Research ReportEffects of alphabeticality, practice and type of instruction on reading an artificial script: An fMRI study
Introduction
Reading acquisition is associated with a change in the cognitive processes involved in reading. In most reading acquisition models, the evolution of skilled reading is related to the distinction between alphabetical and non-alphabetical reading, and the application of letter decoding knowledge [23], [67], [79], [80], [89]. While reading of alphabetical words may involve letter decoding (i.e., letter segmentation and grapheme–phoneme conversion), non-alphabetical words can only be read by the retrieval of word specific representations, consisting of either the whole word or based on salient features within the word. The dual route models consider these to be distinct mechanisms for reading [20], [21], [22], [23], while connectionist models consider them as aspects of a single mechanism [47], [70], [79]. However, regardless of the question of whether letter decoding involves an abstract rule mechanism [20], [21], [22], [23] or is a rule-like behavior based on the statistical regularities of the experienced script [47], [70], [79], most models of reading acquisition agree that the reliance on letter decoding changes in the course of training. Specifically, it has been suggested that in reading alphabetical words, the reliance on segmentation and letter decoding decreases with experience, and that reading familiar words becomes dependent on lexical non-alphabetical processes [23], [80]1. On the basis of this assumption, neuroimaging studies that aimed to examine the difference between alphabetical and non-alphabetical reading often compared words and pseudo-words, or high and low frequency words [4], [31], [32], [59], [63]. The logic of these studies was that while high frequency words are expected to rely on direct retrieval of lexical representations, pseudo-words and low frequency words would rely on letter decoding since they have no effective lexical representation. One should note, however, that the design of natural script studies under this assumption confounds alphabetical reading with low amounts of practice and precludes the separate testing of each effect.
Neuroimaging studies that compared alphabetical and non-alphabetical reading in languages with two script systems (i.e., Japanese and Chinese) lead to conflicting conclusions, showing both similarities [58] and differences [59], [77], [78] between the script systems. Furthermore, even the comparison of Kana and Kanji (in Japanese) and of traditional Chinese characters with Pinyin (the alphabetical script) is confounded with morpho-semantic differences [17], as well as differences in word frequency and familiarity [41]. In addition, reading of traditional Chinese characters may not rely entirely on word-specific recognition processes due to the use of phonological cues in many of the characters. The aim of the current study is to use an artificial script in a functional imaging study to examine the separate effects of alphabeticality and the amount of experience and to test the hypothesis that the reliance on letter decoding decreases in the course of training on reading alphabetical words.
Another factor that may interact with the effects of alphabeticality and practice is the type of instruction given during training, and specifically whether explicit instruction on letter decoding is afforded [51]. There is an ongoing debate about the critical necessity of explicit instruction of phonological decoding rules for the acquisition of reading skills [35]. Previous reading acquisition studies have shown that explicit instruction on phonological decoding enhanced reading acquisition [3], [5], [14], [36], [37], [40], [66], [80], [91] and see [23] for review. Moreover, mere exposure to alphabetical orthography was, in many cases, insufficient for inducing the discovery of the alphabetic principle in children [12], [13], [15], [29], [53], [87]. However, other studies suggest that training on whole word reading may elicit learning of grapheme–phoneme correspondences by young beginning readers [30], [67], [86], [87], [89] and may even be advantageous compared to explicit instruction of letter decoding [40].
In two recent studies [6], [7], we directly addressed the question of whether whole word training results in the formation of letter representations and phonological decoding skills in literate adults. Participants received multi-session training in reading nonsense words, written in an artificial script, in which each phoneme was represented by 2 discrete symbols. Three training conditions were compared in terms of the time-course of learning and the ability to generalize the acquired knowledge (transfer): alphabetical whole words with letter decoding instruction (Explicit); alphabetical whole words without letter instruction (Implicit), and non-alphabetical whole words, with no consistent correspondence of letters to sounds (Arbitrary).
Our results [6], [7] showed that training in the explicit and arbitrary conditions resulted in distinctive learning processes. The pattern of transfer results suggested that training in the explicit condition resulted mainly in learning to recognize the individual letters, but also in some word-specific recognition. Training in the arbitrary condition resulted in word-specific recognition that was based on recognition of the internal structure of symbols in the word. Furthermore, performance in the explicit condition was more accurate, but slower than performance in the arbitrary condition, presumably because it involved letter decoding. Training in the implicit condition, resulted in word-specific recognition in all participants, in addition to non-declarative letter decoding knowledge in some participants. However, letter knowledge in the implicit condition was lower than in the explicit condition, and evolved only under specific facilitating conditions. The three training conditions did not differ only in terms of the type of knowledge that was acquired, but also in terms of preservation of learning gains. The acquired knowledge was better preserved in the explicit compared to the arbitrary and implicit conditions both between sessions, and in terms of long-term testing. This finding suggests that training in the explicit condition reached a higher, more progressed, level of skilled performance [7].
