- Email: [email protected]
Processing of Japanese morphogram and syllabogram in the left basal temporal area: electrical cortical stimulation studies
Cognitive Brain Research 24 (2005) 274 – 283 www.elsevier.com/locate/cogbrainres
Research report
Processing of Japanese morphogram and syllabogram in the left basal temporal area: Electrical cortical stimulation studies Keiko Usuia, Akio Ikedab,T, Motohiro Takayamaa,c, Masao Matsuhashia, Takeshi Satowa,c, Tahamina Beguma, Masako Kinoshitab, Susumu Miyamotoc, Nobuo Hashimotoc, Takashi Nagaminea, Hidenao Fukuyamaa, Hiroshi Shibasakia,b,d a
Human Brain Research Center, Kyoto University Graduate School of Medicine, Shogoin, Sakyo-ku, Kyoto, 606-8507, Japan b Department of Neurology, Kyoto University Graduate School of Medicine, Shogoin, Sakyo-ku, Kyoto, 606-8507, Japan c Department of Neurosurgery, Kyoto University Graduate School of Medicine, Shogoin, Sakyo-ku, Kyoto, 606-8507, Japan d National Institute of Neurological Disorders and Stroke, National Institutes of Health, Bethesda, MD 20892, USA Accepted 1 February 2005 Available online 16 March 2005
Abstract Language functions in the left basal temporal area (LBTA) were investigated using electrical cortical stimulation during functional mapping in six Japanese patients with refractory epilepsy. This study provides the first direct evidence that kana (Japanese syllabogram) is processed in the LBTA. Electrical stimulation of some areas within LBTA induced disturbance in overt reading of kana words only in the first trials, with no errors in the subsequent trials. By contrast, stimulation of the same area caused obvious disturbance in kana non-word reading in all trials. Since a kana word carries both meaning and sound while a kana non-word carries only sounds of a letter string, the contrasting results of partial and complete disturbance imply a possibility that there are two distinct pathways for kana reading: one dealing with both phonological and semantic aspects of the words and the other dealing only with phonological aspect. Kanji words (Japanese morphogram) and objects/pictures were found to be processed in an area different from the area for the kana nonword processing. Furthermore, the present study also identified the common area for processing kanji reading and object/picture naming. There were no errors in matching pictures with kanji words, indicating that concepts of pictures and meanings of kanji words were not interfered by the electrical stimulation of that area. The new insight provides a clue for partial description of processing pathways for language-related visual information in LBTA. Three types of information (morphological, phonological, and semantic) are conveyed together at some stages and are separated into different routes at some other stages. D 2005 Elsevier B.V. All rights reserved. Theme: Neural basis of behavior Topic: Cognition Keywords: Left basal temporal area; Kana (syllabogram) processing; Kanji (morphogram) processing; Electrical cortical stimulation
1. Introduction The existence of language function in the left basal temporal area (LBTA) was first reported in the study of patients with epilepsy using electrical cortical stimulation T Corresponding author. Fax: +81 75 751 9416. E-mail address: [email protected] (A. Ikeda). 0926-6410/$ - see front matter D 2005 Elsevier B.V. All rights reserved. doi:10.1016/j.cogbrainres.2005.02.001
in1980s [15,16]. After the pioneering research, neuroimaging and electrophysiological studies of normal subjects using functional MRI (fMRI) [2,4,5,7,8,18,19,23], positron emission tomography (PET) [4,20,21,29], or magnetoencephalography (MEG) [14,27] have revealed that the area is related to both language and non-language cognitive functions. Although these imaging and electrophysiological techniques are widely used because they are non-invasive and thus
K. Usui et al. / Cognitive Brain Research 24 (2005) 274–283
can be easily applied to both normal subjects and patients, the methods are indirect in terms of the capability of precisely locating a specific function in the brain, and of distinguishing and/or identifying pathways of different but closely related tasks. Electrical cortical stimulation, by contrast, enables us to directly investigate functions of a specific target area in the brain by causing its transient functional impairment, which most of other methods cannot provide. Cortical stimulation studies located the language area in LBTA somewhere between 10 mm and 80 mm from the anterior tip of temporal lobe, including the fusiform gyrus and portions of parahippocampal and inferior temporal gyri [3,25]. While some consensus was obtained about the anatomical location, language functions of the area have been reported to be variable depending on the subject, probably, at least in part, due to different location of electrode placement among subjects [9,13]. Unique characteristics in Japanese orthographic system provide additional dimension in the study of LBTA. Concise description of Japanese two writing systems, kana (syllabogram) and kanji (morphogram), is provided in Usui et al. [28] and the references therein. Each kana letter has one phonological value and somewhat similar to an alphabetical syllable in Western languages. A sequence of kana letters, therefore, can be pronounced very easily, whether it is a word or non-word. Kanji characters, by contrast, have strong association with meaning, and their pronunciations vary depending on the context in which they are used. Kana words differ from kanji words in that the former consists of sequential letters, each of which has no meaning but carries meaning as a cluster. Since kana words can evoke either their corresponding sounds alone or both sounds and meaning (semantic information), it is possible to read them either with or without the retrieval of semantic information. Since these two distinct writing systems are commonly used in daily life, it is of great interest to clarify whether or not both kanji and kana are processed through the same route in the brain. Kanji reading was found to be processed in LBTA in several studies [11,22,24,26,28]. As for kana processing, however, there is no clear evidence that will lead to the identification of responsible cortical area(s), although the lesions in angular gyrus [11] or putamen [12] were associated with disturbance of kana reading or writing.
275
In this study, as a part of presurgical functional mapping in patients with refractory epilepsy, we employed electrical cortical stimulation to clarify the details of language-related functions in LBTA. Our main objectives include commonality and/or distinction of pathways for processing Japanese morphogram (kanji) and syllabogram (kana), and correlations between orthographic and non-orthographic information processing.
2. Subjects and methods 2.1. Subjects Six patients with medically refractory mesial temporal lobe epilepsy participated in the study. Table 1 presents the clinical characteristics of the patients. Their presurgical neurological examinations were unremarkable. The language dominant hemispheres were judged to be the left by Wada test in all cases. Evaluations with prolonged video-scalp EEG monitoring, MRI, and [18F]fluorodeoxyglucose-PET suggested their seizure foci in the left temporal lobe, but did not provide sufficient information for precisely identifying the epileptogenic focus. Therefore, subdural electrodes were implanted in order to localize the epileptogenic zone and eloquent areas of the cortex. This procedure was approved by the Ethics Committee of Kyoto University Graduate School of Medicine (No. 79), and a written informed consent was obtained from each patient. Among the six patients, the results of functional mapping of Patient 1 were previously presented elsewhere for a different purpose [28]. 2.2. Methods 2.2.1. Implantation of electrodes Subdural electrode grids and strips were implanted on the lateral and basal aspect of the left temporal lobe in all six patients. The grids consisted of two or four rows, each row containing 5 to 8 platinum electrodes, with 10 mm center-to-center inter-electrode distance. Disc-shaped electrodes of 4.0 mm in diameter were embedded in sialistic sheet with exposed surface of 2.3 mm in diameter. The strip consisted of a single row of 6 electrodes of the same configuration as used for grids.
Table 1 Demographic profiles of patients Patient number
1 2 3 4 5 6
Age
25 25 30 25 39 27
Gender
F F M F F F
Pathology
Lt. Lt. Lt. Lt. Lt. Lt.
Tumor MTLE MTLE MTLE MTLE MTLE
Language dominance by Wada test
Lt Lt Lt Lt Lt Lt
WAIS-R Performance IQ
Verbal IQ
Full-scale IQ
94 92 104 84 77 95
101 71 88 82 72 73
98 76 94 81 71 81
WAIS-R, Wechsler Adult Intelligence Scale—Revised, Japanese version; Lt, left; MTLE, mesial temporal lobe epilepsy.
276
K. Usui et al. / Cognitive Brain Research 24 (2005) 274–283
The arrangements of electrodes were confirmed by skull radiograph and three-dimensional MRI (3D-MRI), as well as direct visual inspection during the surgical operation. One hundred twenty four slices of 1.5 mm thickness were acquired using GE signa (1.5 T) with the conditions of TR = 11.2 ms, TE = 2.1 ms, and inversion time = 300 ms. Fig. 1 shows the surface-rendered MRI of each patient demonstrating the location of electrodes. The number of electrodes that covered LBTA was 20 in Patient 1 (Fig. 1A), 14 in Patient 2 (Fig. 1B), 16 in Patient 3 (Fig. 1C), and 12 each in Patients 4 (Fig. 1D), 5 (Fig. 1E), and 6 (Fig. 1F). 2.2.2. Screening test for identifying functional cortical area Sessions for functional brain mapping by cortical stimulation started 3–4 days after electrode implantation. Subjects were sitting upright or half-upright in bed while being tested. Details about the method of electrical stimulation have been described elsewhere [10]. Constant electric current of square waves with alternating polarity (width of each pulse 0.3 ms) was delivered at 50 Hz to a pair of neighboring electrodes. The electric current was increased from 0 mA to 15 mA with 1 mA step. The duration of electric stimulus was increased by 1-s increment until it reached the maximum of 5 s. Whenever afterdischarges were induced, the trial was repeated with the same or one step (1 mA) lower current. Every pair of electrodes was first tested for motor responses. Subjects tapped their fingers or pronounced the sound /ra/ repeatedly. If the task was impaired by electrical
stimulation, the cortical area under the stimulated electrodes was judged to be dnegative motor areaT [10]. Those pairs of electrodes in LBTA that did not induce any motor responses were subjected to the screening of language functions using paragraph reading and object/ picture naming. In the paragraph reading, subjects read aloud a part of children’s story consisting of 100–150 words. For the object/picture naming, 6–18 non-animate objects (e.g., watch), each of which was printed in black on a white card, were shown to the subjects. Test items were carefully chosen from materials that are commonly used in daily life of ordinary Japanese so that the performance of subjects would not be affected by the familiarity with the items. After confirming that the tasks were performed without errors before electrical stimulation, the subjects repeated the same tasks during electrical stimulation. In the paragraph reading, electrical stimulation started when subjects finished reading the first sentence. In the object/picture naming, electrical stimulation was given after subjects finished naming the first two items. Test items were given one by one soon after subjects named each given item. When they could not name the given item for more than 2 s, the next item was presented. During each 5 s trial with electrical stimulation, 3–5 items were presented. Responses were rated as errors when subjects made either no response, delay in response for more than 2 s, or incorrect verbal reply. All sessions were video-recorded and electrocorticogram was recorded simultaneously.
