عوامل موثر بر دسترسی واژگانی در نام پریشی خالص

Previous studies involving lexical access in anomic patients had employed regression analyses to separate the lexical variables. But it is now known that when the relevant variables are as closely related as frequency, age of acquisition (AoA), and familiarity, regression analysis loses efficacy. The aim of the present paper was to investigate which of these variables determines lexical access in anomic patients. In order to disentangle the effects of frequency from those of AoA, we conducted two picture-naming studies. In the first study, we used a semifactorial design in which we manipulated one variable and matched the other. In the second, we used an orthogonal factorial design in which the two variables were crossed to detect any interaction between them. The participants were pure anomic patients whose disorders affected lexical access exclusively; they had neither semantic nor phonological deficits. The statistical analyses show a clear effect of AoA, no effect of frequency, and no interaction between the two variables. The results of this study indicate that AoA is the main determinant of lexical access in anomic patients, and this must be taken into account in constructing materials for the assessment and rehabilitation of these patients.

The question of which variables determine lexical access during speech has received considerable interest in recent years. Words, generally, seem to come to mind effortlessly, but sometimes it is so costly to access a word that a speaker may spend minutes or even hours searching for it (the tip-of-the-tongue phenomenon) even when the meaning he or she wishes to convey is very clear. Why can some words be accessed so quickly and automatically, whereas others require an effortful mental search? In investigations of the variables that determine lexical access, the methodology most often used is that of picture-naming, in which pictures of objects are presented on a computer screen and the time participants take to name them is measured. By manipulating the characteristics of words, we can discover which ones produce the longest naming times. Through this methodology, it has been confirmed that variables such as frequency of use and the age at which words are learned are extremely influential: less frequent words and those acquired later in life produce longer latencies (Barry et al., 1997, Cuetos et al., 1999 and Ellis and Morrison, 1998). The variables that determine lexical access can also be investigated with aphasic patients. Whereas in healthy persons, the manipulation of a particular variable may lead to differences of a few milliseconds in a picture-naming task, in aphasic patients the same manipulation may mean the difference between correct naming of the picture and the inability to name it. Studies with aphasic patients not only have the advantage of providing information about the organization and functioning of the language-processing system both in normal conditions and with one of the components altered; but they also provide valuable information about the linguistic problems of such patients. What types of words do aphasic patients access easily, and what types are less accessible? For some time, frequency was considered the principal determinant of lexical access for these patients. The first study that uncovered this relationship was that conducted by Newcombe, Oldfield, and Wingfield (1965), who observed an inverse linear relation between frequency of use and the time patients took to name objects (the higher the frequency, the shorter the naming time). This relationship was also demonstrated in later studies, in which patients' correct responses, rather than their reaction times, functioned as the dependent variable: the higher the frequency of use, the more correct responses the patients produced (Butterworth et al., 1984, Howard et al., 1985 and Kay and Ellis, 1987). However, many of the effects attributed to frequency were actually due to other, closely related variables that were not taken into account, such as familiarity, imageability, and length. All these variables are so closely interrelated (i.e. the most frequent words are also learned earliest in life, are the shortest and most concrete, and correspond to the most familiar objects) that it was not clear that it was frequency, and not one of the other variables, that accounted for the results. One variable that has proven especially relevant to picture-naming in the reaction times of both healthy people and aphasic patients is age of acquisition (AoA): words learned early in life are more accessible than those learned late (Carroll and White, 1973 and Morrison et al., 1992). In fact, the effects of AoA on naming in aphasic patients were reported even before the effects of frequency. More than 40-years-ago, Rochford and Williams (1962) confirmed that the age at which children can name an object predicts the proportion of aphasic patients who can name it. However, frequency was not controlled in that study, so the opposite problem of that pointed out above could occur—that is, an effect attributed to AoA might actually correspond to familiarity, frequency, or some other variable. The effects of each of the variables had to be disentangled to determine the actual influence of each one on patient response. Along these lines, Feyereisen, van der Borght, and Seron (1988) conducted a regression analysis on the results of a task in which 18 aphasic patients had to name 64 pictures. These authors found that both AoA and frequency predicted patient performance, although the effects of the former were considerably greater than those of the latter. The problem with studies of groups of patients is the large number of individual differences that may be present, depending on which cognitive process has been damaged. Correct naming of a picture of an object requires a good number of cognitive operations (such as recognizing the object represented in the picture, activating its meaning, accessing its name, and articulating each of its phonemes), all of which must be performed adequately. Consequently, a patient's difficulties, as well as the variables that determine his or her performance, will vary depending on the process that is damaged. Therefore, the results of the group may not be applicable to the individual patients who make up that group. Some studies have confirmed this hypothesis. Nickels and Howard (1995) conducted two studies, the first with 12 aphasic patients who performed a naming task on 104 pictures five different times. The second involved 15 patients who were to name the objects depicted in 130 pictures