نامگذاری زبان پریشی در اسپانیایی: پیش بینی و اشتباهات

Sixteen Spanish aphasic patients named drawings of objects on three occasions. Multiple regression analyses were carried out on the naming accuracy scores. For the patient group as a whole, naming was affected by visual complexity, object familiarity, age of acquisition, and word frequency. The combination of variables predicted naming accuracy in 15 of the 16 individual patients. Age of acquisition, word frequency, and object familiarity predicted performance in the greatest number of patients, while visual complexity, imageability, animacy, and length all affected performance in at least two patients. High proportions of semantic and phonological errors to particular objects were associated with objects having early learned names while high proportions of no-response errors were associated with low familiarity and low visual complexity. It is suggested that visual complexity and object familiarity affect the ease of object recognition while word frequency affects name retrieval. Age of acquisition may affect both stages, accounting for its influence in patients with a range of different patterns of disorder.

Problems in word retrieval and production are common in aphasic patients. In a task like naming pictures of objects (confrontation naming), most patients will, however, be able to name some items correctly while failing on others. In addition, there is usually a degree of consistency concerning the particular items which a patient can or cannot name. Under such circumstances, some insights into the nature and causes of aphasic naming breakdown might be gleaned by examining which properties of the objects or their names predict success or failure, or the type of errors made to different items (Nickels, 1997; Williams, 1983). Such analyses can be performed at the level of groups of patients or individual cases. Feyereisen, Van der Borght, and Seron (1988) analysed the naming performance of 18 French-speaking aphasic patients attempting to name 64 black-and-white line drawings of familiar objects. Included among Feyereisen et al.'s (1988) analyses were multiple regressions in which a number of properties of the pictures and their names were used as predictors of the accuracy with which items could be named by the group of aphasics as a whole. The dependent variable was a score out of 18 for each item (the number of patients able to name that item correctly). The predictor variables included age of acquisition and word frequency. Feyereisen et al. (1988) found that for their group as a whole, age of acquisition was the most powerful predictor of naming accuracy, while word frequency made a smaller but still significant contribution. Thus, the items that Feyereisen et al.'s (1988) patients were best able to name were those whose names are acquired early in childhood and encountered or used frequently in adulthood. The age of acquisition measure used by Feyereisen et al. (1988) was based on adult estimates of the age at which children learn different object names. Such estimates have been shown to correlate well with objective data on vocabulary acquisition (Carroll & White, 1973; Gilhooly & Gilhooly, 1980; Morrison, Chappell, & Ellis, 1997). Rochford and Williams (1962) and Spreen and Benton (1967; cited in Spreen, 1968) had previously suggested that age of acquisition might affect naming accuracy, but neither of those studies had controlled other factors that correlate with age of acquisition, such as word frequency. Conversely, several studies including Butterworth, Howard, and McLoughlin (1984), Howard, Patterson, Franklin, Orchard-Lisle, and Morton (1984), Howes (1964), and Wepman, Bock, Jones, and Van Pelt (1956) had implicated word frequency as a predictor of aphasic naming but had failed to control age of acquisition. Among sets of object names, age of acquisition and word frequency typically correlate .3–.5 (Cuetos, Ellis, & Alvarez, 1999; Morrison et al., 1997), so a manipulation of one of these lexical properties is likely to be confounded by the other unless it is explicitly controlled. Just because the performance of a group of patients is affected by age of acquisition, word frequency, or some other factor does not mean that every patient in the group is affected by that factor. For example, Howard et al. (1984) found a significant effect of word length on the naming performance of a group of aphasic patients but when the patients were analysed individually that effect was found to be significant for only half of the group. The application of the regression approach to analysing the effects of different object and word properties on naming accuracy at the level of single cases was pioneered by Nickels and Howard (1995). In the first of two studies, Nickels and Howard (1995) presented pictures of 104 familiar objects to 12 aphasic patients on five separate occasions. A response was counted as correct if the target name was produced at any time during the attempt at naming the item. Responses that deviated from the target by a single phoneme were also classed as correct. The mean number of correct responses to each item was then correlated with a range of predictor variables. Naming accuracy for the group of 12 patients correlated most highly with age of acquisition (−.518), but significant correlations were also observed with other factors such as familiarity, concreteness, number of phonemes, and imageability. The raw correlations between naming accuracy and Kucera and Francis (1967) word frequency and the visual complexity of the drawings were smaller and not significant. Raw correlations must be viewed with caution in studies of this type because the predictors are themselves intercorrelated. Thus, in Nickels and Howard's (1995) Study 1, age of acquisition correlated −.724 with familiarity, −.431 with imageability, and −.316 with frequency. Under such circumstances, a significant correlation between a predictor variable and naming accuracy may be mediated by their joint intercorrelations with another variable. To determine which of the predictor variables were exerting a genuine, independent influence on naming accuracy, Nickels and Howard conducted a series of multiple regressions. The first was on the accuracy data for the group as a whole. Here, the significant predictors were age of acquisition, operativity (an estimate of the extent to which an object can be manipulated or operated upon), and word length. Further analyses were then conducted using each patient's score on each item (0–5) as the dependent variable. With visual complexity, familiarity, concreteness, and frequency excluded as predictors (on the grounds that they did not predict naming accuracy in any of the patients), six individual aphasics showed significant effects of age of acquisition on their naming accuracy, two showed effects of imageability, and two showed effects of word length. In Study 2 of Nickels and Howard (1995), 15 aphasic patients named 130 object pictures once each. Only perfectly correct initial responses to targets were counted. The highest correlations for the group as a whole were those of naming accuracy with age of acquisition (−.539) and word length (−.563). In a regression analysis, group naming accuracy was significantly predicted by age of acquisition, imageability, and length. Analysis of individual patient