Fig. 1. Location of subdural electrodes shown on surface-rendered T1-weighted MRIs for each of 6 patients studied. A–F in the order of patient number. Electrodes which induced language impairment during electrical stimulation are indicated by white. In Patients 1, 3, and 4, the numbers of electrodes (e.g., A4/9) are also shown.
K. Usui et al. / Cognitive Brain Research 24 (2005) 274–283
2.2.3. Evaluation of language functions in LBTA In order to investigate language functions in LBTA for more details, the electrodes which induced language impairment in the screening test (see above) were subjected to further examination. In the tasks related to kanji and kana, the results of the subjects whose verbal IQ was 80 or higher were used in order to evaluate those results reliably. The tasks are summarized in Table 2. In Patients 1 and 2, referential electrode derivation was used for stimulation by choosing a reference among those electrodes that did not induce any motor or language impairment during the screening test. The remaining patients (Patients 3–6) were examined by bipolar method using adjacent electrode pairs. Language functions were examined for both visual and auditory modalities. Visual tasks included object/picture naming, reading kanji (morphogram), reading kana words and non-words (syllabogram), copying words, and tool use (Table 2). In object/picture naming, the same line drawings used for the screening sessions were employed. The way of presenting visual stimuli and the timing of initiating electric current during tasks were also the same as in the screening sessions. In reading kanji and kana words, the same 6–18 items used in object/picture naming were
277
employed. In the study of Patient 4, in addition to the original set of kana words (set 1), another set of kana words (set 2) was also used in kana reading. While set 1 included kana words which were totally different from each other both in semantics and phonology, set 2 included kana words which were semantically different but phonologically similar (Table 2). In kana non-word reading, 3letter kana non-words were used. Concrete objects were used in the task of tool use. In the examination of Patient 4, visual matching tasks were also tested (Table 2). The subject was presented with candidate words shown together (6 words each for kanji and kana) and was requested to choose the word that matched with the given target stimulus. Electrical stimulation was applied after the subject completed the first two matchings. Auditory tasks consisted of repeating words and making gestures following verbal commands given by the experimenter (Table 2). In all the tasks described above, it was confirmed that the tasks were performed without errors before electrical stimulation. The tasks were repeated with electrical stimulation and the performance of subjects was evaluated with the same criteria as employed for the screening sessions; responses were rated as errors when patients made either no
Table 2 The tasks used for testing language and related functions in the electrical stimulation study
278
K. Usui et al. / Cognitive Brain Research 24 (2005) 274–283
or no response. In Patient 1, the errors were perseverative paraphasia (30%) and no response (70%). In Patient 3, no response was the only error. In Patient 4, semantic paraphasia (27%), obvious delay (9%), and no response (64%) were observed. Patient 3 showed no response in the stimulation of A3/8 only in the first trial but not in all the subsequent trials. Likewise, Patient 4 showed no response in the stimulation of A11/16 only in the first trial.
response, delay in response for more than 2 s, or incorrect verbal reply. All sessions were video-recorded and electrocorticogram was recorded simultaneously.
3. Results 3.1. Screening test Across the six subjects, stimulation of 14 out of 43 pairs of electrodes placed on LBTA induced language deficiency. Two pairs of electrodes in Patient 1, three pairs in Patient 2, two pairs each in Patients 3 and 4, four pairs in Patient 5, and one pair in Patient 6 showed stimulus-induced disturbance in at least one of the two screening tasks, i.e., paragraph reading or picture naming (Fig. 1).
3.2.2. Reading kanji All the three subjects showed errors in all trials in reading kanji words by stimulation of specific electrodes (all electrodes in Patients 1, A4/9 in Patient 3, and A13/18 in Patient 4). After the cessation of stimulation, all of these three subjects reported that they clearly saw the stimuli and knew what they were. In some trials, however, they could not figure out how to read the given words during electrical stimulation. In some other trials, they simply could not pronounce the words even though they knew how to pronounce them in their mind. The type of errors was either semantic paraphasia, perseverative paraphasia, or no response. In Patient 1, the errors were semantic paraphasia (16%) and no response (84%). In Patient 3, no response was the only error. In Patient 4, 60% of errors were perseverative paraphasia and the rest were no response. In the examination of A11/16 in Patient 4, electrical stimulation induced obvious delay in the first trial. The subject made no errors in the trials afterwards.
3.2. Stimulus-induced language disturbance The results of three subjects, Patients 1, 3, and 4, who satisfied the inclusion criteria (i.e., verbal IQ of more than 80) are shown in Table 3. In visual tasks, tool using and copying were preserved in all cases examined. Picture naming, kanji reading, and kana reading showed different results depending on the subjects, tasks, and the location of electrodes stimulated, as described below. 3.2.1. Object/picture naming All three subjects showed complete disturbance in the stimulation of specific electrodes (all electrodes in Patient 1, A4/9 in Patient 3, and A13/18 in Patient 4). All subjects reported that they clearly saw the stimuli and knew what they were. In some cases, however, they could not name the presented objects or pictures. In some other cases, they could not pronounce the name even though the name itself came to their mind. The type of errors was either semantic paraphasia, perseverative paraphasia, obvious delay for more than 2 s,
3.2.3. Reading kana The results of kana word and non-word reading varied depending on the subjects and the location of electrodes. In kana word (set 1) reading, Patient 1 did not show any disturbance. Patient 3 showed partial disturbance by the stimulation of two pairs of electrodes. In the stimulation of A3/8, he showed obvious delay in response for more than 2 s in the first two trials but became proficient in the latter two
Table 3 Rate of correct responses during electrical stimulation (number of tests with correct responses/number of tests) Tasks
Visual tasks
Patients (electrodes)
Picture naming Reading
Matching
Auditory tasks
kanji words kana words (set 1) kana words (set 2) kana non-words picture with kanji words picture with kana words
Copying Tool use Word repetition Gesture by verbal command
1
3
A12, A13
A3/8
A4/9
A11/16
4 A13/18
0/10 0/12 8/8 – 6/6 – – 6/6 6/6 6/6 6/6
8/9 3/3 2/4 – 0/3 – – – 3/3 3/3 3/3
0/3 0/3 2/4 – – – – – 3/3 3/3 3/3
10/11 7/8 5/6 0/5 0/4 – – 3/3 3/3 3/3 5/5
0/11 0/5 2/4 – 2/3 3/3 0/3 4/4 3/3 3/3 3/3
Responses of the patients were rated as errors when they made no response, delay in response for more than 2 s, or incorrect verbal reply. –: task not performed.
K. Usui et al. / Cognitive Brain Research 24 (2005) 274–283
trials. In the stimulation of A4/9, he showed obvious delay in response in the first two trials. He made no errors in the last two trials. After the trials, however, he reported that he was uncertain whether he read the given words correctly or not. Patient 4 also showed disturbance in the stimulation of two pairs of electrodes. Set 2 kana words (consisting of kana words different in semantics but similar in phonology) were used only for the examination of the electrodes A11/16 in this subject. The stimulation of A11/16 induced disturbance in kana word reading with contrasting results between sets 1 and 2. Kana words in set 1 were read correctly, except in the first trial of obvious delay. In contrast, reading kana words in set 2 was completely disturbed. In the stimulation of A13/ 18 using kana words of set 1, she made no reply in the first trial and showed semantic paraphasia in the second. She made no errors in the last two trials. In kana non-word reading, Patient 1 made no errors. Patient 3 showed obvious delay in response in all trials of the A3/8 stimulation. The electrode A4/9 of this subject was not tested for kana non-word reading due to the limited time allocated. Patient 4 showed erroneous reading in one trial and obvious delay in response in the remaining three trials in the stimulation of A11/16. In the stimulation of A13/18, she showed delay in the first trial but became proficient in the trials afterwards. 3.2.4. Matching Matching was tested only on a specific pair of electrodes (A13/18 in Patient 4) that induced complete impairment in kanji reading but partial impairment in kana reading. The electrodes caused disturbance in matching picture with kana words upon stimulation, but no disturbance with kanji words. 3.2.5. Copying and tool use Copying and tool use were preserved in all cases examined. Those electrodes which induced disturbance in other tasks, such as kanji reading or kana non-word reading, did not induce any impairment in these two tasks. 3.2.6. Auditory tasks All auditory tasks were performed without errors in the three subjects. 3.3. Correlation among different tasks Some electrodes that caused language disturbance upon stimulation did so in more than one task. Kanji word reading was the most visible of such example. All the electrodes that provoked complete disturbance in kanji word reading (all electrodes in Patients 1, A4/9 in Patient 3, and A13/18 in Patient 4) also induced complete disturbance in object/ picture naming. Kana non-word reading showed features somewhat different from the tasks described above. The electrodes
279
which caused impairment in kanji reading induced no disturbance in kana non-word reading in one subject (A12 and A13 in Patient 1) and partial disturbance in another (A13/18 in Patient 4). Other electrodes caused no impairment in kanji reading, but induced disturbance in kana nonword reading (A3/8 in Patient 3 and A11/16 in Patient 4). Thus, kana non-word reading and kanji reading exhibited exactly the opposite performance.
