معناشناختی در مقابل تکنیک های خاص واژه فرم در درمان نام پریشی : مطالعه تک مورد چندگانه

This study compared a semantic and a phonological/orthographic approach to the treatment of word-finding difficulties in a case-series of ten individuals with aphasia, using a cross-over design. The study aims to investigate whether one approach is generally more effective than the other or whether the effectiveness of the two treatments relates systematically to the nature of the underlying functional impairment within the framework of a modular single-word processing model. In both treatments, the main task was spoken naming of pictured objects with different types of cues. In the semantic approach, different aspects of the target semantic concept were used as a cue in picture naming, while in the word-form method, both phonological and orthographic information were provided as a cue. Treatment effects were assessed in terms of both short- and long-lasting effects on spoken picture naming accuracy in each participant after the end of each treatment phase. Here, both item-specific effects and a possible generalisation to untreated pictures were considered. In addition, the immediate effects of the phonological and the semantic prompts were analysed. With regard to the cue effects on immediate naming, the word-form specific cues proved stronger than the semantic cues. The semantic treatment phase on the other hand, produced more stable effects than the word-form specific phase in some participants. A direct relationship between specific treatment effects and underlying functional deficit pattern was not confirmed for all subjects, i.e. participants with post-semantic anomia benefited from the semantic approach and participants with semantic anomia benefited from the phonological/orthographic approach. In the discussion of the results different explanations are considered, including the importance of preserved functions which aphasic participants bring into the treatment, the possible acquisition of a conscious strategy, and the possible influence of order of treatment. The effects of the two treatments are interpreted with regard to their underlying functional mechanisms in a single-word processing model.

