The Word Superiority Effect Experiment - Lab Report Example

WORD SUPERIORITY EFFECT (WSE) ABSTRACT The concept of word superiority effect (WSE) – or word advantage – has undeniably generated considerable interest among linguists, psychologists and more specifically the scholars on cognitive brain functioning. This is specifically true inasmuch as WSE has been observed in different language and across varied experimental conditions. Building on the pioneering observations of Cattell (1886) and the groundbreaking contributions by Reicher (1969) and Wheeler (1970), the ensuing experimentations on WSE have done more than to validate the rudimentary facts that have been already been established so far concerning word advantage. What the present researchers currently present are the many aspects of the phenomenon together with the various factors and variables that come into play with letter recognition. This study involving five hundred ninety-three (593) samples by the students of Psy3304 of the Edith Cowan University intends to simplistically validate the elementary assertion of WSE. In interpreting the data that were culled out of the experimentation, the Paired Sample T-Test was employed. The study concluded that indeed word-context very significantly facilitates letter recognition. INTRODUCTION The concept of word superiority effect (WSE, for brevity) refers to the perceptual facility of accurately identifying letters that are presented in the context of a word than when the same letters are presented alone or in the context of non-words (Galotti, 2009, pp. 81; Goldstein, 2008, p. 64). Also called “word advantage,” WSE means that we see letters better when they fit together with other letters to spell a word than when they occur in isolation or in a jumble array of unrelated letters (Boersma & Hamann, 2009, pp. 295). The first to describe WSE was Cattell (1886), who noted that people are able to perceive more letters from briefly presented words than in meaningless letter strings (Lyddy & Roche-Dwyer, 2008, pp. 1; Grainger et al., 2003, pp. 422). Realizing that word recognition is actually prior to recognition of its constituent letters, Cattell had taken his finding to mean that a letter-based account of visual word recognition is ruled out (Grainger et al., 2003, pp. 422). Reicher (1969) and Wheeler (1970) were responsible in developing the methodology – which, until to date, is still being used – in studying WSE. The so-called Reicher-Wheeler task directs the participants to respond using a two-alternative forced choice task. In this, the participants have to decide which of the two possible letters is present at a given position in a briefly presented string of letters. The Reicher-Wheeler tasks tests only a single letter; thus, the memory load can no longer affect the performance of the participants. Furthermore, the alternative letter is always forming a word – such as D and K in the word WORD – the participants could not get the correct letter by simply making a guess from the partial letter information – e.g., WOR? (Grainger et al., 2003, pp. 422). The WSE has been demonstrated across a range of languages. It is true that early studies on WSE were carried out in languages using the Latin alphabet such as English, French and Italian. But, as Jordan, Paterson and Almabruk (2010) argue, there are also researches that have already established that WSE also applies in non-Latinate language. Actually, they made their experiment in Arabic language – i.e., a cursive language that have an altogether different letter figures from Latin alphabet, and which is read from right to left. Finally, they concluded that WSE that is reported for Latinate language is also observed in Arabic. Lyddy & Roche-Dwyer (2008) have suggested that WSE is present among Irish language speakers and readers. Earlier, Lukatela, Lorenc, Ognjenovic and Turvey (1980) had already established WSE in the Serbo-Croatian orthography. Further, they posited that WSE appears to be indifferent to the linguistic level referenced by the orthography and that substantially WSE resists explanation solely in terms of general properties of the written language. Leh (2009) used WSE as diagnostic tool to examine the modulatory effect of word semantic transparency on the degree to which Chinese bimorphemic compounds are lexically represented as unitized wholes. He found out that WSE is larger for high-frequency than low-frequency Chinese compounds. The WSE has likewise been used to examine bilingual readers’ use of word knowledge as a measure of reading proficiency. Favreau, Komoda and Segalowitz (1980) made use of a modified version of Reicher’s forced-choice procedure to present English (L1)/French bilinguals with first and second language words, anagrams and single letters within ampersands. As a result, they found larger WSE in the bilinguals’ first language that reflected their more skilled and efficient reading. With bilingual speakers and readers of English and Dutch, van Heuven, Dijkstra and Grainger (1998) tested the hypothesis that recognition of words that exclusively belong to one language is affected by the existence of orthographic neighbors from the same or the other language. Their study concluded that increasing the number of orthographic neighbors in Dutch systematically slowed response times to English target words, while an increase in target language neighbors consistently produced inhibitory effects for Dutch and facilitatory effects for English target words. The WSE has also been demonstrated across different experimental conditions. Typically, studies of WSE merely do comparison of letter detection within words with single letters, letters embedded in symbols, or letter strings. The word-letter advantage is the more accurate letter recognition of letters within words – that is, in comparison to “stand alone” letters that occur alone. The word-pseudoword advantage – also referred to as lexicality effect – is the more robust and demonstrated across experimental conditions (Lyddy & Roche-Dwyer, 2008, pp. 2). It has also been documented as the effect holds for pronounceable non-words compared to non-words that do not follow legal letter combinations in a given language (Lyddy & Roche-Dwyer, 2008; pp. 2; Baron & Thurston, 1973). Studies have shown that WSE actually involves lexical access at semantic and phonological levels (see Grainger et al., 2003), as well as word knowledge and experience (Lyddy and Roche-Dwyer, 2008, pp. 2). Researching on readers with acquired dyslexia that affects visual word forms, Hildebrandt et al. (1995) found out that the magnitude of the word-non-word word superiority effect is predictable by one’s performance on a lexical decision task. Martin et al. (2006) have clarified which of the early processing stages of visual word recognition are modulated by top-down lexical effects. They proposed that visual word form representations can constrain letter identification at pre-lexical stage, such as during the extraction of letter-shape information. Likewise, they were able to show that this facilitatory top-down effect is in fact sensitive to stimulus exposure duration. Acha and Perea (2008) studied what is called robust length effect – or the inability of children to read long words faster than short words – that vanish in the adults. They concluded that, in letter recognition, children actually evolve from letter-to-letter reading (to a direct lexical access). The WSE has made strong case for the robust connection between perception and context. With the evidences that the word context is an influence in our perception of letters, WSE has been established to be perceptual in nature. Readers, for example, are said to tend to miss letter “t” more when it is embedded in the word such as “the” or in other high-frequency words than if it is embedded in a non-word. This arises out of our tendency to “unitize,” or to group letters into higher-order units such as groups of letters or entire short words making it difficult to identify letters in high-frequency words. A similar result is found in Chinese language involving target radicals in characters, pseudocharacters and noncharacters. It has been found out that radicals are identified better when they are presented in characters than in pseudocharacters, and better in pseudocharacters than in noncharacters (Carroll, 2008, pp. 97). Moreover, the WSE has led to numerous models of letter perceptions that incorporate context-guided processes (Galotti, 2009). In this paper, we shall limit our discussion with just two competing models of perceptual processing – i.e., the dual-route and the connectionist models. The dual-route model holds that we have a couple of different ways of converting a printed word or letter to speech. The lexical route is the process by which a printed letter, character, or a word activates the entry for the corresponding word in our internal lexicon. The non-lexical route refers to a system of rules that specifies the relationship between letters and sounds, allowing the readers to correctly pronounce nonwords as well as irregular words that disobey the rules of the language. Hence, the dual-route model’s rule system (the lexical route) and memory system (the non-lexical route) that are governed by different principles and are acquired in different ways (Carroll, 2008, p. 98). The connectionist or parallel-distributed-processing model attempts to explain the computational mechanisms underlying various psychological skills such as language production, the acquisition of grammar, and reading. This model consists of three layers – the orthographic representing spelling or the visual features of words, the phonological representing