xml version 1.0 encoding UTF-8
REPORT xmlns http:www.fcla.edudlsmddaitss xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.fcla.edudlsmddaitssdaitssReport.xsd
INGEST IEID E6FDFZFE0_CJPL30 INGEST_TIME 2014-10-03T21:59:36Z PACKAGE UFE0046855_00001
AGREEMENT_INFO ACCOUNT UF PROJECT UFDC
1 EFFECTS OF SYLLABLE LENGTH ON SHORT TERM MEMORY OF NUMBER SEQUENCES: A CROSS LINGUISTIC STUDY OF KOREAN AND JAPANESE BILINGUALS By ALEUNA LEE A THESIS PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL F ULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF ARTS UNIVERSITY OF FLORIDA 2014
2 2014 Aleuna Lee
3 To my husband Han Kyul and my son Hayden
4 ACKNOWLEDGMENTS Above all, I deeply grateful to my advisor, Dr. Wind Cowles for the guidance, resources, and support that she has provided. This project would not have been completed without her continuous help and patience have such a great mentor and her i nspiration and encouragement from the beginning to this day have always been invaluable not only on an academic level but also a personal level. My other committee member Dr. Edith Kaan also has been integral to this entire process. Her suggestions pr ovided important insights throughout the various stages of the project. I appreciate all of the professors in linguistics who helped me to develop my academic knowledge in the field and to finish the thesis work. I also appreciate HaNy Kim, who kindly a llowed me to share her office to write the thesis and George Collins, for proofreading and providing helpful advice. Lastly, I would like to thank my husband Han Kyul who completed my life Mom: thank you always for being my best friend and my dad for the unconditional love and support. I would not be able to finish th is project without them I also want to thank my seven month old son Hayden for being a good baby throughout the process and giving me motivation A special appreciat ion goes to my mo ther in law who came all the way from Korea to support my family and my father in law who encouraged me continuously.
5 TABLE OF CONTENTS ACKNOWLEDGMENTS ........................................... ..................................................4 LIST OF TABLES .......................................................................................................7 LIST OF FIGURES ....................................................................................................8 LIST OF ABBREVIATIONS.........................................................................................9 ABSTRACT................................................................................................................10 CHAPTER 1 INTRODUCTION........................................................... ............... .............. ........ 12 Background.............................................................................. .......................... .12 Working Memory and Phonological L oop M odel... .. ..... ......... .. ..................... ... ...14 Numerical Cognition in B ilinguals .................. ....................... ........................ .......16 Digit words in Korean and Japanese....... ........................ ................. ...................18 Overview of the Current S tudy...... ...... ............................... .................. ...............19 2 MATERIALS AND METHODS ..........................................................................22 Participants.......................................................................... ................. .............. 22 De sign..................................................................................... ............... .......... ...23 Procedure and Materials ........................................ ......... .......... ................. .. .......24 Forward Digit Span T ask ............ ..... ... .............................. ...... .. .............. .... ...24 Backward Digit Span Task................................................................... .........26 Operation T ask....... ................ ...................................... ........................ ......... 26 Operation s timuli ............................................ ... ................... .. ............. 27 Word s timuli ............................................. .... ........ ............ .................. 27 3 RESULTS .............................................. ................................... ............ ..............2 9 Descriptive Statistics...........................................................................................29 An alysis of Group M eans .... ........................................... ........... ...... ................... 31 Correlational A nalysis............................................... ...................... ................. .... 34 Regression Analysis............................................................................................36 4 DISCUSSION ........................................................ .......... .................................. 39 Research Question 1 .... .......................................................................................39 Research Question 2 .............. ...... ................................... ...... ........... ..................42 Limitations and Future Directions....... .... .................... ......................................... 45 Page
6 APPENDIX A FIR ST LANGUAGE DIGIT SPAN STIMULI ............. ............ ................... .. .. ... .. ... 47 B ENGLISH DIGIT SPAN STIMULI .................... ........................ ............................48 C ENGLISH OPERATION TASK STIMULI ... ............... ................................... ........ 49 D FIRST LANGUAGE OPERATION TASK STIMULI ......... ............................ ........ 50 REFERENCE LIST............ .............. .......................................................................... 51 BIOGRAPHICAL SKETCH ................................................................................... ....55
