down. Older adults were found to make more spelling errors than did young adults. The authors
123
J Psycholinguist Res
discussed these results in terms of a transmission deficit hypothesis of orthographic retrieval. They further suggest symetrical effects of aging in written and spoken language production. To our knowledge, the transmission-deficit model of Burke et al. (2000) has never been tested for written language comprehension. Examining whether semantic-to-orthographic activation changes with aging therefore raises the question of whether the transmission-deficit theory developed initially to account for a decline in language production can be extended to word recognition. Up to now, little information has been available on possible age differences in activation transmission efficiency. With regard to the visual word recognition literature, some evidence for a deficit in activation transmission comes from studies that have investigated the effect of word frequency (e.g., Allen et al. 2004; Robert et al. 2009) and orthographic neighborhood (e.g., Robert and Mathey 2007). These studies reported that both the effects of word frequency and orthographic neighborhood were reduced for the elderly, which suggests that bottom-up activation toward orthographic units is less efficient during aging. Given that bottom-up activation toward orthography seems to be altered in older people, it remains to determine whether and to what extent top-down activation toward orthography is also modified in older people. The aim of the present study was to examine whether and to what extent the effect of NOF in the lexical decision task changes with aging. In an interactive-activation framework of visual word recognition, the effect of NOF can be considered as an estimate of the efficiency of activation from semantics to orthography (see Pexman et al. 2002). If we assume that feedback activation efficiency decreases with aging, then the semantic units should send less activation toward the orthographic units of the word in older adults than in younger ones. In other words, the differential processing of words with a high NOF and a low NOF should be less salient with aging. It is therefore expected that older adults would exhibit a lesser effect of NOF than young adults.
Methods Participants A total of 50 adults participated in the experiment. All were native French speakers and reported having normal or corrected-to-normal vision. Twenty-five young adults (age, M = 19.2 years, S D = 1.7, 22 women) were students and averaged 12.1 years of education (S D = 0.6). Twenty-five older adults (age, M = 66.3 years, S D = 6.4, 20 women) were recruited from the senior citizen’s education courses at the University of Bordeaux and averaged 14.3 years of education (SD = 4.6). Their mean score on the Mini-Mental State Examination (Folstein et al. 1975) was 28.5 points out of 30 (SD = 0.9). All participants completed the French version of the Mill Hill vocabulary test (Deltour 1998). Younger adults scored lower on this test (M = 33.5 out of 44, S D = 4.1) than the older adults (M = 39.0 out of 44, S D = 3.1), t (48) = 5.3, p < .001. Also, young and older participants differed on education years, t (48) = 2.3, p < .05. Stimuli To select the word materials, we chose 52 words from a feature database that was constituted with a procedure similar to that used by Farah and McClelland (1991). The database included 72 concepts (Rico Duarte 2004). For each concept, the definition was taken from French dictionaries and broken down into individual feature information. Then, the list of concepts
123
J Psycholinguist Res Table 1 Characteristics of the word materials Word condition Low-NOF
p High-NOF
NOF
14.69 (3.52)
26.69 (7.00)
Semantic Richness and Aging: The Effect of Number of Features in the Lexical Decision Task Christelle Robert · Liliana Rico Duarte
© Springer Science+Business Media New York 2015
Abstract The aim of this study was to examine whether the effect of semantic richness in visual word recognition (i.e., words with a rich semantic representation are faster to recognize than words with a poorer semantic representation), is changed with aging. Semantic richness was investigated by manipulating the number of features of words (NOF), i.e., the number of characteristics that describe the meaning of words. Half of the words had a high NOF and the other half had a low NOF. Young adults (19.6 years) and older adults (66.3 years) performed a lexical decision task. An interaction was found between age group and NOF on word latencies. More precisely, a facilitatory effect of NOF was observed for the young adults, but not for the older ones. These data are consistent with the assumption of an age-related decline in feedback activation from semantics to orthography. Keywords Semantic richness · Visual word recognition · Feedback activation · Lexical decision task
Introduction Many theories assume that word meaning is represented by semantic features (e.g., McRae et al. 1999). Semantic features refer to attributes or characteristics that describe the meaning of a word. For example, the word “apple” has several features such as “is a fruit”, “has a core”, “has pips”, and so on. To generate features, a first method is to ask people to list what they know about the things to which words refer (McRae et al. 2005). Another method is to use linguistically based definitions of the things and to break down these definitions into individual feature information (Farah and McClelland 1991). Considering semantic C. Robert (B) Laboratoire de Psychologie, Santé et Qualité de Vie EA4139, University of Bordeaux, 3 place de la Victoire, 33076 Bordeaux, France e-mail: [email protected] L. Rico Duarte University of Toulouse, Toulouse, France
123
J Psycholinguist Res
features, the notion of semantic richness refers to the number of features (NOF) a concept has (e.g., Grondin et al. 2009). In fact, some words have a rich semantic representation (e.g., dolphin has 23 features) whereas others have a poorer one (e.g., nightingale has only 6 features). The effects of semantic richness have been investigated in visual word recognition studies conducted with young adults. These studies have shown that the richness of a word’s semantic representation affects its processing. Specifically, words are recognized faster when they have a richer semantic representation. For instance, Pexman et al. (2003) found that semantic richness was a significant predictor of decision latencies for concrete objects when participants had to decide whether each word referred to a concrete object or to something abstract. In the same vein, the facilitatory effect of concreteness reported in naming tasks (e.g., Strain et al. 1995) and lexical decision tasks (e.g., Binder et al. 2005) has been ascribed to richer semantic representations for concrete words. A facilitatory effect of semantic richness was also reported