pronunciation and consisting of phonemes or phonological features, and the semantic representing meaning. The orthographic layer consists of letters or visual features of words. Unlike the dual route model, the connectionist model maintains that there is only a single route for perceptual processing. Too, it assumes neither an existence of mental lexicon nor an assumption of phonological or orthographic rules. What it presumes is that the learner starts with no knowledge of the relationship between print and sound. And, it is through experience that the learner gradually comes to develop weights between letters and sounds approximating those of a mature learner. It follows then that word pronunciation is not represented in terms of linguistic rules but by a system of connections between different layers (Carroll, 2008, pp. 98-99). Moreover, WSE has influenced role of top-down processes – e.g., the interactive activation model (IAM) – in speeding letter detection in word conditions. This means that letters are identified much faster when it is presented within a word on account of the facilitatory effect of strong word level activation feeding back to letter level analysis. In the Reicher-Wheeler task, word level activation facilitates the correct response alternative and – at the same time – inhibits the activation of the incorrect choice. Pseudoword advantages occur on account of the fact that the legal orthographic patterns, which are shared with real words, activate to influence bottom-up processing. Favoring a non-interactive account, activation verification model (AVM) accounts that WSE results from letter level or word level activation. And, in what appears as synthesis of the IAM and AVM, the dual read-out model maintains that letter identification follows critical activation at a letter response or word response threshold. Now, for this model of context-guided process, WSE occurs as words can reach the critical activation level required at shorter presentation intervals. The letter in a particular position can be identified once the word has been activated. At longer exposures, though, sufficient input needs to be provided so as to allow efficient data-driven letter identification without the benefit of a word context – with WSE due to degraded stimuli instead of brief exposures ( see Lyddy and Roche-Dwyer, 2008, pp. 2). METHOD The experiment used 593 separate test stimuli consisting of target letter(s) in word and target letters in isolation. The design was essentially very simple, with just two level of a within subject variable. The experiment was computer-aided, with the participating students of Edith Cowan University (ECU). All the instructions that were provided the subjects were provided by the computer system. Finally, the data that was generated from the experiment was analyzed by conducting a Paired Sample T-Test – an inferential statistics that assess whether the means of two groups are statistically different from each other (Archambault, 2002). It was done using PASW/SSPS. RESULT Following is the result of the Paired Samples T-Test for the variables for this experimentation. The mean scores for the word and letter are compared. We specifically want to determine whether the letter in the context of a word is indeed more recognizable that when it is made to appear alone. The descriptive statistics for both variables are presented in Table 1. Note that the scores in “word” variable – i.e., the mean – are higher. Table 1 The descriptive statistics of the paired variables Mean N Std. Deviation Std. Error Mean Pair 1 word letter 73.3340 72.1537 593 593 13.84322 13.11354 .56847 .53851 The correlation between the two variables is presented in Table 2. Table 2 Paired samples correlation N Correlation Sig. Pair 1 word & letter 593 .562 .000 There is a strong positive correlation between the paired samples. This means that the people who did well in letter recognition in the context of a word also were able to do the same on stand-alone letter(s). Table 3 presents the results of the Paired Samples T-Test. This test is based on the difference between the two variables. Under the column(s) “Paired Differences,” we see the descriptive statistics for the difference between the two variables. Table 3 Paired Samples Test Paired Differences Mean Standard Deviation Standard Error Mean 95% Confidence Interval of the Difference t df Sig. (2-tailed) Lower Upper Pair 1 Word-Letter 1.18029 12.62725 .51854 .16189 2.19869 2.276 592 .023 The T value is 2.276. There is 592 degrees of freedom. And the significance is .023. Now, if the significance value is less than .05, there is significant difference between the two variables. However, if the significance value is greater than .05, there is no significant difference (Archambault, 2002). In this experimentation, we have established that the significance