7 LIST OF TABLES Table page 2 1 Means per each group................................................................ .. ................. 22 3 1 Descriptive data for variab les of interest ( mean data) .. ........................ ......... 30 3 2 Analysis of variance in the first language condition ...................................... 32 3 3 Pairwise comparisons...... .............................................................................. 32 3 4 Digit span and WMC measures from bilingual speakers .............................. 33 3 5 Correlations among measures ...................................................................... 35 3 6 Regression results from L1 digit span of bilinguals ....................................... 37 3 7 Regression results from L2 digit span of bilinguals ....................................... 37
8 LIST OF FIGURES Figure page 3 1 L1 digit span performance ....................... .................................................... 38 3 2 L2 digit span scatterplot ................................................................................ 38
9 LIST OF ABB REVIATIONS BD Backward Digit Span FD Forward Digit Span L1 First Language L2 Second Language OP Operation Task RT Reaction Time WMC Working Memory Capacity
10 Abstract of Thesis Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requir ements for the Degree of Master of Arts EFFECTS OF SYLLABLE LENGTH ON SHORT TERM MEMORY OF NUMBER SEQUENCES: A CROSS LINGUISTIC STUDY OF KOREAN AND JAPANESE BILINGUALS By Aleuna Lee May 20 14 Chair: Heidi Wind Cowles Major: Linguistics Despite substantial attention on the effects of word length on short term memory of numbers, it still remains unclear wh at accounts for differences in number memory for example whether it is phonological efficiency of number words, individual difference s numerals is in need of investigation. As such, the present study looked for specific differences in number memory between speakers of Korean, Japanese and English and evaluate based on measures of working memory capacity. The goals of the study were to investigate the effects of syllable length on a digit span task and mental calculation by comparing performance from Korean and Japanese bilinguals, and explore the underlying mechanism of number processing among bilinguals by examining the differences in performance between first language s and second language s Forward and backward digit span tasks were used to measure short term memory of number sequenc es and operation task estimated the working memory capacity of the participants. As Korean counting words have shorter syllable length, it was hypothesized that digit span and mental calculation score s would be larger among Korean participants than Japanes e speakers.
11 The results revealed that the total number of digit sequences correctly recalled were greater among Korean English speakers than Japanese English speakers and English monolinguals when the first language was used During the second language trial s Korean speakers outperform ed Japanese speakers only in backward digit span task The results of digit span between two groups showed that the type of language used in the task was influential during first language processing, while it was the varia tion in working memory capacity that made the difference between groups in the second language task This pattern supports the effects of syllable length and is consistent with the prior observations that working memory should be highly correlated with dig it span task, because it is one of the most interference such as language.
12 CHAPTER 1 INTRODUCTION Background Attempts to investigate the mental representation and proc essing of numerals have been made extensively on both theoretical and empirical grounds; however, the question of how to conceptualize the cognitive architecture of number processing and mental representation is still the source of much debate. Proposed by McCloskey et al. (1992), the Abstract Code Model assumes that the architecture of number processing is interrelated with various components of number comprehension, calculation, and production. According to the model, all numerical inputs such as non verb al digits or verbal number words are converted into the abstract representation. Then, it is the abstract code that is involved in numerical tasks (e.g., calculation process, numerical comparison, naming numbers). The hypothesis of abstract format represen tation leads to claim an independency of surface formats of inputs (e.g., Arabic numerals, written or verbal number words) and comprehension and calculation process. For example, in case of presentation of a multiplication problem, there is no effect of in put format in mental computation Complex Model by Campbell and Clark (1988), which reje cts a central abstract representation and supports an interactive network of various representations such as phonological, articulatory, visual, and graphemic codes, Dehaene (1992) outlined the Triple code Model. This model assumes that numeric mental proc esses are based on three representational codes such as auditory verbal word frame (e.g., /six/), visual Arabic number form, and an analog magnitude representation. The auditory verbal word codes are assumed to support written or spoken number words,
13 and m ay contribute to addition and multiplication processes. The speculation mediated primarily by language The controversy has derived partly from notation specific representations such as non verbal numerals (e.g., Arabic number forms: 1,2,3) and verbal number forms produce spoken numerals and written numerals ( Campbell, 1994; Koechlin et al., 1999 ) and yet are themselves different representations. Then, a question arises as to how Arabic numerals and counting words are mentally represented during the processing of numbers and whether there is any variation in representation between different language speakers. Assuming cross linguistic universality of Arabic numerals, investigation of the relationship between linguistic features of number words and numerical cognition might provide meaningful insights to under stand encoding specificity in number processing. Thus, several studies have sought to characterize a role of language in numerical cognition such as short term memory of numbers or mental arithmetic. mory capacity is cited and influential works in memory and cognition, whether memory capability is actually uniform in holding numbers across languages is unclear. In partic ular, many researchers have demonstrated how linguistic properties of digit names, such as word length or phonological similarity, limit short term memory of number sequences. This effect of articulatory duration and phonological similarity is explained wi thin the discussed in the following section.