data using discriminant analysis, with familiarity, concreteness, and frequency excluded as predictors, found significant independent contributions of word length in nine patients and age of acquisition in six patients. Nickels and Howard (1995) suggested that age of acquisition might influence semantic processing, or the mapping between semantics and the output lexicon, while word length might exert its effect at the level of phonological encoding or articulation. Ellis, Lum, and Lambon Ralph (1996) gave 139 object pictures to six aphasic patients for naming on three separate occasions. Five predictor variables were used to analyse the results. Object familiarity provided an estimate of how often people encounter or think about the various objects in the pictures. In place of the standard adult ratings of age of acquisition, an objective measure was employed that was derived from children's naming of these and other object pictures (Morrison et al., 1997). The word frequency measure employed by Ellis et al. (1996) was the spoken frequency count from the Celex database, which is based on a contemporary corpus of British English (Baayen, Piepenbroack, & Van Rijn, 1993; Sinclair, 1987) and is therefore more appropriate for the analysis of English patient data than the Kucera and Francis (1967) count, which is based on samples of written (not spoken) American English. Imageability and word length completed the set of predictors. For the group of six patients as a whole, age of acquisition correlated most highly with naming accuracy and was the only variable to make a significant independent contribution to predicting naming performance. Combined scores for each item for each patient across the three administrations were analysed by multiple regression. Three of the patients showed significant effects of age of acquisition, two showed effects of imageability, one showed an effect of familiarity, while another patient showed an effect of word frequency. Finally, logistic regression was used to analyse each patient's performance on the separate administrations of the item set. Age of acquisition was again the most effective predictor, but Ellis et al. (1996) noted that at this level the results could be inconsistent across different administrations of the same pictures to an individual patient. They suggested that results combined over a number of administrations of the same items provide a more reliable indication of which factors affect naming performance in individual cases. The present study set out to investigate further the determinants of aphasic naming accuracy in a set of 16 aphasic patients who were all native speakers of Spanish resident in Spain. Each patient named 140 black-and-white line drawings of objects on three separate occasions. After the removal of nine items with polysemic names, 131 items were used in the analyses. We report the results of analyses of naming accuracy for the group as a whole and for individual cases. Data were available for each item regarding the visual complexity of the picture, the familiarity, animacy and imageability of the object, the age of acquisition of the name and its frequency in adult language sample, and the number of syllables in the name. The measures of visual complexity, familiarity, imageability, and age of acquisition were based on adult ratings of the properties in question. For the key predictor of age of acquisition there is good evidence to suggest that adult ratings are both reliable and valid (Carroll & White, 1973; Gilhooly & Gilhooly, 1980; Morrison et al., 1997). Word frequencies were taken from the Alameda and Cuetos (1995) dictionary of word frequencies. The animacy variable distinguished living from nonliving objects. Dissociations between performance on living and nonliving items have aroused a great deal of theoretical interest (Caramazza & Shelton, 1998; Lambon Ralph, Howard, Nightingale, & Ellis, 1998b), but also a considerable amount of methodological discussion and criticism. If other factors are not controlled, then the living items in picture naming tests, which will include a number of wild and farm animals, tend to be objects that are encountered less often in daily life, and therefore rated as less familiar than the nonliving things, which include household objects and types of vehicle seen on a regular basis (Funnell & Sheridan, 1992; Howard, Best, Bruce, & Gatehouse, 1995). Those differences in average familiarity may be reflected in differences in frequency of the object names, etc. The advantage of the present method is that if animacy emerges as a significant predictor of naming performance on a group or individual basis, it is in the context of statistical control of visual complexity, familiarity, imageability, age of acquisition, word frequency, and word length (cf. Howard et al., 1995). A study of object naming speed in normal young adults by Cuetos et al. (1999) employed the same items that were used in the present study. That study found object naming speed to be predicted by age of acquisition and word frequency, along with object familiarity and number of syllables. Age of acquisition and word frequency have also been found to be significant predictors of naming speed in other studies of object naming in normal English-speaking adults (e.g., Barry, Morrison, & Ellis, 1997; Ellis & Morrison, 1998; Snodgrass & Yuditsky, 1996). Object familiarity was also significant in some of the analyses reported by Ellis and Morrison (1998). Effects of word length tend not to have been observed in studies of English object naming: Cuetos et al. (1999) suggested that this may be because Spanish object names cover a wider range of syllable lengths in a more even manner, making an effect of length easier to detect in Spanish.

The three most powerful determinants of aphasic naming accuracy revealed in the present study were age of acquisition, object familiarity, and word frequency. We would associate object familiarity (and visual complexity) with the ease of recognising pictured objects (that is, with the ease or difficulty with which their semantic representations can be accessed from the visual features available in the picture). We would associate word frequency effects with the ease or difficulty of accessing spoken word forms from semantic representations. We would also associate age of acquisition with accessing spoken word forms, but would suggest that it might also affect the level of object recognition, with early learned objects being easier to recognise as well as to name. If that were the case, it might help explain the ubiquity of age of acquisition effects in our patient group and in the literature generally. The effect of imageability seen in two patients may also be to do with more imageable objects having richer semantic representations. Effects of visual complexity may arise because greater image detail means activation of more visual semantic information and therefore a higher likelihood of the object being recognised. An animacy effect (better naming of nonliving than living things) was seen in two patients whose category-specific deficits were apparent even with object familiarity, age of acquisition, word frequency, and other factors controlled. Finally, a subset of patients showed independent effects of word length, which we are inclined to attribute to phonological output problems.