In the current study, we used the Morse-like artificial script, studied in Bitan and Karni [7], to test the interaction between the effects of the alphabeticality and the amount of experience following training in either explicit letter instruction or whole word training conditions on brain activation during reading. The separate manipulation of alphabeticality and the amount of practice enabled the examination of the hypothesis that reading of familiar (well practiced) alphabetical words does not necessarily involve letter decoding. The use of an artificial script enabled us to control the amount of practice participants received on specific words (alphabetical and non-alphabetical) by comparing trained words to less trained words. Furthermore, the inclusion of arbitrary items afforded a condition wherein the script was devoid of any alphabetical or phonological cues, which is not the case in high frequency words in natural scripts. Finally, the use of non-sense words in a phonological “translation” task eliminated the effect of semantic processes, which is confounded in the comparison of words and pseudo-words.
Our results showed that alphabeticality, the amount of practice and the type of instruction, may each (independently) affect the patterns of brain activation evoked by reading. Our results suggest that explicit training on alphabetical words relied more on letter decoding in initial as compared to later stages of reading, with the reading of highly familiar, well-trained alphabetical words much less dependent on word segmentation and letter decoding. Nevertheless, explicitly well-trained alphabetical words elicited a different pattern of activation compared to the one elicited by non-alphabetical words in the arbitrary condition, even though both of them presumably resulted in reading that did not rely on letter decoding. The pattern of activation following the implicit training on alphabetical words suggests that the reliance on letter decoding persisted to later stages of training, as compared to the explicit condition.
Section snippets
Subjects
16 right-handed female volunteers, ages 22–26, native Hebrew speakers with normal linguistic and reading skills participated in the experiment and were paid for their time. Each subject participated in two training conditions: an alphabetical condition and an arbitrary condition, serially.
Stimuli
The stimuli and the procedure of the behavioral phase were identical to those used in Bitan and Karni [7]. The training stimuli consisted of two sets of 12 nonsense words written in an artificial Morse-like
Behavioral
All training conditions induced significant improvement in the translation task, both in terms of accuracy and in terms of reaction time (RT), with no speed-accuracy tradeoffs. The GLM analyses with group and condition order as between-subject variables, and training-condition, session and block as within subject variables, showed significant effect of session (F(4,40) = 135.9 and F(4,40) = 49.5, P < 0.001, for accuracy and RT, respectively) and block (F(5,50) = 30.1 and F(5,50) = 27.1, P
Discussion
The results of the current study show that the amount of experience afforded for a specific set of stimuli is an important factor, interacting with the type of script (alphabetical vs. non-alphabetical) and the type of instructions (explicit vs. implicit) in determining the involvement of different cortical regions in reading. More experience resulted in lesser recruitment of the right SPL in: (a) all transfer conditions compared to the trained conditions, and (b) in the artificial script
Conclusions
Our results suggest that the posterior LIFG is involved in letter decoding as part of its broader function in segmentation and decoding of sequences. The results provide a neural support for the notion that reading of well-trained alphabetical words does not rely on letter decoding. Nevertheless, our results suggest that even when reading of trained alphabetical words is done with minimal recourse to letter decoding, it may still elicit a different pattern of activation from reading
Acknowledgments
This research was conducted while the first author was affiliated with the Neurobiology department in Weizmann Institute, Israel. The authors would like to thank Dr. James Booth for his valuable comments on the manuscript.
References (92)
- et al.
Functional activation of the human brain during mental rotation
Neuropsychologia
(1997) - et al.
Training and transfer-of-learning effects in disabled and normal readers: evidence of specific deficits
J. Exp. Child Psychol.
(1997) - et al.
Alphabetical knowledge from whole words training: effects of explicit instruction and implicit experience on learning script segmentation
Cogn. Brain Res.
(2003) - et al.
Procedural and declarative knowledge of word recognition and letter decoding in reading an artificial script
Cogn. Brain Res.
(2004) - et al.
Implicit memory: intention and awareness revisited
Trends Cogn. Sci.
(2001) - et al.
Relative language proficiency modulates BOLD signal change when bilinguals perform semantic judgments, blood oxygen level dependent
NeuroImage
(2001) - et al.
Assembling and encoding word representations: fMRI subsequent memory effects implicate a role for phonological control
Neuropsychologia
(2003) - et al.
Implicit learning: news from the front
Trends Cogn. Sci.
(1998) - et al.
Effects of word form on brain processing of written Chinese
NeuroImage
(2002) - et al.
Phonological processing during language production: fMRI evidence for a shared production–comprehension network
Brain Res. Cogn. Brain Res.
(2003)