Carpenter (1980) is the most commonly used tool for WMC. Other span tasks for WM include operation span and counting span. Among all types of WM tools, RSTs are found to be highly related to RC. Daneman and Carpenter (1980) define RSTs as dual-tasks that require participants to complete two simultaneous missions. RSTs require participants to
“fulfill both processing and storage requirement” simultaneously (Conway et al., 2005, p.581). Namely, two tasks have to be done by the participants simultaneously: to read a series of sentences and to recall or recognize the final words later on. Participants’ WMCs were the maximum number of ending words that he or she recalled or recognized from multiple-choice options.
A RST adopted from the study of Harrington and Sawyer (1992) was administered in
33
the current study. The RST consisted of 42 unrelated sentences with 21 grammatical sentences and 21 ungrammatical sentences. All 42 sentences were randomly assigned into three trials. Each trial had 2 sentences, 3 sentences, 4 sentences and 5 sentences; the number of sentences in each trial was increasing from 2 to 5. There were 42 sentences in total. Every sentence ended with different words. Conway and his colleagues (2005) indicate that the item/sentence number of “two to five or six” was sufficient to test WMC span in order to “create the potential for ceiling effects among those participants in the upper end of the performance distribution” (p.773). The adopted list of 42 sentences was shown in the Appendix A. A brief summary of relevant literature about RST was in Table 3.1 below and a sample test of RST was in Table 3.2 in the following.
Table 3.1 Summary of Literature Review on RST
Daneman and Carpenter (1980) 1. RST, a dual-task, could better assess the influence of WMC on RC.
2. RST is able to predict participants’ prose comprehension skills.
Friedman, & Miyake (2005) The dual-task design of RST requires participants to complete two sub-tasks
simultaneously: one is to store information and the other is to process it.
Alptekin & Ercetin (2011) The sub-task of information storage in RST requires participants to memorize the ending words of sentences or to recognize them. The other sub-task of information process requires the participants to judge the grammaticality.
34
Table 3.2 Example Sentences of RST with WMC=3
These 42 sentences, each of which consists of 11-13 words, were all simple
sentences in active voice, not passive voice. The sentences were 3 to 4 words shorter and syntactically simpler than the sentences in the RST used by Daneman and Carpenter (1980). Harrington and Sawyer (1992) suggested that sentences of RSTs be shorter in order to avoid the possible “floor effect” (p.28). Floor effect refers to extremely low scores for individual participants because of task difficulty in studies. Therefore, true WMC span for each one can be measured (Harrington & Sawyer, 1992). Besides, each of the 42 sentences ended with different words, and the ending words were designed to
“avoid phonologically similar words in the same set” (Harrington & Sawyer 1992, p.28).
For instance, minimal pairs of ball and fall, and around and ground were not included in the same set. In terms of grammaticality, the grammatical sentences were correct in terms of semantic and syntactic sense while the ungrammatical sentences were designed by reversing the last four to six words of grammatically correct sentences. According to Harrington and Sawyer (1992), syntactic accuracy is considered as an intruder into participants’ storing performance and it can also prevent participants from merely trial 1
All morning the two children sat and talked under a tree.
*He played baseball all day at the park and sore got a arm.
At night the prisoners escaped through a hole in the wall.
trial 2
*His younger brother played guitar a rock and roll in band.
The people in northern Europe always like to travel by train.
*The last thing he did was take to a nice hot bath.
trial 3
The clerk in the department store put the presents in a bag.
*I saw a child and her father the river near playing ball.
*Suddenly the taxi opened its the door in front of bank.
35
memorizing the ending words without truly understanding the sentences. For example, grammatical sentences are like “The clerk in the department store put the present in a
bag.” By reversing the last four to six words, ungrammatical sentences are like “*The last thing she did drink a was to hot cup of tea” (Harrington & Sawyer 1992, p.29).
5These 42 sentences came in three trials (see exemplary RST trials in Table 3.2). All the three trials had sentences of 2, 3, 4 and 5 increasing in the number of the sentences (see Appendix A for all the 42 sentences). The participants took the RST independently and were informed beforehand to expect the increasing number of sentences. Each sentence was displayed on one slide and shown only once. Every sentence was shown on one PowerPoint slide for 10 seconds. Blank slides were inserted in slides to separate sets of 2, 3, 4, and 5 sentences. When seeing the blank slides, the participants wrote down the ending words they remembered or selected the ending words they remembered.