value is very significant. Hence, we say that there is significant difference between the two variables scores or means. There is substantial difference in the recognition of letter in the context of a word and when the letter is made to appear alone. DISCUSSION This experimentation has established that that there is substantial difference insofar as the recognition of letter in the context of a word and/or a single letter that occurs alone. Actually, this result is neither novel nor revolutionary. As it is said at the outset of this paper, the experimentation intended simply to make the students of ECU undergo the process that has established the fundamental assertions of WSE. Besides, the simple design of the experimentation clearly rules out that it actually wanted to arrive at any breakthrough on the subject matter. Thus, it should not surprise anyone that the conclusion of this experimentation is merely in parallels with those previously reported particularly for visual stimuli (see Rogers, Thistle and Nielson, 1976) – the methods of which are described by Prinzmetal and Silvers (1994). Accordingly, the methods of one of the most popular of these paradigms consist of presentation of stimulus for a fraction of a second and which are followed by some type of visual mask. The stimulus may consist of alphanumeric characters, objects or even faces. The subject’s task is to report some aspect of the stimulus, such as the identity of a letter, a digit, a facial feature, or a line segment. Such methods essentially formed the steps that the participants in this simple experimentation were asked to undertake. The experimentation followed the paradigm of Reicher-Wheeler task – which Jordan and Bevan (1996) and Jordan and Patching (2004) particularly found out to be very useful in studying word-letter phenomenon (WLP). And, this simple experimentation has concluded on WLP! Johnston and McClelland (1973, pp. 375) define WLP as the facility or the correctness and easiness of picking up more information from each letter in a word than from a single letter presented alone. It is distinguished from word apprehension effect (WAE), where words are perceived more accurately than are strings of an equal number of unrelated letters. The existence of WLP is actually a suggestion that, when people perceive words, the whole is in some way more than the sum of its parts. In fact, even the conscious effort to provide pre-cues to help one out in going through the forced choice paradigm of Reicher-Wheeler task with tachistoscopic exposures is found to have detrimental effect as the subject is unnecessarily induced to search for where in the word the forced choice would be plausible (Holender, 1979). If only WLP were not resisting further interpretation (see Wheeler, 1970), it could have led the way to providing the important clue to the nature of the reading process (Johnston & McClelland, 1973). Be that as it may, a rundown of what McClelland and Rumelhart (1981, pp. 375-377) called basic findings on the role of contexts in letter perception or recognition fittingly closes this discussion. They maintained that the perceptual advantage of letters in (the context of) the words is explainable by the fact that they are indeed more perceptible. That is, they receive more activation than representations of either single letters or letters in an unrelated context. Likewise, they held that WLP is actually independent of the subject’s familiarity of the word as visual configuration. Hence, the word shape is irrelevant (see Larson [2004] who has diverging view on the effect of word shape to letter perception). Further, they emphasized the role of patterned masking. Over single letters and nonwords, the WLP appears to depend upon the visual masking conditions used. Moreover, the word advantage is said to apply to pronounceable nonwords – e.g., REET or MAVE. Letters in these pronounceable nonwords – also called pseudowords – have a large advantage over letters in unpronounceable non-words (or the unrelated letter strings). Finally, McClelland and Rumelhart noted the absence of effects of contextual constraint under patterned-mask conditions. Considered to be important, their finding is that letters in highly constraining word contexts have little or no advantage over letters in weakly constraining word contexts under the distinct-target/patterned-mask conditions that produce a large word advantage. REFERENCES Acha, J. and Perea, M. (2008). The effects of length and transposed-letter similarity in lexical decision: evidence with beginning, intermediate and adult readers. British Journal of Psychology, 99 (2), pp. 245-264. Archambault, S. (2002). Paired sample t-test. Retrieved on 27 August 2011, from http://www.wellesley.edu/Psychology/Psych205/pairttest.html Baron, J. and Thurston, I. (1973). An analysis of the word-superiority