4. Discussion Taking full advantage of electrical cortical stimulation, the present study addressed the issues that require further investigation for better understanding of LBTA; not just the existence of a specific function but the correlation among different functions and the identification of processing stages or components involved in the functions in this area. Stimulation-induced impairment is an indication that the area beneath the electrode carries a function examined by the particular task. Therefore, difference in performance observed for different electrodes indicates the localization of a specific function within LBTA. 4.1. Kana (Japanese syllabogram) processing vs. Kanji (Japanese morphogram) and object/picture processing in LBTA We previously reported that the kanji-processing route runs through LBTA and that kanji reading shares at least one anatomical area with object/picture naming [28]. The present study also identified the common area for kanji reading and object/picture naming. Patients 3 and 4 showed stimulus-induced impairment both in kanji reading and object/picture naming in a certain area of LBTA. In the case of Patient 1, the electrodes which induced disturbance in kanji reading and object/picture naming had no effect on kana reading (both word and non-word), suggesting that processes of kanji and kana involve different pathways. The present study provided additional direct significant evidence that kana is processed in LBTA, which has not been previously reported. The results of Patients 3 and 4 showed that electrical stimulation applied to some areas within LBTA induced complete disturbance in kana non-word reading without hindering kanji reading. These findings indicate that processings of kanji and kana involve different pathways in LBTA at least at some stage of processing (Fig. 2). Moreover, the manner of disturbance observed in different tasks (e.g., complete disturbance in kana non-word reading and partial disturbance in kana word reading) sheds light on the underlying mechanism for language processing, which will be discussed in the next section.
280
K. Usui et al. / Cognitive Brain Research 24 (2005) 274–283
reference to either semantic or phonological information, because only the information of visual shape is necessary. These facts suggest that the combined signal containing morphological, phonological, and semantic information is eventually separated to go through different routes. In Patient 4, we carried out additional tests to clarify the details of language processing associated with this area.
Fig. 2. Two distinct language functions in LBTA. A common processing area for objects/pictures and kanji and another area associated with kana processing are located in LBTA.
4.2. Hypothesis for language-related visual information pathways in LBTA In cognitive research, theoretical models involving several stages or components of processing have been applied in order to improve the understanding of language functions. A dual-route model of reading, for example, claims that two separate pathways exist for word reading; in a direct route, a word as a whole is converted into its phonological representation, while, in an indirect pathway, each orthographic constituent of a word is sequentially transformed into its corresponding phonological constituent [1,6]. Although our study covered only a fraction of the brain, it revealed direct evidence of distinct function-brain association for visual language processing in LBTA. 4.2.1. Pathways for processing kanji words and objects/ pictures The findings of this study enable us to construct a hypothetical partial schema of kanji and object/picture processings (Fig. 3A). Visual entities such as kanji words and objects/pictures initially carry one specific information of visual shape (or morphological information). The fact that a person reads a word, names an object, or matches a picture with a specific word indicates that, at some stage of signal processing in the brain, the signal carries two other types of information, i.e., semantic information such as concept or meaning, and phonological information such as pronounceable name or label, together with morphological information. The initial signal containing morphological information, therefore, possibly initiates an access to other information at a certain stage of signal processing. As shown in Table 3, electrical stimulation applied to a specific cortical area in LBTA caused complete disturbance both in kanji reading and object/picture naming but had no effects on the tasks such as using tools or copying. The task of copying, for example, can be carried out without
Fig. 3. Hypothetical partial schemas of object/picture and kanji word processing (A) and kana non-word processing (B). Flows of information processing are shown from top to bottom. Visual input such as objects/ pictures evokes processing signals related to morphological, semantic, and phonological information in the brain. Correct output (e.g., matching or copy) indicates successful processing and incorrect output (e.g., reading/ naming) indicates failure in processing.
K. Usui et al. / Cognitive Brain Research 24 (2005) 274–283
Matching task between pictures and kanji words was performed without errors, which indicates that concepts of pictures and meanings of kanji words were correctly recognized. Semantic information of the assigned items, whether they are orthographic or non-orthographic, was identified correctly. In the same cortical site, however, electrical stimulation induced complete disturbance in kanji reading and object/picture naming. The finding indicates that this particular area is responsible for converting the correctly retrieved semantic information into phonological representation. In other words, phonological processing was blocked by electrical stimulation of that area, while morphological and semantic information processings were not affected. 4.2.2. Pathways for processing kana non-word A hypothetical partial schema of kana non-word processing based on the present findings is shown in Fig. 3B. Since kana non-words carry no meaning (i.e., no semantic information) at all, reading of kana non-words necessitates one-by-one conversion of each kana letter to the corresponding syllable. The task of kana non-word reading is, therefore, the process of converting morphological, or letterlevel orthographic, information into phonological representation. During electrical stimulation, the recognition of orthographic information was assumed to be intact because there was no error in copying kana non-words. The impairment of kana non-word reading thus suggests that the electrically stimulated region possesses a function of converting pure orthographic information (i.e., without semantic information) into phonological representation. In other words, processing of phonological information evoked by orthographic information is blocked by electrical stimulation of this particular area. 4.2.3. Possible cause of partial hindrance in kana word reading and related pathways The two properties of kana words as described in the Introduction enable native Japanese speakers to read kana words in two distinct ways; one is to read letter-by-letter, which can be accomplished without consideration of the meaning of word, and the other is to read a word as a whole. In the latter case, kana words are directly connected to its meaning, i.e., phonological information is assumed to be retrieved through semantic information. Although native Japanese speakers can intentionally choose how to read kana words between the two ways of reading, they usually read without intending to choose one way or the other. Instead, in daily life, the two ways of reading kana words are supposed to be chosen automatically, depending on the words, sentences, or context. For example, familiar words tend to be read as a whole, while unfamiliar words and phonologically similar words tend to be read letter-by-letter. Since familiarity with a word strongly depends on past experience and cultural/educational background of each individual, whether a kana word
281
is read letter-by-letter or a word as a whole also depends on individuals. The two ways of kana word reading provide one possible explanation of the present results, which showed partial hindrance in kana word reading. Two subjects (Patients 3 and 4) had difficulty in overt reading of kana words (set 1) in the first trials under electrical stimulation, suggesting that the signal evoked by visual information was blocked in this particular area. After a couple of trials, however, they made no errors even during electrical stimulation. This finding indicates that the signal evoked by visual information bypassed the location where it was blocked during the first trials. Since a kana word is a cluster of sequential letters, which carries meaning not as each letter but as a cluster, we carried out a detailed examination in one subject (Patient 4) how the performance of the subject is affected by the similarity or difference among the group of words. When the subject read a group of kana words with difference both in semantics and phonology (kana words in set 1), she could read the words without error except in the first trial. By contrast, the disturbance in reading of kana word was observed when reading a group of kana words with difference in semantics but similar in phonology (set 2). The comparison of results in reading kana words in sets 1 and 2 performed by Patient 4 discloses an aspect of kana word processing. Mild impairment in reading kana words in set 1 and complete impairment in set 2 indicates that phonological similarity between words interferes with the processing of kana words. The findings described above lead to important assumptions on kana word processing. (1) Visual information evokes phonological information alone, or both phonological and semantic information. (2) Existence of two signal pathways; one carrying only phonological information, and the other carrying both phonological and semantic information. The latter is assumed to process a kana word not as a sequence of letters but as a whole. (3) A pathway for phonological information is the primary route (i.e., the first one to be taken) for overt kana word reading. (4) The other pathway for phonological and semantic information is a secondary route, which may be activated if the primary route is blocked. Based on the assumptions mentioned above, the hindrance observed in reading of phonologically similar kana words implies that reading of this group of words primarily goes through the route for visual information and corresponding sounds. Since many of the words in Japanese language are able to be represented both by kanji and kana, it is highly likely that ordinary Japanese adults, even though the usage may vary among individuals, automatically use
282
K. Usui et al. / Cognitive Brain Research 24 (2005) 274–283
these two pathways depending on the type of words and/or sentences. The results of matching tasks also give us some clue in kana word processing. Matching pictures with kana words failed, while matching pictures with kanji words was preserved. As mentioned in Section 4.2.1, semantic information of both pictures and kanji words was successfully processed. In contrast, the failure in matching pictures with kana words indicates that the retrieval of semantic information of kana words is impeded. The fact suggests that association between a word and its meaning is not so direct in kana words as in kanji words. Based on the assumptions mentioned above, a hypothetical partial schema for kana word reading is provided in Fig. 4. Our results and hypothesis give a significant indication that the three types (or two types in kana nonwords) of information are conveyed together at some stages and are separated into different routes at some other stages in the network of brain. 4.3. Auditory processing Since most of the previous studies have focused on visual language processing in terms of functions in LBTA, this area of the brain is considered to be mainly related to visual functions. Only a small number of reports using electrical cortical stimulation of LTBA pointed out auditory language processing in addition to the visual one [17]. Our studies revealed that simple verbal commands were carried out without difficulty. The same electrodes induced complete disturbance of object/picture naming, kanji reading, and partial impairment of kana reading. Tool using and copying of kanji/kana words and kana non-word were performed without errors. The patient’s responses to the
Fig. 4. A hypothetical partial schema for kana word reading. Flows of information processing are shown from top to bottom. The results of this study indicate the two possible pathways for processing kana words. The pathway involving morphological and phonological information is assumed to be primary and the other involving semantic information secondary.