14 The current study explores how and to what extent specific language features such as syllable length contribute to processing of sh ort term memory of number sequences and mental calculations. These features vary from language to language, and in particular between English, Korean and Japanese. In addition, while Arabic numerals are used by speakers of all three languages, Korean and Japanese have distinctive digit names with different syllable lengths to compare. Despite substantial attention on the effects of word length on short term memory of numbers, which accounts for differences in number memory whether it is phonological effici ency of exposure to numerals is still ambiguous. The effects of language alone may not be able to explain the performance in numerical cognition, in particular variation in worki ng memory capacity and/ or better mathematics performance of Asian language speakers were assumed to be potential contributors in memory for number sequences that should be controlled for. As such, this study will look for specific differences in number me mory between speakers of Korean, Japanese, and English and evaluate based on measures of working memory capacity. Working Memory and Phonological Loop Model Working memory is a limited capacity ability to temporarily maintain and process the information involved in an ongoing task (Otten & Van Berkum, 2009). The concept of working memory was first introduced by Miller, Galanter, and Pribram in 1960 and was referred to the memory used to plan and execute behaviors (Cowan, 2008). Expanded from Miller et al ., Baddeley and Hitch (1974) proposed an influential framework of the working memory system known as the three component model of working memory; the phonological loop, the visuospatial sketchpad and the central executives, emphasizing the contribution of interdependent multiple modules
15 in the storage system. A fourth component was added later by Baddeley (2000) ; the episodic buffer which is responsible for integrating information from different sources and long term memory into an episodic representation. Based on the sounds or phonological information of language, the phonological loop compromises two subsidiary systems; a phonological store that holds temporary memory traces for a few seconds, and an articulatory rehearsal process that allows phonological codes to be revived. That is, visual information will be transformed into phonological codes through vocalizing them covertly or overtly, and hence phonological storage becomes crucial in retention of the visual information. This also implies that if the number of items to be retrieved increases, items that are subvocalized first are decayed before they rehearsed. The phonological loop model is supported by a wide range of empirical evidence (e.g ., Baddeley et al., 1975; Schweickert & Boruff, 1986; Longoni et al., 1993; Baddeley, 1966a, 1968; Conrad & Hull, 1964; Schweickert et al., 1990). The first important factor that influences the immediate memory trace is the effect of phonological similarity; lists of phonologically similar words such as pea, bee, ke y, fee are remembered less well than words that are pronounced dissimilar such as man, sky, old, fix (e.g., Conrad, 1964; Baddel e y, 2003). This fact supports the primary construct of the phonological loop model that visual information is represented and encoded phonologically rather than visually or semantically in the working memory storage system. The same study also examined the significance of semantic similarity in the immediate recall; however, similarity of meaning was found less important than pho nological similarity in short term memory (Baddel e y, 1966a). The effect of semantic similarity becomes more evident in the long term memory system in which the influence of phonological similarity is no longer valid (Baddel e y, 1966b).
16 The word length effe ct also explains the articulatory rehearsal process that immediate serial recall is better when articulatory duration is shorter. That is, limitation of rehearsing time of longer words results in more decay of items than shorter words. Baddeley et al. (197 5) investigated the effect of word length and found a strong correlation between performance of cognitive task and word length across a wide range of materials. Baddeley et al. focused on the memory span task as a measure of cognitive processing and the re sults showed that as the duration to pronounce the item increased, serial recall performance decreased. Many studies also confirmed that articulatory duration of words determined memory span across languages. For example, Neveh Benjamin and Ayres (1986) st udied the relationship between articulatory rate and immediate memory span among four language groups: English, Spanish, Hebrew, and Arabic. The results found a larger memory span for speakers of languages where reading rate was faster. A comparison of Chi nese and English (Hoosain, 1982; Stigler et al., 1986) also supported the prediction that differences in digit span memory between speakers of two languages are attributed to differences in time to pronounce number words. Further, Chincotta and Underwood ( 1997) tested native speakers of six languages Chinese, English, Finnish, Greek, Spanish, and Swedish. The digit span was recorded and the results confirmed superior digit span in Chinese as a function of shorter pronunciation duration of Chinese. Numerica l Cognition in Bilinguals So far, we have discussed cross linguistic variations in memory of numbers depending on the time to articulate items for monolinguals. Studies in bilinguals also support a causal relationship between speech rate and digit span: d igit span was larger when participants were instructed to answer in the language in which