Meanwhile, each participant was asked to judge the grammaticality of sentences while they read the sentences. They read and wrote down their judgment on answer sheets.
Grammatical judgment used here “grammatically represents the processing measurement of their reading span” (Alptekin & Ercetin 2011, p.249). During the presentation of the sentences, participants read, processed, and judged the grammaticality. Right after they saw the blank sheet at the end of each set, they needed to recall or recognize the final words in the same order of the presentation. To recall, the participants wrote down the ending words. As for the recognition-RST, participants judged the grammaticality when they read them. The answer sheet for the recognition-RST was a multiple-choice test, and four options were provided for each ending word of sentences. The participants chose the ending words after 2 sentences, 3 sentences, 4 sentences and 5 sentences. There was a warm-up practice of three sentences demonstrated in the same way of the later RST for
5Examples of un/grammatical sentences with syntactic accuracy as intruders into WM storage are “He looked across the room and saw a person holding a gun” and “*The girl picked up her bag and down to went the gym(Alptekin & Ercetin, 2011, p.249)”
36
the participants at the very beginning of answering the RST. The participants could understand how to complete the RST by doing the warm-up practice.
As far as the scoring methods of RST are concerned, two out of four commonly used methods (e.g., total words, proportion words, correct sets words, and truncated span) were discussed and testified to have satisfactory reliability and validity—total words and proportion words. There were three trials of sentence sets in the RST. The number of sentences in one set was increasing as the RST proceeded. To be specific, in the first trial the participants read one set of 2 sentences, 3 sentences, 4 sentences and 5 sentences. In the second trial, the participants also read one set of 2 sentences, 3 sentences, 4 sentences and 5 sentences. In the last trial, the participants also read one set of 2 sentences, 3 sentences, 4 sentences and 5 sentences. There were 42 sentences in total. With the first method of total words, participants’ WMC refers to the total words they remembered from all the three trials. For instance, when a participant successfully remembered two out of five words on a trial in a recall-RST or recognized two out of five words on a trial from the provided list in a recognition-RST, his or her WMC was two. 42 sentences meant 42 ending words, and the maximum points for one’s WMC was 42.With the second method of proportion words, the participants’ WMC would be able to reach the average percentage of correctly recalled or recognized words across the number of words in all trials. For example, if the participants remember two out of five words in one trial, the WMC for him or her was .4 (40%). And all the correctness percentage across all the trials was averaged as that participants’ WMC. With this scoring method, failure to recall words at beginning part of a RST, two sentences in one set for example, would result in lower scores than at later level, five sentences in one set for example. With this scoring method, the maximum scores under this proportion method would be 1.00. As Conwey et al. (2005) indicated the two methods of total words and proportion words present a normal distribution, good reliability, and reasonably higher correlation with RC
37
performances. Friedman and Miyake (2005) suggest the total number of correctly recalled (or recognized) words across be the better scoring method because it has better reliability and is able to demonstrate stable correlation with RC.
Afterwards, a Cronbach’s alpha test was employed to ensure the internal consistence of reliability in this psychometric test. In other words, Cronbach’s alpha helped ensure that the result truly represents every individual participant’s WMC. Although every participant had their own particular strategies of conducting the RST developed during the test-taking process. The Cronbach’s alpha suggested the RST used in this study have satisfactory reliability. The Cronbach’s alpha was 0.663 for the RST and 0.781 for the RST in the pilot study, which fell within the acceptable range of scores being higher than 0.5. Furthermore, the interview with the participants also supported the construct validity of RSTs in terms of testing WMC. WMC is composed of three main
components—phonological loop, visuals spatial sketchpad, and central executive interacting with episodic long-term memory (LTM), visual semantics and languages. In order to memorize the ending words during the RST in this study, some participants used the clues of sounds of the ending words to memorize while some linked the meanings of the ending words into a meaningful sentence or a visual picture. Still some used both sound and semantic clues to keep the ending words in mind. The three main techniques used to memorize the ending words is consistent with the WMC
components—phonological loop, visual spatial sketchpad, episodic buffer— and visual semantics. Therefore, the construct validity and reliability of the RST was guarded and applicable.