Word-finding difficulties are a common symptom in aphasia (Goodglass and Wingfield, 1997 and Nickels, 1997). It is generally known that they can result from different underlying functional impairments. In “semantic anomia” word-retrieval deficits reflect incorrect, incomplete, or underspecified semantic representations (Hodges et al., 1995 and Howard and Orchard-Lisle, 1984), hence, lexical target entries do not receive enough activation from the semantic system to be retrieved from the output lexicon. Word-finding difficulties can also result from a post-semantic deficit at the level of the phonological (or orthographic) output lexicon itself (“post-semantic anomia”). In this case, semantics may be totally preserved whereas access to the output lexicon (“classical anomia”) (e.g. Lambon Ralph, Sage, & Roberts, 2000) or the lexical representations themselves (“lexical anomia”) (e.g. Howard, 1995) may be impaired (see also Gainotti et al., 1986, Howard and Gatehouse, 2006 and Howard and Orchard-Lisle, 1984). The functional deficit underlying a patient's anomia is not predictable from surface symptoms in spontaneous speech or spoken naming (Nickels, 2003 and Nickels, 2004). For example, central-semantic deficits and deficits at the level of the phonological output lexicon may both surface as semantic paraphasias (Caramazza and Hillis, 1990, Howard and Franklin, 1988 and Nickels, 2004), and patients may also produce semantically related word errors when they consciously try to compensate for their word-retrieval impairment (Goodglass and Wingfield, 1997 and Nickels, 2003). Similarly, phonological errors in naming may also be caused by different functional impairments: They may result from a deficit at the level of the phonological output lexicon, or from a post-lexical deficit at the phonological encoding or the phonological output buffer level (Buckingham, 1992, Buckingham, 1993, Butterworth, 1979, Kohn and Smith, 1994, Nickels, 2003, Nickels and Howard, 1995 and Shallice et al., 2000). In the treatment of word-finding difficulties, tasks focusing on word form (phonological/lexical techniques) or word meaning (semantic techniques) have both been demonstrated to be effective (e.g. Hillis, 1998, Hillis and Caramazza, 1994, Howard et al., 1985a and Nickels and Best, 1996b). However, no single task has been found to be universally effective for all individuals with impaired word retrieval (Nickels, 2002 and Nickels and Best, 1996a). Some authors postulate a direct relationship between the underlying functional deficit and specific effects of a particular treatment task (e.g. Miceli et al., 1996 and Nettleton and Lesser, 1991). Semantic tasks are argued to be effective for individuals with semantic impairment, and phonological tasks for those with impaired retrieval of phonology. However, other authors have found both semantic and phonological techniques to be effective for the same patient (e.g. Hillis, 1998). With regard to the stability of effects, some authors reported longer-lasting effects for semantic than for phonological tasks (e.g. Howard et al., 1985a and Patterson et al., 1983). Generalisation effects also appeared to occur more often after semantic than after phonological treatments (Nickels and Best, 1996a and Nickels and Best, 1996b). However, several authors provided evidence that word-form specific approaches (phonological, orthographic) may result in long-term improvement and may generalise to untreated items as well (e.g. Best et al., 1997, Hillis, 1998, Miceli et al., 1996 and Raymer et al., 1993; see also Howard, 2005). Howard (2005) therefore suggested that the effectiveness of semantic and phonological techniques may be based on a common mechanism. Some authors argue that generalisation effects are more likely to occur in patients with an underlying semantic deficit (and after semantic therapy) than in patients with (post-semantic) lexical access deficits (e.g. Miceli et al., 1996). The reason is that the semantic system and the phonological lexicon differ in representational characteristics and type of processing during retrieval of a word. According to current accounts, the representations of concepts in the semantic system are considered to be distributed and highly interconnected by features overlapping between semantically related concepts, such as dog and cat, which overlap in the features “pet”, “four legs”, “fur”, etc.. Underspecified or impoverished semantic representations might lead to problems in distinguishing between such related concepts and to semantic errors in picture naming. After remediation of a semantic concept, however, naming should improve for treated and semantically related, untreated words, i.e. cat should improve after the treatment of dog, if a relevant distinguishing feature has been restored by therapy, for example that a dog barks. In contrast, no such generalisation should occur at the level of the phonological lexicon as an outcome of lexical treatment, because the access procedures to representations of words in the phonological lexicon are viewed as discrete. Therefore, remediation of the semantic network should result in a generalised improvement in picture naming, at least to semantically related items. In contrast, functional improvements at the level of the phonological output lexicon should result in a better access to the trained lexical entries only (Hillis, 1998, Howard, 2005 and Miceli et al., 1996). Furthermore, the acquisition of a conscious strategy should result in generalised improvement of both trained and untrained words (e.g. Howard, 2005). On the whole, there is still a problem of predicting the effectiveness of semantic and phonological techniques from the underlying functional deficits of a patient. This may be due to a number of reasons. First, the localisation of the underlying functional deficit within the framework of a model may be incomplete. For example, patients with a central-semantic deficit may have an additional, undetected deficit at the level of the phonological output lexicon (e.g. Hillis, 2001) and therefore benefit less from semantic treatment than patients with a purely semantic deficit would (e.g. Drew & Thompson, 1999). Second, both “semantic” and “phonological” approaches are based on heterogeneous techniques which may differ in their mechanisms of action. Some of these techniques are more likely to result in direct improvements in accessing certain lexical entries, for example picture naming with a cueing hierarchy (e.g. Wambaugh et al., 2001), word-to-picture matching with semantic distractors (Marshall, Pound, White-Thomson, & Pring, 1990), or repetition of the target word (Miceli et al., 1996), while other techniques may help patients to acquire a conscious strategy, such as “Semantic Feature Analysis” (SFA) (e.g. Boyle, 2004, Boyle and Coelho, 1995, Coelho et al., 2000 and Law et al., 2006), picture naming with a letter board (e.g. Best et al., 1997), orthographic (self-) cueing (Law et al., 2007 and Nickels, 1992), or word-form specific feature analysis (Bachmann & Lorenz, 2009). Furthermore, most approaches tap into more than one level of processing and may therefore have effects at several processing levels (e.g. Hillis, 1989 and Howard et al., 2006; but for a purely semantic approach see Crofts et al., 2004 and Le Dorze et al., 1994). Another reason why knowledge about the differential effectiveness of model-based treatment methods is still rare is that in most of the studies only one treatment method was investigated, while only few studies have compared the effects of different techniques within the same individuals (but see Crofts et al., 2004, Howard et al., 1985a, Howard et al., 1985b and Lorenz and Nickels, 2007). However, several authors have a more principal concern that current models of language processing may still not be detailed enough to be used as the basis for therapy (Caramazza, 1989, Hillis, 1993, Hillis, 2001, Holland, 1994 and Wilson and Patterson, 1990). The mechanisms underlying the effectiveness of model-based treatment remain unclear because normal processing mechanisms are not sufficiently specified (Caramazza, 1989, Hillis, 1998 and Hillis, 2001). Nevertheless, only a model-based single-case approach seems to convey a better understanding of the underlying mechanisms in normal and impaired language processing and of the underlying mechanisms of effectiveness of one type of treatment in one type of patient (e.g. Best and Nickels, 2000 and Nickels and Best, 1996b). In the present study we compared a semantic and a phonological/orthographic treatment of anomia in a case-series of ten aphasic participants. In both approaches, the major training task was spoken naming of object pictures with different types of cues. Task structure was the same in both treatments, but types of naming cues differed. In the word-form specific method, partial information of the target word was given as a cue in picture naming, while in the semantic approach partial information about the target semantic concept was provided. We predicted that the semantic treatment would result in an extra boost at the semantic level, whereas the phonological treatment would especially strengthen lexical–phonological processing in the subjects. For participants with semantic deficits we therefore predicted an advantage of the semantic approach over the phonological approach, whereas the phonological treatment was expected to be particularly effective in participants with post-semantic anomia. We further predicted that generalisation effects (improved naming of untrained stimuli) would occur in participants with semantic anomia after semantic treatment, while no generalisation was expected to occur after phonological treatment and for participants with post-semantic anomia. Furthermore, we expected generalisation effects to occur in participants who acquire a conscious strategy to facilitate word retrieval.

The phonological treatment was highly effective in producing item-specific improvements in picture naming in most of the participants, but the effects were only short-lasting in most cases. In contrast, the semantic treatment produced item-specific effects in a smaller number of participants, but the effects were more stable. Furthermore, after the first two phases, no evidence for specific generalisation to untreated pictures was present in any of the subjects. A clear relationship between type of functional deficit and specific treatment effects was not observable. Some of the participants benefited from both methods, irrespective of their underlying functional deficit, which was possibly due to the fact that both approaches can be effective at different levels of processing or that both treatments may have overlapping functional effects. However, the approaches were not similarly effective in every participant. Preserved conceptual-semantic processing and fluency of spontaneous speech seemed to have made it more likely for the semantic treatment to become effective. For phonological therapy, the ability to profit immediately from word-form specific cues seemed to predict treatment effectiveness. Therefore, not only the impaired but also the preserved functions have to be considered in predictions of the effectiveness of phonological or semantic treatments.