tasks indicate that attention and motor functions were intact. Since the number of subjects and the area covered by electrodes are limited, a possibility that LBTA carries some functions of processing auditory language information might still remain. Our study so far, however, has no clear evidence that LBTA has functions of processing auditory language information. 4.4. Concluding remarks In our study of the language-related functions in LBTA, we have to rely on subjects who have epileptic foci in the left temporal lobe. We proved left language lateralization by Wada test. The number of cases which satisfy these conditions is small among all the cases we deal with in our institution. Although a large number is preferable to obtain statistically meaningful result and we understand that additional condition may reduce the number available in our study, we should not sacrifice quality to number. We further employed a stringent criteria for the verbal IQ over 80 because language-related functions of this degree of sophistication are specifically characteristic to human being and we tried to ensure the reliability of the results by using the data of subjects with appropriate level of intelligence. As a result of the requirements mentioned above, the number of the subjects was reduced, and a complete general description of visual language processing in human brain is far beyond the scope of this study. Non-invasive methods, however, are indirect in terms of the capability of precisely locating a specific function in the brain and of distinguishing and/or identifying pathways of different but closely related tasks. Since electrical cortical stimulation study is essential to overcome these limitations, we believe that a small number should not deter the development of hypothetical proposals at an early stage of the study. Although statistically weak in terms of the number of samples and our interpretation is by no means the only possible explanation of the observed phenomena, the new insight provides a clue for at least partial description of the processing pathways in LBTA. In summary, our study provides direct evidence that LBTA carries functions of processing visual language information. The main findings were (a) kana words and non-words are processed, (b) a common area for kanji reading and object/picture naming exists, and (c) kanji and kana are processed via different routes. The findings of this detailed examination also lead to valuable indications and/or assumptions; (i) the existence of two pathways for kana reading: one dealing with phonological aspect of the words and the other dealing with semantic and phonological aspects; (ii) visual entities carrying semantic information, whether they are orthographic or non-orthographic, are processed. Our findings have significantly improved our understanding of language-related functions in LBTA, which has
K. Usui et al. / Cognitive Brain Research 24 (2005) 274–283
been unavailable so far. Further studies using more subjects are essential to provide details of the processing pathways, which will lead to a network model of information processing in the brain with more sophistication.
Acknowledgments This study was supported by Grants-in-Aid for Scientific Research (C2) 1559150 from Japan Ministry of Education, Culture, Sports, Science and Technology for AI. This study was also supported by the Grant-in-Aid for Scientific Research (B2) 14370205 from the Japan Society for the Promotion of Science and the Science of Mind from the Ministry of Health, Labor and Welfare for HF. We thank Drs. Nobuhiro Mikuni, Riki Matsumoto, Jun-ichi Yamamoto, and Takefumi Hitomi for their help in experiments.
References [1] S.Y. Bookheimer, T.A. Zeffiro, T. Blaxton, W. Gaillard, W. Theodore, Regional cerebral blood flow during object naming and word reading, Hum. Brain Mapp. 3 (1995) 93 – 106. [2] J.R. Booth, D.D. Burman, J.R. Meyer, D.R. Gitelman, T.B. Parrish, M.M. Mesulam, Modality independence of word comprehension, Hum. Brain Mapp. 16 (2002) 251 – 261. [3] T.H. Burnstine, R.P. Lesser, J. Hart Jr., S. Uematsu, S.J. Zinreich, G.L. Krauss, et al., Characterization of the basal temporal language area in patients with left temporal lobe epilepsy, Neurology 40 (1990) 966 – 970. [4] R. Cabeza, L. Nyberg, Imaging cognition II: an empirical review of 275 PET and fMRI studies [review], J. Cogn. Neurosci. 12 (2000) 1 – 47. [5] L. Cohen, S. Dehaene, L. Naccache, S. Lehericy, G. DehaeneLambertz, M.A. Henaff, et al., The visual word form area: spatial and temporal characterization of an initial stage of reading in normal subjects and posterior split-brain patients, Brain 123 (2000) 291 – 307. [6] M. Coltheart, B. Curtis, P. Atkins, M. Haller, Models of reading aloud: dual-route and parallel-distributed processing approaches, Psychol. Rev. 100 (1993) 589 – 608. [7] S. Dehaene, L. Naccache, L. Cohen, D.L. Bihan, J.F. Mangin, J.B. Poline, D. Riviere, Cerebral mechanisms of word masking and unconscious repetition priming, Nat. Neurosci. 4 (2001) 752 – 758. [8] C.J. Fiebach, A.D. Friedrici, K. Mqller, D.Y. von Cramon, fMRI evidence for dual routes to the mental lexicon in visual word recognition, J. Cogn. Neurosci. 14 (2002) 11 – 23. [9] J. Hart Jr., N.E. Crone, R.P. Lesser, J. Sieracki, D.L. Miglioretti, C. Hall, et al., Temporal dynamics of verbal object comprehension, Proc. Natl. Acad. Sci. U. S. A. 95 (1998) 6498 – 6503. [10] A. Ikeda, H.O. Lqders, R.C. Burgess, H. Shibasaki, Movement-related potentials recorded from supplementary motor area and primary motor area. Role of supplementary area in voluntary movements, Brain 115 (1992) 1017 – 1043.
283
[11] M. Iwata, Kanji versus kana: neuropsychological correlates of the Japanese writing system, Trends Neurosci. 7 (1984) 290 – 293. [12] K. Kokubo, K. Suzuki, A. Yamadori, K. Satou, Pure kana agraphia as a manifestation of graphemic buffer impairment, Cortex 37 (2001) 187 – 195. [13] G.L. Krauss, R. Fisher, C. Plate, J. Hart, S. Uematsu, B. Gordon, et al., Cognitive effects of resecting basal temporal language areas, Epilepsia 37 (1996) 476 – 483. [14] S. Kuriki, F. Takeuchi, Y. Hirata, Neural processing of words in the human extrastriate visual cortex, Brain Res. Cogn. Brain Res. 6 (1998) 193 – 203. [15] H.O. Lqders, R.P. Lesser, J. Hahn, D.S. Dinner, H. Morris, S. Resor, et al., Basal temporal language area demonstrated by electrical stimulation, Neurology 36 (1986) 505 – 510. [16] H.O. Lqders, R.P. Lesser, J. Hahn, D.S. Dinner, H.H. Morris, E. Wyllie, et al., Basal temporal language area, Brain 114 (1991) 743 – 754. [17] B.A. Malow, T.A. Blaxton, S. Sato, S.Y. Bookheimer, C.V. Kufta, C.M. Figlozzi, et al., Cortical stimulation elicits regional distinctions in auditory and visual naming, Epilepsia 37 (1996) 245 – 252. [18] B.D. McCandliss, L. Cohen, S. Dehaene, The visual word form area: expertise for reading in the fusiform gyrus, Trends Cogn. Sci. 7 (2003) 293 – 299. [19] K.B. McDermott, S.E. Petersen, J.M. Watson, J.G. Ojemann, A procedure for identifying regions preferentially activated by attention to semantic and phonological relations using functional magnetic resonance imaging, Neropsychologia 41 (2003) 293 – 303. [20] C.J. Moore, C.J. Price, Three distinct ventral occipitotemporal regions for reading and object naming, NeuroImage 10 (1999) 181 – 192. [21] S. Murtha, H. Chertkow, M. Beauregard, A. Evans, The neural substrate of picture naming, J. Cogn. Neurosci. 11 (1999) 399 – 423. [22] K. Nakamura, M. Honda, T. Okada, T. Hanakawa, K. Toma, H. Fukuyama, et al., Participation of the left posterior inferior temporal cortex in writing and mental recall of kanji orthography: a functional MRI study, Brain 123 (2000) 954 – 967. [23] T.A. Polk, M. Stallcup, G.K. Aguirre, D.C. Alsop, M. D’Esposito, J.A. Detre, et al., Neural specialization for letter recognition, J. Cogn. Neurosci. 14 (2002) 145 – 159. [24] Y. Sakurai, T. Momose, M. Iwata, Y. Sudo, K. Ohtomo, I. Kanazawa, Different cortical activity in reading of kanji words, kana words and kana nonwords, Brain Res. Cogn. Brain Res. 9 (2000) 111 – 115. [25] L. Sch7ffler, H.O. Lqders, G.J. Beck, Quantitative comparison of language deficits produced by extraoperative electrical stimulation of Broca’s, Wernicke’s, and basal temporal language areas, Epilepsia 37 (1996) 463 – 475. [26] Y. Soma, M. Sugishita, K. Kitamura, S. Maruyama, H. Imanaga, Lexical agraphia in the Japanese language. Pure agraphia for kanji due to left posterioinferior temporal lesions, Brain 112 (1989) 1549 – 1561. [27] A. Tarkiainen, P.L. Cornelissen, R. Salmelin, Dynamics of visual feature analysis and object-level processing in face versus letter-string perception, Brain 125 (2002) 1125 – 1136. [28] K. Usui, A. Ikeda, M. Takayama, M. Matsuhashi, J. Yamamoto, T. Satoh, et al., Conversion of semantic information into phonological representation: a function in left posterior basal temporal area, Brain 126 (2003) 632 – 641. [29] C. Whatmough, H. Chertkow, S. Murtha, K. Hanratty, Dissociable brain regions process object meaning and object structure during picture naming, Neuropsychologia 40 (2002) 174 – 186.