a single time. In both studies, AoA emerged as the main predictor for the group. However, in the first study length and operativity (i.e. how manipulable the objects are, in the sense that a hammer is very manipulable and a cloud is very unmanipulable) were also significant, and in the second imageability and length were also significant. Frequency was not significant in either of the studies. When the authors conducted regression analyses independently for each patient, they found that, individually, none of them showed effects of the variables that had predicted the overall performance of the group. In the first study, AoA was significant in five of the patients and frequency was significant in only one of them. In the second study, AoA was significant in three of the patients (although in the correlation analysis it was significant in 13 of the 15 patients); frequency, in contrast, never reached significance. In another study (Ellis, Lum, & Lambon Ralph, 1996), six patients were given the task of naming 139 pictures in three different sessions. Again, AoA was the variable that contributed most to patient performance. In fact, AoA and imageability were the only variables that proved to be significant in the regression analysis conducted on the group as a whole; none of the other variables (familiarity, frequency, and length) was significant. However, in the regression analysis on each individual patient, none showed the effects of the two variables that were significant for the group as a whole: imageability was not significant for any individual patient, AoA was significant in three of the patients, familiarity in two, and frequency in one. We found similar results in a similar study of Spanish aphasics (Cuetos, Aguado, Izura, & Ellis, 2002). In a picture-naming task, we presented 140 pictures to 16 patients in three different sessions, separated by 1-week intervals. As in the previous studies, we found that the principal determinant of group performance was AoA, although object familiarity, name frequency, and visual complexity also had significant effects. In the regression analysis for each of the 16 patients individually, we found that only one patient showed effects of the four variables that were significant for the group as a whole. Individually, AoA had a significant effect in ten of the patients, familiarity and frequency in six, visual complexity in five, word length in three, and imageability in two. This heterogeneity among patients is probably due to the fact that patients who participate in these studies suffer from a wide variety of disorders. Although they all have difficulty naming pictures, the causes of these difficulties may have very different origins. It would be worthwhile to use more homogeneous groups of patients. Nickels and Howard (1995) divided their patients on the basis of fluency (fluent vs. nonfluent), but it would be more informative to divide them as a function of which process has been damaged (e.g. semantic, lexical, phonological) and to identify the determinant variables of performance in each case. Thus, to study the effects of the lexical variables, one would need to recruit aphasic patients whose problem resides exclusively in lexical access, with neither a semantic nor a phonological component—that is, pure anomic patients. Likewise, to investigate the effects of semantic variables, patients with semantic disorders should be recruited, and to investigate variables that function at the phonemic level subjects with phonological disorders would be studied. On the other hand, all of the studies described used regression analysis as a way of isolating the different variables. Regression analysis is certainly a good tool for obtaining preliminary information about the different roles played by a series of variables in patient performance, but to obtain more conclusive data factorial designs must be used that allow each variable to be separated from the others. This is especially true when the variables in question are as interrelated as frequency, familiarity, imageability, AoA, and the like. As Ellis et al. (1996) wrote, “when a set of candidate variables has been identified, their influence should be corroborated using sets of items which manipulate those factors factorially” (p. 174). In this study, we wanted to focus on a very specific and important process—namely, access to lexical representations during oral production. With this goal in mind, we selected patients who had specific problems with this process: so-called pure anomic patients or lexical-selection anomia patients, who do not have semantic problems (they know very well what they mean to say) or phonological problems (they access and pronounce words perfectly); their only problem lies in finding the right word for the meaning they wish to express. They most certainly have problems with lexical access (Kay & Ellis, 1987). There are several variables that could influence this process: frequency of use, AoA, familiarity, imageability, and length. Nevertheless, judging from the previous studies that we have described, the two that seem to have an effect on lexical access are frequency and AoA. Familiarity, defined as one's contact with objects or the use one makes of them, is clearly a semantic variable, since its influence is greater in those whose semantic system is damaged, such as semantic dementia patients (Lambon Ralph, Graham, Ellis, & Hodges, 1998). Imageability, defined as the ability to evoke an image of the concept represented by a word, is considered a semantic variable (Howard & Franklin, 1988), since its value depends on semantic features (Plaut & Shallice, 1993); however, Franklin, Howard, and Patterson (1994) affirmed that the effect of imageability may not originate solely in the semantic system, but may also occur within the input or output lexicons. Along these lines, Franklin, Howard, and Paterson (1995) described a pure anomic patient whose problems were limited to abstract words. However, in previous studies we have seen that imageability had few effects on naming in aphasic patients (Cuetos et al., 2002). Finally, length is considered a phonological variable (Alario et al., 2004). In contrast, most authors consider frequency and AoA to have an effect on lexical access (Alario et al., 2004 and Levelt et al., 1999), although some researchers suggest that AoA effects could be localized prior to the lexicon (Nickels & Howard, 1995) or in the retrieval of phonological word form (Hirsh & Ellis, 1994;Kremin et al., 2003). To test the effects of these two variables (AoA and frequency), we used semifactorial and factorial designs instead of regression analysis.