effect. Cognitive Psychology, 4, pp. 207-228. Boersma, P. and Hamann, S. (2009). Phonology in perception. Berlin: Mouton de Gruyter. Carr, T., Davidson, B. and Hawkins, H. (1978). Perceptual flexibility in word recognition: strategies affect orthographic computation but not lexical access. Journal of Experimental Psychology: Human Perception and Performance, 4, pp. 674-690. Carroll, D. (2008). Psychology of language, 5th Ed. Belmont: Thomson Higher Education. Cattell, J.M. (1886). The time taken up by cerebral operations. Mind, 11, 220-242. Favreau, M., Komoda, M. and Segalowitz, N. (1980). Second language reading: implications of the word superiority effect in skilled bilinguals. Canadian Journal of Psychology, 34, pp. 370-380. Galotti, K.M. (2009). Cognitive psychology: in and out of laboratory. Toronto: Nelson Education, Ltd. Goldstein, B. (2008). Cognitive psychology: connecting mind, research and everyday living, 2nd Ed. Belmont: Thomson Higher Education. Grainger, J. et al. (2003). Word superiority, pseudoword superiority, and learning to read: a comparison of dyslexic and normal reader. Brain and Language, 87, pp. 432-440. Hildebrandt, N., Caplan, D., Sokol, S. and Terreano, L. (1995). Lexical factors in the word superiority effect. Memory and Cognition, 23, pp. 23-33. Holender, D. (1979). Identification of letters in words and of single letters with pre- and post-knowledge vs. post-knowledge of the alternatives. Perception and Psychophysics, 25 (4), pp. 313-318. Johnston, J. and McClelland, J. (1973). Visual factors in word perception. Perception and Psychophysics, 14 (2), pp. 365-370. Jordan, T.R., and Bevan, K.M. (1996). Position-specific masking and the word-letter phenomenon: re-examining the evidence from the Reicher-Wheeler paradigm. Journal of Experimental Psychology: Human Perception and Performance, 22 (6), pp. 1461-1433. Jordan, T.R. and Patching, G.R. (2004). What do lateralized displays tell us about visual word perception? A cautionary indication from the word-letter effect. Neuropsychologia, 42 (11), pp. 1504-1514. Jordan, T.R., Paterson, K.B., and Almabruk, A. (2010). Revealing the superior perceptibility of words in Arabic. Perception, 39, pp. 426-428. Larson, K. (2004). The science of word recognition. Microsoft Typography. Retrieved 29 August 2011, from http://www.microsoft.com/typography/ctfonts/wordrecognition.aspx Laszio, S. and Federmeier, K.D. (2007). The acronym superiority effect. Psychonomic Bulletin and Review, 14 (6), pp. 1158-1163. Lee Woon Mok (2009). Word superiority effect as a function of semantic transparency of Chinese bimorphemic compound words. Language and Cognitive Processes, 24 (7/8), pp. 1039-1081. Lukatela, G., Lorenc, B., Ognjenovic, P. and Turvey, M. (1980). A word superiority effect in a phonetically precise orthography. Status Report on Speech Research, SR 63/64, pp. 263-273. Lyddy, F. and Roche-Dwyer, C. (2008). A bilingual word superiority effect in Irish speakers. Written Language & Literacy, 11 (1), pp. 1-14. Martin, C.M. et al. (2006). Perceptual and lexical effects in letter identification: an event-related potential study of the word superiority effect. Brain Research, 1098, pp. 153-160. McClelland, J. and Rumelhart, D. (1981). An interactive activation model of context effects in letter perception: Part I an account of basic findings. Psychological Review, 88 (5), pp. 375-408. Massol, S. et al. (2011). When less is more: feedback, priming, and the pseudoword superiority effect: Brain Research, 1386, pp. 153-164. Prinzmetal, W. and Silvers, B. (1994). The word without the tachistoscope. Perception and Psychophysics, 55 (3), pp. 296-312. Reicher, G.M. (1969). Perceptual recognition as a function of meaningfulness of stimulus material. Journal of Experimental Psychology, 81 (2), pp. 275-280. Rogers, J., Thistle, A. and Neilson, B. (1976). Word-letter phenomenon with speech stimuli: a word-segment effect. Journal of Acoustical Society of America, 60 (1), pp. 27-28. van Heuven, W., Dijkstra, T. and Grainger, J. (1998). Orthographic neighborhood effects in bilingual word recognition. Journal of Memory and Language, 39, pp. 458-483. Wheeler, D.D. (1970). Process in word recognition. Cognitive Psychology, 1, pp. 59-85. Read More

The so-called Reicher-Wheeler task directs the participants to respond using a two-alternative forced choice task. In this, the participants have to decide which of the two possible letters is present at a given position in a briefly presented string of letters. The Reicher-Wheeler tasks tests only a single letter; thus, the memory load can no longer affect the performance of the participants. Furthermore, the alternative letter is always forming a word – such as D and K in the word WORD – the participants could not get the correct letter by simply making a guess from the partial letter information – e.g., WOR?