Research report
Processing of Japanese morphogram and syllabogram in the left basal temporal area: Electrical cortical stimulation studies Keiko Usuia, Akio Ikedab,T, Motohiro Takayamaa,c, Masao Matsuhashia, Takeshi Satowa,c, Tahamina Beguma, Masako Kinoshitab, Susumu Miyamotoc, Nobuo Hashimotoc, Takashi Nagaminea, Hidenao Fukuyamaa, Hiroshi Shibasakia,b,d a
Human Brain Research Center, Kyoto University Graduate School of Medicine, Shogoin, Sakyo-ku, Kyoto, 606-8507, Japan b Department of Neurology, Kyoto University Graduate School of Medicine, Shogoin, Sakyo-ku, Kyoto, 606-8507, Japan c Department of Neurosurgery, Kyoto University Graduate School of Medicine, Shogoin, Sakyo-ku, Kyoto, 606-8507, Japan d National Institute of Neurological Disorders and Stroke, National Institutes of Health, Bethesda, MD 20892, USA Accepted 1 February 2005 Available online 16 March 2005
Abstract Language functions in the left basal temporal area (LBTA) were investigated using electrical cortical stimulation during functional mapping in six Japanese patients with refractory epilepsy. This study provides the first direct evidence that kana (Japanese syllabogram) is processed in the LBTA. Electrical stimulation of some areas within LBTA induced disturbance in overt reading of kana words only in the first trials, with no errors in the subsequent trials. By contrast, stimulation of the same area caused obvious disturbance in kana non-word reading in all trials. Since a kana word carries both meaning and sound while a kana non-word carries only sounds of a letter string, the contrasting results of partial and complete disturbance imply a possibility that there are two distinct pathways for kana reading: one dealing with both phonological and semantic aspects of the words and the other dealing only with phonological aspect. Kanji words (Japanese morphogram) and objects/pictures were found to be processed in an area different from the area for the kana nonword processing. Furthermore, the present study also identified the common area for processing kanji reading and object/picture naming. There were no errors in matching pictures with kanji words, indicating that concepts of pictures and meanings of kanji words were not interfered by the electrical stimulation of that area. The new insight provides a clue for partial description of processing pathways for language-related visual information in LBTA. Three types of information (morphological, phonological, and semantic) are conveyed together at some stages and are separated into different routes at some other stages. D 2005 Elsevier B.V. All rights reserved. Theme: Neural basis of behavior Topic: Cognition Keywords: Left basal temporal area; Kana (syllabogram) processing; Kanji (morphogram) processing; Electrical cortical stimulation
1. Introduction The existence of language function in the left basal temporal area (LBTA) was first reported in the study of patients with epilepsy using electrical cortical stimulation T Corresponding author. Fax: +81 75 751 9416. E-mail address: [email protected] (A. Ikeda). 0926-6410/$ - see front matter D 2005 Elsevier B.V. All rights reserved. doi:10.1016/j.cogbrainres.2005.02.001
in1980s [15,16]. After the pioneering research, neuroimaging and electrophysiological studies of normal subjects using functional MRI (fMRI) [2,4,5,7,8,18,19,23], positron emission tomography (PET) [4,20,21,29], or magnetoencephalography (MEG) [14,27] have revealed that the area is related to both language and non-language cognitive functions. Although these imaging and electrophysiological techniques are widely used because they are non-invasive and thus
K. Usui et al. / Cognitive Brain Research 24 (2005) 274–283
can be easily applied to both normal subjects and patients, the methods are indirect in terms of the capability of precisely locating a specific function in the brain, and of distinguishing and/or identifying pathways of different but closely related tasks. Electrical cortical stimulation, by contrast, enables us to directly investigate functions of a specific target area in the brain by causing its transient functional impairment, which most of other methods cannot provide. Cortical stimulation studies located the language area in LBTA somewhere between 10 mm and 80 mm from the anterior tip of temporal lobe, including the fusiform gyrus and portions of parahippocampal and inferior temporal gyri [3,25]. While some consensus was obtained about the anatomical location, language functions of the area have been reported to be variable depending on the subject, probably, at least in part, due to different location of electrode placement among subjects [9,13]. Unique characteristics in Japanese orthographic system provide additional dimension in the study of LBTA. Concise description of Japanese two writing systems, kana (syllabogram) and kanji (morphogram), is provided in Usui et al. [28] and the references therein. Each kana letter has one phonological value and somewhat similar to an alphabetical syllable in Western languages. A sequence of kana letters, therefore, can be pronounced very easily, whether it is a word or non-word. Kanji characters, by contrast, have strong association with meaning, and their pronunciations vary depending on the context in which they are used. Kana words differ from kanji words in that the former consists of sequential letters, each of which has no meaning but carries meaning as a cluster. Since kana words can evoke either their corresponding sounds alone or both sounds and meaning (semantic information), it is possible to read them either with or without the retrieval of semantic information. Since these two distinct writing systems are commonly used in daily life, it is of great interest to clarify whether or not both kanji and kana are processed through the same route in the brain. Kanji reading was found to be processed in LBTA in several studies [11,22,24,26,28]. As for kana processing, however, there is no clear evidence that will lead to the identification of responsible cortical area(s), although the lesions in angular gyrus [11] or putamen [12] were associated with disturbance of kana reading or writing.
275
In this study, as a part of presurgical functional mapping in patients with refractory epilepsy, we employed electrical cortical stimulation to clarify the details of language-related functions in LBTA. Our main objectives include commonality and/or distinction of pathways for processing Japanese morphogram (kanji) and syllabogram (kana), and correlations between orthographic and non-orthographic information processing.
2. Subjects and methods 2.1. Subjects Six patients with medically refractory mesial temporal lobe epilepsy participated in the study. Table 1 presents the clinical characteristics of the patients. Their presurgical neurological examinations were unremarkable. The language dominant hemispheres were judged to be the left by Wada test in all cases. Evaluations with prolonged video-scalp EEG monitoring, MRI, and [18F]fluorodeoxyglucose-PET suggested their seizure foci in the left temporal lobe, but did not provide sufficient information for precisely identifying the epileptogenic focus. Therefore, subdural electrodes were implanted in order to localize the epileptogenic zone and eloquent areas of the cortex. This procedure was approved by the Ethics Committee of Kyoto University Graduate School of Medicine (No. 79), and a written informed consent was obtained from each patient. Among the six patients, the results of functional mapping of Patient 1 were previously presented elsewhere for a different purpose [28]. 2.2. Methods 2.2.1. Implantation of electrodes Subdural electrode grids and strips were implanted on the lateral and basal aspect of the left temporal lobe in all six patients. The grids consisted of two or four rows, each row containing 5 to 8 platinum electrodes, with 10 mm center-to-center inter-electrode distance. Disc-shaped electrodes of 4.0 mm in diameter were embedded in sialistic sheet with exposed surface of 2.3 mm in diameter. The strip consisted of a single row of 6 electrodes of the same configuration as used for grids.
Table 1 Demographic profiles of patients Patient number
1 2 3 4 5 6
Age
25 25 30 25 39 27
Gender
F F M F F F
Pathology
Lt. Lt. Lt. Lt. Lt. Lt.
Tumor MTLE MTLE MTLE MTLE MTLE
Language dominance by Wada test
Lt Lt Lt Lt Lt Lt
WAIS-R Performance IQ
Verbal IQ
Full-scale IQ
94 92 104 84 77 95
101 71 88 82 72 73
98 76 94 81 71 81
WAIS-R, Wechsler Adult Intelligence Scale—Revised, Japanese version; Lt, left; MTLE, mesial temporal lobe epilepsy.
276
K. Usui et al. / Cognitive Brain Research 24 (2005) 274–283
The arrangements of electrodes were confirmed by skull radiograph and three-dimensional MRI (3D-MRI), as well as direct visual inspection during the surgical operation. One hundred twenty four slices of 1.5 mm thickness were acquired using GE signa (1.5 T) with the conditions of TR = 11.2 ms, TE = 2.1 ms, and inversion time = 300 ms. Fig. 1 shows the surface-rendered MRI of each patient demonstrating the location of electrodes. The number of electrodes that covered LBTA was 20 in Patient 1 (Fig. 1A), 14 in Patient 2 (Fig. 1B), 16 in Patient 3 (Fig. 1C), and 12 each in Patients 4 (Fig. 1D), 5 (Fig. 1E), and 6 (Fig. 1F). 2.2.2. Screening test for identifying functional cortical area Sessions for functional brain mapping by cortical stimulation started 3–4 days after electrode implantation. Subjects were sitting upright or half-upright in bed while being tested. Details about the method of electrical stimulation have been described elsewhere [10]. Constant electric current of square waves with alternating polarity (width of each pulse 0.3 ms) was delivered at 50 Hz to a pair of neighboring electrodes. The electric current was increased from 0 mA to 15 mA with 1 mA step. The duration of electric stimulus was increased by 1-s increment until it reached the maximum of 5 s. Whenever afterdischarges were induced, the trial was repeated with the same or one step (1 mA) lower current. Every pair of electrodes was first tested for motor responses. Subjects tapped their fingers or pronounced the sound /ra/ repeatedly. If the task was impaired by electrical
stimulation, the cortical area under the stimulated electrodes was judged to be dnegative motor areaT [10]. Those pairs of electrodes in LBTA that did not induce any motor responses were subjected to the screening of language functions using paragraph reading and object/ picture naming. In the paragraph reading, subjects read aloud a part of children’s story consisting of 100–150 words. For the object/picture naming, 6–18 non-animate objects (e.g., watch), each of which was printed in black on a white card, were shown to the subjects. Test items were carefully chosen from materials that are commonly used in daily life of ordinary Japanese so that the performance of subjects would not be affected by the familiarity with the items. After confirming that the tasks were performed without errors before electrical stimulation, the subjects repeated the same tasks during electrical stimulation. In the paragraph reading, electrical stimulation started when subjects finished reading the first sentence. In the object/picture naming, electrical stimulation was given after subjects finished naming the first two items. Test items were given one by one soon after subjects named each given item. When they could not name the given item for more than 2 s, the next item was presented. During each 5 s trial with electrical stimulation, 3–5 items were presented. Responses were rated as errors when subjects made either no response, delay in response for more than 2 s, or incorrect verbal reply. All sessions were video-recorded and electrocorticogram was recorded simultaneously.
