Lexical Processing with Deaf and Hearing

3 downloads 22 Views 175KB Size Report
representation in order to access the word in the lexicon? .... 15-60 ms. Then this masked word is replaced with a simple pattern mask ( ..... 17. soal jeck sole.

Lexical processing… 31

LEXICAL PROCESSING WITH DEAF AND HEARING: PHONOLOGY AND ORTHOGRAPHIC MASKED PRIMING Jody H. Cripps, Kara A. McBride, and Kenneth I. Forster University of Arizona This preliminary study investigates lexical retrieval of written English with native English speakers (i.e., hearing) and American Sign Language (ASL) signers (i.e., deaf) by using masked priming techniques. Repetition and pseudohomophone priming were tested. These types of priming were employed in order to investigate phonological and/or orthographic effects. A significant facilitative phonological effect for hearing participants and a significant inhibitory orthographic effect for deaf participants were found, showing clearly that the modality differences of participants who use sign or spoken languages are a factor in the lexical processing of written English.

INTRODUCTION Exposure to language for most people is first through the spoken word, and only later in life do people learn to read. Thus, early in life, words are accessed exclusively through phonology. Throughout one’s lifetime, lexical access continues to be accomplished phonologically when spoken language is being processed. But when written words are being processed, does the mature reader achieve lexical access directly from the visual cues, or is it necessary to recode the visual representation into a phonological representation in order to access the word in the lexicon? Everyone is familiar with “hearing words in one’s head” when reading. Is this part of the process of lexical access? Or is this instead a post-access phenomenon, a checking mechanism perhaps? Or might it be that words are searched through two parallel processes, one using the visual cue and one using a recoded phonological representation? This paper describes an experiment that was designed to examine phonological and orthographic issues in lexical processing. It is important to keep in mind that hearing people read in their first language (L1), which is based on their spoken language. This does not apply to deaf people because they use sign languages. There is lack of a standard written version of American Sign Language (ASL) (Baker-Shenk & Cokely, 1980; Supalla & Bahan, 1992); a deaf person always reads in his or her second language (L2). Thus, for the deaf participants, any priming with written words occurs within the L2 (e.g., written English). Modality differences (i.e., sign versus spoken) here play an important role which do not apply to previous bilingual priming studies with hearing people. Given these modality differences, this experiment explores what potential differences there are in a hearing population’s and a deaf population’s responses to written English words; written English is used

Arizona Working Papers in SLAT – Vol. 12

32 Cripps, McBride, Forster

instead of ASL graphemes (see Supalla, Wix, & McKee, 2001). It is important to note that the system of ASL graphemes is new and in the developing stages. It has been used with deaf children for a decade but has not been used with deaf adults. Therefore there is no official writing system in ASL, and in order to study how the deaf process written language, written English must be used. PHONOLOGICAL AND ORTHOGRAPHIC MASKED PRIMING STUDIES There are a number of studies that suggest that people use both phonological and visual cues in visual word recognition (e.g., Ferrand & Grainger 1993; Lukatela, Frost, & Turvey, 1998; Grainger & Ferrand, 1994; Lukatela & Turvey, 1994; Van Orden, Pennington, & Stone, 1990). These studies showed facilitation in processing a target word when the word that came before it, the prime, overlapped phonologically with the target. However, because these studies used languages with alphabetic writing systems, in which the written versions of words are based on the phonological characteristics of the words, it is very hard to tease apart the influence of visual and phonological cues in reading. That is, the homophonic words sail and sale are identical in phonology, but they also overlap in orthography, so if faster responses are found to sale when it is primed by sail, it is difficult to decide which factor is responsible. Conventionally, investigators use a control prime that is matched for orthographic overlap with the phonological prime (e.g., saul), but in the absence of any theory of orthographic similarity, one cannot be certain that the matching is accurate. One way to find out is to carry out the experiment with a sample of deaf participants. Because deaf participants could not be employing phonological cues from the spoken language in visual word recognition, any tendency to obtain more priming in the homophonic condition (sail-SALE) than in the control condition (saulSALE) would indicate that this result is due to inadequate orthographic matching. Thus, deaf participants provide a way of calibrating for orthographic similarity without contamination from phonological properties. One of the earliest experiments to investigate the question of what role phonology plays in visual word recognition was a lexical decision task experiment by Rubenstein, Lewis, and Rubenstein (1971). In a lexical decision task, subjects indicate, as quickly and accurately as possible, whether each string of letters that they see, shown one by one on a screen, is a word or not. Rubenstein, et al. (1971) found a delay effect with pseudohomophones; subjects were slower to classify as nonwords items that were homophonic with real words (for instance LEEF). It was postulated that items are recoded phonologically and that this recoding is used in attempting lexical access through the phonological input system. In the case of a pseudohomophone, a word with a matching phonological representation is located, but then a postaccess check detects a mismatch between this word’s orthography and that of the input, and it is at this stage that the time cost occurs.