for concrete words when the NOF was manipulated (Grondin et al. 2009). Concrete words with a high NOF were recognized faster than those with a lower NOF. Pexman et al. (2002) argued that the effect of semantic richness can be explained within an interactive-activation framework of word recognition, i.e., the effect of semantic richness may be ascribed to feedback activation from semantic units to orthographic units. In fact, many-toone mappings from semantics to orthography for words with a high NOF lead to a more rapid accumulation of activation at the orthographic level for these words. Since responses in visual word recognition tasks are mainly based on the activation of orthographic units, words with a high NOF should be recognized faster than those with a low NOF. Also, by manipulating the polysemy of words, Pecher (2001) found that visual word processing is influenced by feedback semantics. The author further showed that the top-down semantic effect was stronger in lexical decision than in naming. Lexical decision may require more complete recognition of a word than the naming task. The present research therefore addressed the issue of an eventual age-related change in the efficiency of semantic-to-orthography feedback activation by examining the magnitude of the effect of NOF in a lexical decision task performed by young and older adults. In the aging literature, some authors have suggested that age-related changes in language performance are caused by a failure of activation from semantic units to phonological units on the part of older adults (e.g., Burke et al. 2000). Empirical evidence for a transmission deficit from semantics to phonology and/or orthography has been provided when investigating the tip-of-the-tongue (TOT) phenomenon (Burke et al. 2000). In fact, a TOT is experienced when the person knows the definition, grammatical characteristics (Miozzo and Caramazza 1997) and sometimes some sounds of the target word (Burke et al. 1991) but is not able to produce this word. According to the transmission deficit theory, a TOT is due to the fact that semantic units corresponding to the word are activated (giving access to semantic activation about the word) but activation of its phonological information is incomplete (preventing its production). Activation of phonological units might fail because the connections from semantics to phonology have become weak, thereby reducing the transmission of activation. Aging is a factor that contributes to the weakening of such connections and might explain the increase in TOTs that older people experience. The transmission deficit theory predicts an age-related decline in orthographic production (i.e., spelling), parallel to the decline in phonological production (i.e., TOTs). Like phonological nodes, orthographic nodes are integrated in a hierarchically organized network with a semantic system. Aging weakens connections, and orthographic nodes like phonological nodes are especially vulnerable to transmission deficits. To test these assumptions, MacKay and Abrams (1998) presented a tape-recorded series of difficult-to-spell English words to young and older adults who had to write the words down. Older adults were found to make more spelling errors than did young adults. The authors
123
J Psycholinguist Res
discussed these results in terms of a transmission deficit hypothesis of orthographic retrieval. They further suggest symetrical effects of aging in written and spoken language production. To our knowledge, the transmission-deficit model of Burke et al. (2000) has never been tested for written language comprehension. Examining whether semantic-to-orthographic activation changes with aging therefore raises the question of whether the transmission-deficit theory developed initially to account for a decline in language production can be extended to word recognition. Up to now, little information has been available on possible age differences in activation transmission efficiency. With regard to the visual word recognition literature, some evidence for a deficit in activation transmission comes from studies that have investigated the effect of word frequency (e.g., Allen et al. 2004; Robert et al. 2009) and orthographic neighborhood (e.g., Robert and Mathey 2007). These studies reported that both the effects of word frequency and orthographic neighborhood were reduced for the elderly, which suggests that bottom-up activation toward orthographic units is less efficient during aging. Given that bottom-up activation toward orthography seems to be altered in older people, it remains to determine whether and to what extent top-down activation toward orthography is also modified in older people. The aim of the present study was to examine whether and to what extent the effect of NOF in the lexical decision task changes with aging. In an interactive-activation framework of visual word recognition, the effect of NOF can be considered as an estimate of the efficiency of activation from semantics to orthography (see Pexman et al. 2002). If we assume that feedback activation efficiency decreases with aging, then the semantic units should send less activation toward the orthographic units of the word in older adults than in younger ones. In other words, the differential processing of words with a high NOF and a low NOF should be less salient with aging. It is therefore expected that older adults would exhibit a lesser effect of NOF than young adults.
Methods Participants A total of 50 adults participated in the experiment. All were native French speakers and reported having normal or corrected-to-normal vision. Twenty-five young adults (age, M = 19.2 years, S D = 1.7, 22 women) were students and averaged 12.1 years of education (S D = 0.6). Twenty-five older adults (age, M = 66.3 years, S D = 6.4, 20 women) were recruited from the senior citizen’s education courses at the University of Bordeaux and averaged 14.3 years of education (SD = 4.6). Their mean score on the Mini-Mental State Examination (Folstein et al. 1975) was 28.5 points out of 30 (SD = 0.9). All participants completed the French version of the Mill Hill vocabulary test (Deltour 1998). Younger adults scored lower on this test (M = 33.5 out of 44, S D = 4.1) than the older adults (M = 39.0 out of 44, S D = 3.1), t (48) = 5.3, p < .001. Also, young and older participants differed on education years, t (48) = 2.3, p < .05. Stimuli To select the word materials, we chose 52 words from a feature database that was constituted with a procedure similar to that used by Farah and McClelland (1991). The database included 72 concepts (Rico Duarte 2004). For each concept, the definition was taken from French dictionaries and broken down into individual feature information. Then, the list of concepts
123
J Psycholinguist Res Table 1 Characteristics of the word materials Word condition Low-NOF
p High-NOF
NOF
14.69 (3.52)
26.69 (7.00)