Fig. 1. Location of subdural electrodes shown on surface-rendered T1-weighted MRIs for each of 6 patients studied. A–F in the order of patient number. Electrodes which induced language impairment during electrical stimulation are indicated by white. In Patients 1, 3, and 4, the numbers of electrodes (e.g., A4/9) are also shown.
K. Usui et al. / Cognitive Brain Research 24 (2005) 274–283
2.2.3. Evaluation of language functions in LBTA In order to investigate language functions in LBTA for more details, the electrodes which induced language impairment in the screening test (see above) were subjected to further examination. In the tasks related to kanji and kana, the results of the subjects whose verbal IQ was 80 or higher were used in order to evaluate those results reliably. The tasks are summarized in Table 2. In Patients 1 and 2, referential electrode derivation was used for stimulation by choosing a reference among those electrodes that did not induce any motor or language impairment during the screening test. The remaining patients (Patients 3–6) were examined by bipolar method using adjacent electrode pairs. Language functions were examined for both visual and auditory modalities. Visual tasks included object/picture naming, reading kanji (morphogram), reading kana words and non-words (syllabogram), copying words, and tool use (Table 2). In object/picture naming, the same line drawings used for the screening sessions were employed. The way of presenting visual stimuli and the timing of initiating electric current during tasks were also the same as in the screening sessions. In reading kanji and kana words, the same 6–18 items used in object/picture naming were
277
employed. In the study of Patient 4, in addition to the original set of kana words (set 1), another set of kana words (set 2) was also used in kana reading. While set 1 included kana words which were totally different from each other both in semantics and phonology, set 2 included kana words which were semantically different but phonologically similar (Table 2). In kana non-word reading, 3letter kana non-words were used. Concrete objects were used in the task of tool use. In the examination of Patient 4, visual matching tasks were also tested (Table 2). The subject was presented with candidate words shown together (6 words each for kanji and kana) and was requested to choose the word that matched with the given target stimulus. Electrical stimulation was applied after the subject completed the first two matchings. Auditory tasks consisted of repeating words and making gestures following verbal commands given by the experimenter (Table 2). In all the tasks described above, it was confirmed that the tasks were performed without errors before electrical stimulation. The tasks were repeated with electrical stimulation and the performance of subjects was evaluated with the same criteria as employed for the screening sessions; responses were rated as errors when patients made either no
Table 2 The tasks used for testing language and related functions in the electrical stimulation study
278
K. Usui et al. / Cognitive Brain Research 24 (2005) 274–283
or no response. In Patient 1, the errors were perseverative paraphasia (30%) and no response (70%). In Patient 3, no response was the only error. In Patient 4, semantic paraphasia (27%), obvious delay (9%), and no response (64%) were observed. Patient 3 showed no response in the stimulation of A3/8 only in the first trial but not in all the subsequent trials. Likewise, Patient 4 showed no response in the stimulation of A11/16 only in the first trial.
response, delay in response for more than 2 s, or incorrect verbal reply. All sessions were video-recorded and electrocorticogram was recorded simultaneously.
3. Results 3.1. Screening test Across the six subjects, stimulation of 14 out of 43 pairs of electrodes placed on LBTA induced language deficiency. Two pairs of electrodes in Patient 1, three pairs in Patient 2, two pairs each in Patients 3 and 4, four pairs in Patient 5, and one pair in Patient 6 showed stimulus-induced disturbance in at least one of the two screening tasks, i.e., paragraph reading or picture naming (Fig. 1).
3.2.2. Reading kanji All the three subjects showed errors in all trials in reading kanji words by stimulation of specific electrodes (all electrodes in Patients 1, A4/9 in Patient 3, and A13/18 in Patient 4). After the cessation of stimulation, all of these three subjects reported that they clearly saw the stimuli and knew what they were. In some trials, however, they could not figure out how to read the given words during electrical stimulation. In some other trials, they simply could not pronounce the words even though they knew how to pronounce them in their mind. The type of errors was either semantic paraphasia, perseverative paraphasia, or no response. In Patient 1, the errors were semantic paraphasia (16%) and no response (84%). In Patient 3, no response was the only error. In Patient 4, 60% of errors were perseverative paraphasia and the rest were no response. In the examination of A11/16 in Patient 4, electrical stimulation induced obvious delay in the first trial. The subject made no errors in the trials afterwards.
3.2. Stimulus-induced language disturbance The results of three subjects, Patients 1, 3, and 4, who satisfied the inclusion criteria (i.e., verbal IQ of more than 80) are shown in Table 3. In visual tasks, tool using and copying were preserved in all cases examined. Picture naming, kanji reading, and kana reading showed different results depending on the subjects, tasks, and the location of electrodes stimulated, as described below. 3.2.1. Object/picture naming All three subjects showed complete disturbance in the stimulation of specific electrodes (all electrodes in Patient 1, A4/9 in Patient 3, and A13/18 in Patient 4). All subjects reported that they clearly saw the stimuli and knew what they were. In some cases, however, they could not name the presented objects or pictures. In some other cases, they could not pronounce the name even though the name itself came to their mind. The type of errors was either semantic paraphasia, perseverative paraphasia, obvious delay for more than 2 s,
3.2.3. Reading kana The results of kana word and non-word reading varied depending on the subjects and the location of electrodes. In kana word (set 1) reading, Patient 1 did not show any disturbance. Patient 3 showed partial disturbance by the stimulation of two pairs of electrodes. In the stimulation of A3/8, he showed obvious delay in response for more than 2 s in the first two trials but became proficient in the latter two
Table 3 Rate of correct responses during electrical stimulation (number of tests with correct responses/number of tests) Tasks
Visual tasks
Patients (electrodes)
Picture naming Reading
Matching
Auditory tasks
kanji words kana words (set 1) kana words (set 2) kana non-words picture with kanji words picture with kana words
Copying Tool use Word repetition Gesture by verbal command
1
3
A12, A13
A3/8
A4/9
A11/16
4 A13/18
0/10 0/12 8/8 – 6/6 – – 6/6 6/6 6/6 6/6
8/9 3/3 2/4 – 0/3 – – – 3/3 3/3 3/3
0/3 0/3 2/4 – – – – – 3/3 3/3 3/3
10/11 7/8 5/6 0/5 0/4 – – 3/3 3/3 3/3 5/5
0/11 0/5 2/4 – 2/3 3/3 0/3 4/4 3/3 3/3 3/3
Responses of the patients were rated as errors when they made no response, delay in response for more than 2 s, or incorrect verbal reply. –: task not performed.
K. Usui et al. / Cognitive Brain Research 24 (2005) 274–283
trials. In the stimulation of A4/9, he showed obvious delay in response in the first two trials. He made no errors in the last two trials. After the trials, however, he reported that he was uncertain whether he read the given words correctly or not. Patient 4 also showed disturbance in the stimulation of two pairs of electrodes. Set 2 kana words (consisting of kana words different in semantics but similar in phonology) were used only for the examination of the electrodes A11/16 in this subject. The stimulation of A11/16 induced disturbance in kana word reading with contrasting results between sets 1 and 2. Kana words in set 1 were read correctly, except in the first trial of obvious delay. In contrast, reading kana words in set 2 was completely disturbed. In the stimulation of A13/ 18 using kana words of set 1, she made no reply in the first trial and showed semantic paraphasia in the second. She made no errors in the last two trials. In kana non-word reading, Patient 1 made no errors. Patient 3 showed obvious delay in response in all trials of the A3/8 stimulation. The electrode A4/9 of this subject was not tested for kana non-word reading due to the limited time allocated. Patient 4 showed erroneous reading in one trial and obvious delay in response in the remaining three trials in the stimulation of A11/16. In the stimulation of A13/18, she showed delay in the first trial but became proficient in the trials afterwards. 3.2.4. Matching Matching was tested only on a specific pair of electrodes (A13/18 in Patient 4) that induced complete impairment in kanji reading but partial impairment in kana reading. The electrodes caused disturbance in matching picture with kana words upon stimulation, but no disturbance with kanji words. 3.2.5. Copying and tool use Copying and tool use were preserved in all cases examined. Those electrodes which induced disturbance in other tasks, such as kanji reading or kana non-word reading, did not induce any impairment in these two tasks. 3.2.6. Auditory tasks All auditory tasks were performed without errors in the three subjects. 3.3. Correlation among different tasks Some electrodes that caused language disturbance upon stimulation did so in more than one task. Kanji word reading was the most visible of such example. All the electrodes that provoked complete disturbance in kanji word reading (all electrodes in Patients 1, A4/9 in Patient 3, and A13/18 in Patient 4) also induced complete disturbance in object/ picture naming. Kana non-word reading showed features somewhat different from the tasks described above. The electrodes
279
which caused impairment in kanji reading induced no disturbance in kana non-word reading in one subject (A12 and A13 in Patient 1) and partial disturbance in another (A13/18 in Patient 4). Other electrodes caused no impairment in kanji reading, but induced disturbance in kana nonword reading (A3/8 in Patient 3 and A11/16 in Patient 4). Thus, kana non-word reading and kanji reading exhibited exactly the opposite performance.