SLAT Student Association

Lexical processing… 33

However, post-access processes are not the primary object of investigation. The question of whether phonological recoding is a necessary part of lexical access must be investigated at the pre-lexical stage. It is the preaccess phenomena that can properly be said to lead to lexical access, while post-access processes can involve other mechanisms, including episodic memory, decision making, strategy use and guessing. The difference between pre-access and post-access phenomena is a matter of milliseconds. One way to get at these questions, and the method employed in the present study, is with the masked priming paradigm (Forster & Davis, 1984). In the forward masked priming paradigm, before subjects see the letter string upon which the decision is made (the target), another word, called the prime, is flashed for a time so brief (usually 30 to 60 ms) that the subject is not consciously aware of seeing it. Nonetheless, masked primes have been shown to have an effect on reaction times to target words. When the masked prime and the target are the same, reaction times are reliably quicker (Bodner & Masson, 2004). Other relationships between masked primes and targets have resulted in quicker reaction times. Semantic priming has been demonstrated (Perea & Gotor, 1997). Orthographic overlap (Andrews, 1997) and phonological overlap (Perfetti, Bell, & Delaney, 1988) have also been shown to result in priming. It is assumed that the way in which these unconsciously perceived primes affect the subject’s response time is due only to the unconscious and automatic processes of lexical access and not to other processes. Depending on which experimental task the masked priming paradigm is used with, it may still be unable to isolate issues enough to answer the question of whether or not phonological recoding plays an essential role in visual word reading. For example, with Chinese, in which there is no systematic connection between phonology and the written system, phonology has been shown to affect reaction times in the naming task but not the lexical decision task (Shen & Forster, 1999). Subjects in a naming task read words out loud, which means that phonological recoding is required for this task. This requirement makes the naming task fundamentally different from silent visual word recognition precisely in terms of phonology’s role in the performance of the tasks. The lexical decision task is a better method for the present inquiry. Lexical access is required in order to perform this task, but phonological recoding is not necessarily required. Lukatela, Frost, and Turvey (1998) reported evidence for phonological recoding in a lexical decision task that used the masked priming paradigm. The subjects in this experiment responded more quickly to items when there was phonological overlap between the prime and the target. However, the effect was significant only when the experiment was done with the additional condition of dim lighting. As they say in a footnote, “It was only by reducing the room illumination that provided by a single desk lamp at floor level that we could obtain reliable priming differences” (p. 671). The studies discussed thus far used forward masked priming. In contrast, Perfetti, Bell, and Delaney (1988) argued for using backward masked