(Grainger et al., 2003, pp. 422). The WSE has been demonstrated across a range of languages. It is true that early studies on WSE were carried out in languages using the Latin alphabet such as English, French and Italian. But, as Jordan, Paterson and Almabruk (2010) argue, there are also researches that have already established that WSE also applies in non-Latinate language. Actually, they made their experiment in Arabic language – i.e., a cursive language that have an altogether different letter figures from Latin alphabet, and which is read from right to left.

Finally, they concluded that WSE that is reported for Latinate language is also observed in Arabic. Lyddy & Roche-Dwyer (2008) have suggested that WSE is present among Irish language speakers and readers. Earlier, Lukatela, Lorenc, Ognjenovic and Turvey (1980) had already established WSE in the Serbo-Croatian orthography. Further, they posited that WSE appears to be indifferent to the linguistic level referenced by the orthography and that substantially WSE resists explanation solely in terms of general properties of the written language.

Leh (2009) used WSE as diagnostic tool to examine the modulatory effect of word semantic transparency on the degree to which Chinese bimorphemic compounds are lexically represented as unitized wholes. He found out that WSE is larger for high-frequency than low-frequency Chinese compounds. The WSE has likewise been used to examine bilingual readers’ use of word knowledge as a measure of reading proficiency. Favreau, Komoda and Segalowitz (1980) made use of a modified version of Reicher’s forced-choice procedure to present English (L1)/French bilinguals with first and second language words, anagrams and single letters within ampersands.

As a result, they found larger WSE in the bilinguals’ first language that reflected their more skilled and efficient reading. With bilingual speakers and readers of English and Dutch, van Heuven, Dijkstra and Grainger (1998) tested the hypothesis that recognition of words that exclusively belong to one language is affected by the existence of orthographic neighbors from the same or the other language. Their study concluded that increasing the number of orthographic neighbors in Dutch systematically slowed response times to English target words, while an increase in target language neighbors consistently produced inhibitory effects for Dutch and facilitatory effects for English target words.

The WSE has also been demonstrated across different experimental conditions. Typically, studies of WSE merely do comparison of letter detection within words with single letters, letters embedded in symbols, or letter strings. The word-letter advantage is the more accurate letter recognition of letters within words – that is, in comparison to “stand alone” letters that occur alone. The word-pseudoword advantage – also referred to as lexicality effect – is the more robust and demonstrated across experimental conditions (Lyddy & Roche-Dwyer, 2008, pp. 2). It has also been documented as the effect holds for pronounceable non-words compared to non-words that do not follow legal letter combinations in a given language (Lyddy & Roche-Dwyer, 2008; pp.

2; Baron & Thurston, 1973). Studies have shown that WSE actually involves lexical access at semantic and phonological levels (see Grainger et al.

Read More