4. Discussion Taking full advantage of electrical cortical stimulation, the present study addressed the issues that require further investigation for better understanding of LBTA; not just the existence of a specific function but the correlation among different functions and the identification of processing stages or components involved in the functions in this area. Stimulation-induced impairment is an indication that the area beneath the electrode carries a function examined by the particular task. Therefore, difference in performance observed for different electrodes indicates the localization of a specific function within LBTA. 4.1. Kana (Japanese syllabogram) processing vs. Kanji (Japanese morphogram) and object/picture processing in LBTA We previously reported that the kanji-processing route runs through LBTA and that kanji reading shares at least one anatomical area with object/picture naming [28]. The present study also identified the common area for kanji reading and object/picture naming. Patients 3 and 4 showed stimulus-induced impairment both in kanji reading and object/picture naming in a certain area of LBTA. In the case of Patient 1, the electrodes which induced disturbance in kanji reading and object/picture naming had no effect on kana reading (both word and non-word), suggesting that processes of kanji and kana involve different pathways. The present study provided additional direct significant evidence that kana is processed in LBTA, which has not been previously reported. The results of Patients 3 and 4 showed that electrical stimulation applied to some areas within LBTA induced complete disturbance in kana non-word reading without hindering kanji reading. These findings indicate that processings of kanji and kana involve different pathways in LBTA at least at some stage of processing (Fig. 2). Moreover, the manner of disturbance observed in different tasks (e.g., complete disturbance in kana non-word reading and partial disturbance in kana word reading) sheds light on the underlying mechanism for language processing, which will be discussed in the next section.
280
K. Usui et al. / Cognitive Brain Research 24 (2005) 274–283
reference to either semantic or phonological information, because only the information of visual shape is necessary. These facts suggest that the combined signal containing morphological, phonological, and semantic information is eventually separated to go through different routes. In Patient 4, we carried out additional tests to clarify the details of language processing associated with this area.
Fig. 2. Two distinct language functions in LBTA. A common processing area for objects/pictures and kanji and another area associated with kana processing are located in LBTA.
4.2. Hypothesis for language-related visual information pathways in LBTA In cognitive research, theoretical models involving several stages or components of processing have been applied in order to improve the understanding of language functions. A dual-route model of reading, for example, claims that two separate pathways exist for word reading; in a direct route, a word as a whole is converted into its phonological representation, while, in an indirect pathway, each orthographic constituent of a word is sequentially transformed into its corresponding phonological constituent [1,6]. Although our study covered only a fraction of the brain, it revealed direct evidence of distinct function-brain association for visual language processing in LBTA. 4.2.1. Pathways for processing kanji words and objects/ pictures The findings of this study enable us to construct a hypothetical partial schema of kanji and object/picture processings (Fig. 3A). Visual entities such as kanji words and objects/pictures initially carry one specific information of visual shape (or morphological information). The fact that a person reads a word, names an object, or matches a picture with a specific word indicates that, at some stage of signal processing in the brain, the signal carries two other types of information, i.e., semantic information such as concept or meaning, and phonological information such as pronounceable name or label, together with morphological information. The initial signal containing morphological information, therefore, possibly initiates an access to other information at a certain stage of signal processing. As shown in Table 3, electrical stimulation applied to a specific cortical area in LBTA caused complete disturbance both in kanji reading and object/picture naming but had no effects on the tasks such as using tools or copying. The task of copying, for example, can be carried out without
Fig. 3. Hypothetical partial schemas of object/picture and kanji word processing (A) and kana non-word processing (B). Flows of information processing are shown from top to bottom. Visual input such as objects/ pictures evokes processing signals related to morphological, semantic, and phonological information in the brain. Correct output (e.g., matching or copy) indicates successful processing and incorrect output (e.g., reading/ naming) indicates failure in processing.
K. Usui et al. / Cognitive Brain Research 24 (2005) 274–283
Matching task between pictures and kanji words was performed without errors, which indicates that concepts of pictures and meanings of kanji words were correctly recognized. Semantic information of the assigned items, whether they are orthographic or non-orthographic, was identified correctly. In the same cortical site, however, electrical stimulation induced complete disturbance in kanji reading and object/picture naming. The finding indicates that this particular area is responsible for converting the correctly retrieved semantic information into phonological representation. In other words, phonological processing was blocked by electrical stimulation of that area, while morphological and semantic information processings were not affected. 4.2.2. Pathways for processing kana non-word A hypothetical partial schema of kana non-word processing based on the present findings is shown in Fig. 3B. Since kana non-words carry no meaning (i.e., no semantic information) at all, reading of kana non-words necessitates one-by-one conversion of each kana letter to the corresponding syllable. The task of kana non-word reading is, therefore, the process of converting morphological, or letterlevel orthographic, information into phonological representation. During electrical stimulation, the recognition of orthographic information was assumed to be intact because there was no error in copying kana non-words. The impairment of kana non-word reading thus suggests that the electrically stimulated region possesses a function of converting pure orthographic information (i.e., without semantic information) into phonological representation. In other words, processing of phonological information evoked by orthographic information is blocked by electrical stimulation of this particular area. 4.2.3. Possible cause of partial hindrance in kana word reading and related pathways The two properties of kana words as described in the Introduction enable native Japanese speakers to read kana words in two distinct ways; one is to read letter-by-letter, which can be accomplished without consideration of the meaning of word, and the other is to read a word as a whole. In the latter case, kana words are directly connected to its meaning, i.e., phonological information is assumed to be retrieved through semantic information. Although native Japanese speakers can intentionally choose how to read kana words between the two ways of reading, they usually read without intending to choose one way or the other. Instead, in daily life, the two ways of reading kana words are supposed to be chosen automatically, depending on the words, sentences, or context. For example, familiar words tend to be read as a whole, while unfamiliar words and phonologically similar words tend to be read letter-by-letter. Since familiarity with a word strongly depends on past experience and cultural/educational background of each individual, whether a kana word
281
is read letter-by-letter or a word as a whole also depends on individuals. The two ways of kana word reading provide one possible explanation of the present results, which showed partial hindrance in kana word reading. Two subjects (Patients 3 and 4) had difficulty in overt reading of kana words (set 1) in the first trials under electrical stimulation, suggesting that the signal evoked by visual information was blocked in this particular area. After a couple of trials, however, they made no errors even during electrical stimulation. This finding indicates that the signal evoked by visual information bypassed the location where it was blocked during the first trials. Since a kana word is a cluster of sequential letters, which carries meaning not as each letter but as a cluster, we carried out a detailed examination in one subject (Patient 4) how the performance of the subject is affected by the similarity or difference among the group of words. When the subject read a group of kana words with difference both in semantics and phonology (kana words in set 1), she could read the words without error except in the first trial. By contrast, the disturbance in reading of kana word was observed when reading a group of kana words with difference in semantics but similar in phonology (set 2). The comparison of results in reading kana words in sets 1 and 2 performed by Patient 4 discloses an aspect of kana word processing. Mild impairment in reading kana words in set 1 and complete impairment in set 2 indicates that phonological similarity between words interferes with the processing of kana words. The findings described above lead to important assumptions on kana word processing. (1) Visual information evokes phonological information alone, or both phonological and semantic information. (2) Existence of two signal pathways; one carrying only phonological information, and the other carrying both phonological and semantic information. The latter is assumed to process a kana word not as a sequence of letters but as a whole. (3) A pathway for phonological information is the primary route (i.e., the first one to be taken) for overt kana word reading. (4) The other pathway for phonological and semantic information is a secondary route, which may be activated if the primary route is blocked. Based on the assumptions mentioned above, the hindrance observed in reading of phonologically similar kana words implies that reading of this group of words primarily goes through the route for visual information and corresponding sounds. Since many of the words in Japanese language are able to be represented both by kanji and kana, it is highly likely that ordinary Japanese adults, even though the usage may vary among individuals, automatically use
282
K. Usui et al. / Cognitive Brain Research 24 (2005) 274–283
these two pathways depending on the type of words and/or sentences. The results of matching tasks also give us some clue in kana word processing. Matching pictures with kana words failed, while matching pictures with kanji words was preserved. As mentioned in Section 4.2.1, semantic information of both pictures and kanji words was successfully processed. In contrast, the failure in matching pictures with kana words indicates that the retrieval of semantic information of kana words is impeded. The fact suggests that association between a word and its meaning is not so direct in kana words as in kanji words. Based on the assumptions mentioned above, a hypothetical partial schema for kana word reading is provided in Fig. 4. Our results and hypothesis give a significant indication that the three types (or two types in kana nonwords) of information are conveyed together at some stages and are separated into different routes at some other stages in the network of brain. 4.3. Auditory processing Since most of the previous studies have focused on visual language processing in terms of functions in LBTA, this area of the brain is considered to be mainly related to visual functions. Only a small number of reports using electrical cortical stimulation of LTBA pointed out auditory language processing in addition to the visual one [17]. Our studies revealed that simple verbal commands were carried out without difficulty. The same electrodes induced complete disturbance of object/picture naming, kanji reading, and partial impairment of kana reading. Tool using and copying of kanji/kana words and kana non-word were performed without errors. The patient’s responses to the
Fig. 4. A hypothetical partial schema for kana word reading. Flows of information processing are shown from top to bottom. The results of this study indicate the two possible pathways for processing kana words. The pathway involving morphological and phonological information is assumed to be primary and the other involving semantic information secondary.