Arizona Working Papers in SLAT – Vol. 12

34 Cripps, McBride, Forster

priming. That is when the target word is presented for a very brief amount of time (usually 15-30 ms) and then is followed by a nonword, which appears for 15-60 ms. Then this masked word is replaced with a simple pattern mask (such as #######). Usually, participants perceive only the target word and are not consciously aware of the masked nonword. Using this method, Perfetti, et al. (1988) found that phonological similarity had a strong facilitatory effect. Similarly, Frost (2003) conducted a masked priming study with Hebrew using both forward masked priming and backward masked priming. From his findings, he then claimed that there is a significant phonological effect. However, word and nonword targets exhibited very similar effects. If the phenomenon of study were a purely lexical access issue, words and nonwords should display different effects. Other studies’ results instead provide support for the theory that an orthographic route is the primary route to visual word recognition (Davis, Castles, & Iakovidis, 1998; Forster & Davis, 1984; Forster & Taft, 1994). The search model (Forster, 1976) proposes that visual word recognition relies on orthography. Only after lexical access has been achieved through the orthographic cues does information about the word’s phonology and semantics become available. Davis, et al. (1998) conducted a masked priming study with children and adults to determine whether phonological similarity affected priming. They found little evidence for a phonological priming effect with either children or adults. They did find that the younger the children were, the more there was a tendency in the data to show a phonological priming effect, but it did not reach significance. Davis, et al. (1998) noted that the slow readers perhaps use phonological information more often than the more skilled readers do. Still, they found that the repetition priming effect (e.g., washWASH) was faster than the sound-same prime (e.g., wosh-WASH). Thus, they claimed that these children’s primary process for lexical access was the orthographic one. Another group of researchers has proposed the use of a connectionist framework of lexical processing which employs both phonological and orthographic routes (Plaut, 1997; Seidenberg & McCelland, 1989; Van Orden et al., 1990). This framework represents phonological, orthographic, and semantic units, which are all active during lexical processing, and all influence each other. In between these three kinds of representational units are hidden units. The hidden units’ function is to meditate the representational units (Seidenberg & McCelland, 1989). Lexical decision is based on the interaction of all routes, and none of these routes is the primary route in this framework. Simultaneous use of phonological and orthographic information need not necessarily require a connectionist framework, however. Coltheart, Davelaar, Jonasson, and Besner (1977), finding pseudohomophones to act as primes (for example when leef primes LEAF), proposed the dual route model. By this model, the pronunciation of visually presented words is calculated by grapheme-phoneme correspondence (GPC) rules, while at the same time the word is accessed directly by means of the orthographic form. The word corresponding to the computed pronunciation and that which was directly SLAT Student Association

Lexical processing… 35

accessed by means of the orthographic form are both activated; one or the other may reach threshold first, depending on the nature of the items and the task. In short, there are three major possibilities for the role of phonology in visual word recognition. It might be that phonology is the primary route for all word recognition. It might instead be that orthography is the primary route, with phonology as a post-access check or an alternative route. Alternatively, phonology and orthography may both be utilized even in the earliest stages of lexical access. In all of the studies mentioned above, the question of whether phonology or orthography is the primary route to lexical access remains clouded by the fact that in the languages tested, orthography reflects phonology. When a homophone or pseudohomophone primes a word, it may be due to the overlap in orthography. Shen and Forster (1999) addressed this issue when they tested for priming effects in homophones in Chinese. Because no priming effects were found in the lexical decision task, it appears that native Chinese speakers do not employ a phonological route in lexical access in visual word recognition. Still, this does not mean that the same is necessarily true for visual word processing for people whose first language is English or another language that uses a phonologically based writing system. Hearing people are able to use both phonological and orthographic information to control lexical processing while reading a word, while deaf people cannot. How deaf people’s lexical processing differs from hearing people’s lexical access has yet to be fully examined. To test whether access to spoken-language phonology alters the process of visual word recognition, a study comparing hearing and deaf people’s lexical decision task is needed. This present study looked at hearing versus deaf people’s reaction times to words when they were preceded by phonologically similar primes (blooBLUE) and when they were not. The assumption was that if the effects that have been ascribed to phonological processing are in fact simply a product of orthographic overlap, then the hearing and deaf participants’ results would be quite similar in this part of the experiment. If on the other hand the results showed to find significant differences in the two populations’ reaction times, this would lend support to the claim that phonology plays an essential role in visual word reading for hearing people. THE EXPERIMENT Materials and Design The experiment was designed to investigate the issue of phonological priming. Phonologically similar prime-target pairs were taken from Perfetti, et al.'s (1988) experiment in which evidence for phonological priming was said to be found. The original experiment in 1988 employed backward masked priming. In the current experiment, the same materials were used but with forward masked priming (see Appendix A for the list of word pairs). Within this section of the experiment, repetition priming was also tested. Repetition

Arizona Working Papers in SLAT – Vol. 12

36 Cripps, McBride, Forster

priming, when the prime and target are equivalent, should produce very strong priming (e.g., Forster & Davis, 1984). Words for this section were chosen from the MRC Psycholinguistic Database web site, http://www.psy.uwa.edu.au/mrcdatabase/uwa_mrc.htm. This web site allowed the present study restricts word choice to more frequent words (see Appendix B for word list). All nonwords for the experiment were taken from the ARC Nonword Database (http://www.maccs.mq.edu.au/~nwdb/). The choice of nonwords was restricted to letter combinations that are phonologically possible in English. Two counterbalanced lists were constructed such that if a target was preceded by its pseudohomophone/identity prime on List A, it was preceded by its control prime on List B, and vice versa (see Table 1 for examples). No word appeared twice within the materials. For both conditions, there were 12 targets per list with related primes, or exemplars, and 12 targets per list with unrelated primes. This gives a total of 48 pairs of words per list. Additionally, there was an equal number of nonword pairs, and there was also a block of practice items at the beginning of the experiment. After the practice items, pairs were presented at random to the subjects, using DMDX software developed at the University of Arizona by J.C. Forster (Forster & Forster, 2003). The primes were displayed for 67 ms after a 606 ms hash mark display. Response times (RTs) were recorded to the nearest millisecond. Also, a questionnaire was administered to the participants. The purpose of the questionnaire was to gather information on gender, age, education level, and language backgrounds. This questionnaire was given to the deaf participants because it is quite common that deaf people have a variety of language backgrounds and it is important to know when they acquired or learned ASL. This gives a general picture of the deaf participants’ language background. Table 1. Breakdown of Item Types and Numbers, with Examples Condition

Examples List A

Identity

Pseudohomophone

List B

12 exemplars sample-SAMPLE

12 control items caught-SAMPLE

12 control items victory-HIGHWAY

12 exemplars highway-HIGHWAY

12 exemplars bloo-BLUE

12 control items scron-BRAKE

12 control items caft-BLUE

12 exemplars braik-BRAKE

SLAT Student Association

Lexical processing… 37

Subjects were debriefed after completing the lexical decision task. Most hearing participants showed surprise to learn about the presence of the masked primes, but three of them claimed to have been aware that there was something between the display span and the words on which they were told to make a lexical decision. One participant claimed to have known that it was letters. Most of the deaf participants had noted that there was something unusual in the display span, but none of them said it was a string of letters. Once they were told about how the masked priming procedure works, most of them said that, when they saw it again, they could detect the presence of flashing letters before the target word display. Participants There were two groups of participants: hearing people and deaf people. The hearing participants were twenty native English speakers, eight of whom were female. Their ages ranged from 18 to 62. Twelve of them were advanced speakers of at least one other language (Spanish, French, Portuguese, Japanese, Korean, and Mandarin Chinese). One participant had also a very basic knowledge of ASL. The deaf participants were 14 ASL users from the deaf community from the southwestern United States. Half of these participants were female. The age of these participants ranged from 23 to 53 years old. All of them except for one were born deaf. The one exception became deaf before the age of 18 months. Six of them were exposed to ASL prior to three years of age. All of these deaf participants have used ASL more than ten years. All of them also were either graduates of or currently enrolled in college or university (AA to Ph.D.) and had mastered written English as their L2. RESULTS Only the participants’ responses to words were analyzed in the present study. Data from trials in which the subject responded incorrectly were discarded. When the subject’s response time was far from his or her mean response time, those times were replaced with the value equal to the cutoffs 2 SD units above or below the mean for that participant. Table 2 shows the result of the repetition and phonological priming conditions for hearing and deaf participants. The facilitatory effect of repetition priming is significant for both groups and of a similar magnitude, 44 ms for hearing subjects and 42 ms for deaf subjects. This effect was highly significant in the hearing population in both the subject analysis, F1(1,18)=55.72, p

Suggest Documents


Mary Poppins Returns (2018) HDCAMRip Full Movie Watch Online English Full Length Film | Running Man Episode 22 | 亚特兰大第二季