tasks indicate that attention and motor functions were intact. Since the number of subjects and the area covered by electrodes are limited, a possibility that LBTA carries some functions of processing auditory language information might still remain. Our study so far, however, has no clear evidence that LBTA has functions of processing auditory language information. 4.4. Concluding remarks In our study of the language-related functions in LBTA, we have to rely on subjects who have epileptic foci in the left temporal lobe. We proved left language lateralization by Wada test. The number of cases which satisfy these conditions is small among all the cases we deal with in our institution. Although a large number is preferable to obtain statistically meaningful result and we understand that additional condition may reduce the number available in our study, we should not sacrifice quality to number. We further employed a stringent criteria for the verbal IQ over 80 because language-related functions of this degree of sophistication are specifically characteristic to human being and we tried to ensure the reliability of the results by using the data of subjects with appropriate level of intelligence. As a result of the requirements mentioned above, the number of the subjects was reduced, and a complete general description of visual language processing in human brain is far beyond the scope of this study. Non-invasive methods, however, are indirect in terms of the capability of precisely locating a specific function in the brain and of distinguishing and/or identifying pathways of different but closely related tasks. Since electrical cortical stimulation study is essential to overcome these limitations, we believe that a small number should not deter the development of hypothetical proposals at an early stage of the study. Although statistically weak in terms of the number of samples and our interpretation is by no means the only possible explanation of the observed phenomena, the new insight provides a clue for at least partial description of the processing pathways in LBTA. In summary, our study provides direct evidence that LBTA carries functions of processing visual language information. The main findings were (a) kana words and non-words are processed, (b) a common area for kanji reading and object/picture naming exists, and (c) kanji and kana are processed via different routes. The findings of this detailed examination also lead to valuable indications and/or assumptions; (i) the existence of two pathways for kana reading: one dealing with phonological aspect of the words and the other dealing with semantic and phonological aspects; (ii) visual entities carrying semantic information, whether they are orthographic or non-orthographic, are processed. Our findings have significantly improved our understanding of language-related functions in LBTA, which has
K. Usui et al. / Cognitive Brain Research 24 (2005) 274–283
been unavailable so far. Further studies using more subjects are essential to provide details of the processing pathways, which will lead to a network model of information processing in the brain with more sophistication.
Acknowledgments This study was supported by Grants-in-Aid for Scientific Research (C2) 1559150 from Japan Ministry of Education, Culture, Sports, Science and Technology for AI. This study was also supported by the Grant-in-Aid for Scientific Research (B2) 14370205 from the Japan Society for the Promotion of Science and the Science of Mind from the Ministry of Health, Labor and Welfare for HF. We thank Drs. Nobuhiro Mikuni, Riki Matsumoto, Jun-ichi Yamamoto, and Takefumi Hitomi for their help in experiments.
References [1] S.Y. Bookheimer, T.A. Zeffiro, T. Blaxton, W. Gaillard, W. Theodore, Regional cerebral blood flow during object naming and word reading, Hum. Brain Mapp. 3 (1995) 93 – 106. [2] J.R. Booth, D.D. Burman, J.R. Meyer, D.R. Gitelman, T.B. Parrish, M.M. Mesulam, Modality independence of word comprehension, Hum. Brain Mapp. 16 (2002) 251 – 261. [3] T.H. Burnstine, R.P. Lesser, J. Hart Jr., S. Uematsu, S.J. Zinreich, G.L. Krauss, et al., Characterization of the basal temporal language area in patients with left temporal lobe epilepsy, Neurology 40 (1990) 966 – 970. [4] R. Cabeza, L. Nyberg, Imaging cognition II: an empirical review of 275 PET and fMRI studies [review], J. Cogn. Neurosci. 12 (2000) 1 – 47. [5] L. Cohen, S. Dehaene, L. Naccache, S. Lehericy, G. DehaeneLambertz, M.A. Henaff, et al., The visual word form area: spatial and temporal characterization of an initial stage of reading in normal subjects and posterior split-brain patients, Brain 123 (2000) 291 – 307. [6] M. Coltheart, B. Curtis, P. Atkins, M. Haller, Models of reading aloud: dual-route and parallel-distributed processing approaches, Psychol. Rev. 100 (1993) 589 – 608. [7] S. Dehaene, L. Naccache, L. Cohen, D.L. Bihan, J.F. Mangin, J.B. Poline, D. Riviere, Cerebral mechanisms of word masking and unconscious repetition priming, Nat. Neurosci. 4 (2001) 752 – 758. [8] C.J. Fiebach, A.D. Friedrici, K. Mqller, D.Y. von Cramon, fMRI evidence for dual routes to the mental lexicon in visual word recognition, J. Cogn. Neurosci. 14 (2002) 11 – 23. [9] J. Hart Jr., N.E. Crone, R.P. Lesser, J. Sieracki, D.L. Miglioretti, C. Hall, et al., Temporal dynamics of verbal object comprehension, Proc. Natl. Acad. Sci. U. S. A. 95 (1998) 6498 – 6503. [10] A. Ikeda, H.O. Lqders, R.C. Burgess, H. Shibasaki, Movement-related potentials recorded from supplementary motor area and primary motor area. Role of supplementary area in voluntary movements, Brain 115 (1992) 1017 – 1043.
283
[11] M. Iwata, Kanji versus kana: neuropsychological correlates of the Japanese writing system, Trends Neurosci. 7 (1984) 290 – 293. [12] K. Kokubo, K. Suzuki, A. Yamadori, K. Satou, Pure kana agraphia as a manifestation of graphemic buffer impairment, Cortex 37 (2001) 187 – 195. [13] G.L. Krauss, R. Fisher, C. Plate, J. Hart, S. Uematsu, B. Gordon, et al., Cognitive effects of resecting basal temporal language areas, Epilepsia 37 (1996) 476 – 483. [14] S. Kuriki, F. Takeuchi, Y. Hirata, Neural processing of words in the human extrastriate visual cortex, Brain Res. Cogn. Brain Res. 6 (1998) 193 – 203. [15] H.O. Lqders, R.P. Lesser, J. Hahn, D.S. Dinner, H. Morris, S. Resor, et al., Basal temporal language area demonstrated by electrical stimulation, Neurology 36 (1986) 505 – 510. [16] H.O. Lqders, R.P. Lesser, J. Hahn, D.S. Dinner, H.H. Morris, E. Wyllie, et al., Basal temporal language area, Brain 114 (1991) 743 – 754. [17] B.A. Malow, T.A. Blaxton, S. Sato, S.Y. Bookheimer, C.V. Kufta, C.M. Figlozzi, et al., Cortical stimulation elicits regional distinctions in auditory and visual naming, Epilepsia 37 (1996) 245 – 252. [18] B.D. McCandliss, L. Cohen, S. Dehaene, The visual word form area: expertise for reading in the fusiform gyrus, Trends Cogn. Sci. 7 (2003) 293 – 299. [19] K.B. McDermott, S.E. Petersen, J.M. Watson, J.G. Ojemann, A procedure for identifying regions preferentially activated by attention to semantic and phonological relations using functional magnetic resonance imaging, Neropsychologia 41 (2003) 293 – 303. [20] C.J. Moore, C.J. Price, Three distinct ventral occipitotemporal regions for reading and object naming, NeuroImage 10 (1999) 181 – 192. [21] S. Murtha, H. Chertkow, M. Beauregard, A. Evans, The neural substrate of picture naming, J. Cogn. Neurosci. 11 (1999) 399 – 423. [22] K. Nakamura, M. Honda, T. Okada, T. Hanakawa, K. Toma, H. Fukuyama, et al., Participation of the left posterior inferior temporal cortex in writing and mental recall of kanji orthography: a functional MRI study, Brain 123 (2000) 954 – 967. [23] T.A. Polk, M. Stallcup, G.K. Aguirre, D.C. Alsop, M. D’Esposito, J.A. Detre, et al., Neural specialization for letter recognition, J. Cogn. Neurosci. 14 (2002) 145 – 159. [24] Y. Sakurai, T. Momose, M. Iwata, Y. Sudo, K. Ohtomo, I. Kanazawa, Different cortical activity in reading of kanji words, kana words and kana nonwords, Brain Res. Cogn. Brain Res. 9 (2000) 111 – 115. [25] L. Sch7ffler, H.O. Lqders, G.J. Beck, Quantitative comparison of language deficits produced by extraoperative electrical stimulation of Broca’s, Wernicke’s, and basal temporal language areas, Epilepsia 37 (1996) 463 – 475. [26] Y. Soma, M. Sugishita, K. Kitamura, S. Maruyama, H. Imanaga, Lexical agraphia in the Japanese language. Pure agraphia for kanji due to left posterioinferior temporal lesions, Brain 112 (1989) 1549 – 1561. [27] A. Tarkiainen, P.L. Cornelissen, R. Salmelin, Dynamics of visual feature analysis and object-level processing in face versus letter-string perception, Brain 125 (2002) 1125 – 1136. [28] K. Usui, A. Ikeda, M. Takayama, M. Matsuhashi, J. Yamamoto, T. Satoh, et al., Conversion of semantic information into phonological representation: a function in left posterior basal temporal area, Brain 126 (2003) 632 – 641. [29] C. Whatmough, H. Chertkow, S. Murtha, K. Hanratty, Dissociable brain regions process object meaning and object structure during picture naming, Neuropsychologia 40 (2002) 174 – 186.