17 articulation duration for digit names were shorter (Chincotta & Hoosain, 1995; Chincotta & Underwood, 1996; Ellis & Hennelly, 1980). Although the word length effects are reported in several bilingual studies, da Costa Pinto (1991) observed the superiority of first language in digit span even in cases of faster reading rates in second languages. These findings suggested that massive practice in the native language to a llow reducing speech duration and more likely to remember larger digits. The results assumed a role of proficiency, level of familiarity, and degree of dominancy of first or second language in performance of digit span. Moreover, Chincotta, et al. (1997) d emonstrated that not only shorter digit names but also larger storage capacity in the first language and competence level of the language also contribute to performance in digit span. Studies in mental arithmetic also revealed effects of word length. For example several researchers have reported a strong first language preference when bilinguals are exposed to two languages simultaneously in performing mental calculation (Kolers, 1968; Vaid & Menon, 2000). However, other studies have claimed that language preference is observed when digit names of one language are shorter than the other language (Ellis, 1992; Ellis & Hennelly, 1980). For example, Ellis (1992) examined balanced bilinguals of Welsh and English and found a lower error rate and shorter reactio n time in mental calculation when tested in the language with shorter number names. Although the influence of word length in memory capacity has been extensively demonstrated in studies with both monolinguals and bilinguals, most studies have used the te rms syllable length, word length, reading rate, speech rate, and articulatory duration interchangeably in the same context. Further, measures of articulatory duration still leave the possibility of individual variation and inadequate
18 estimates of speech ra tes between two languages. For instance, in studies by Hoosain (1982) and Ellis & Hennelly (1980) which measured speech rates of English counting words, the numbers did not match between the two studies (375 vs. 321 sec per monosyllabic words). Also, Neveh Benjamin & Ayres (1986) as well as Hoosain (1982) overlooked the significance of quantitative information of word length and did not provide them in the study. Despite the meaningful conclusions provided by previous studies, these cross linguistic analys es have compared languages that were rather culturally, pedagogically, and typologically distant. In addition, previous experimental work has claimed mathematical competence of Asian students and investigated what causes the phenomenon. Some researchers di scussed that cultural expectations and emphases (Hess et al., 1986; Stevenson et al., 1986) are known to be related to differences in mathematical ability. Despite the advantages of Asian students, previous research focused on performance of numerical task s between Asians and groups of participants who are non Asians. In order to address the potential limitations from previous studies, the present study deliberately chose Japanese and Korean speakers, which share marginal cultural and pedagogical difference s in mathematics (Clarke, 2003), while maintaining distinctive phonological features to compare. Digit Words in Korean and Japanese There are two reading systems in Korean numerals : Native Korean and Sino Korean. Based on Chinese characters ( , ) t he Sino Korean reading system is preferred when communicating sequences of numbers and calculations. The Sino Korean numerals consist of monosyllabic words, such as il (1), i (2), sam (3), sa (4), o (5), yuk (6), chil (7), pal (8), gu (9).
19 The Japanese nu meral system also has two readings; Sino Japanese derived from Chinese numerals and Native Japanese. Similar to the Korean system, verbal number reading in Japanese is based on Sino Japanese numerals. Although number reading in both languages is based on t he same Chinese characters, the readings sound relatively different. Unlike Korean digit names, Japanese number words include several disyllabic words such as ichi (1), roku (6), nana (7), hachi (8). Since the native systems in both languages are generally used for ordinal numbers (i.e., first, second, third, etc.), The Sino reading system that is favored in reading Arabic numerals is the one that we will explore in the study. Overview of the Current S tudy As noted previously, studies comparing two languag e groups focused on languages that are culturally and pedagogically different, and many researchers inadequately measured word length. Thus, in the present study, comparisons were strictly made within the boundary of syllable length between Japanese and Ko rean, maintaining consistency throughout the analysis. Further, little effort has been made to consider individual variation in working memory capacity. Therefore, the current study implemented an operation task as a measure of working memory capacity and examined whether numerical cognition performance is attributed to differences in working memory capacity, rather than the articulatory efficiency of digit names. The current study had several guiding research questions: Research Question 1: How do the sur face features of counting words facilitate or inhibit memory and retention of number sequences? Controlling for cultural and pedagogical differences in numbers between two language groups (e.g., Korean and Japanese), does the comparison show differences i n performance where vocal articulation
20 of counting words is involved, such as on a digit span task or mental arithmetic? More specifically, as Korean digit words take less time to articulate than those of Japanese, will Korean participants outperform Japa nese participants in numerical cognition tasks? Assuming the significant role of syllable length, are there any other factors that may affect the performance? Research Question 2: What role do counting words play in access to numerical information process ing among bilinguals? Do results from the second language condition for bilinguals also support effects of syllable length? Since the same English stimuli are tested in Korean and Japanese speakers, does this comparison show any significant differences? I s it the first language advantage or efficiency of word length that affects memory of numbers and mental calculation? Does overall performance of bilinguals surpass the results of monolinguals due to superior executive functioning of bilinguals? Or, are t here any first language disadvantages because of possible interference of the first language? Does degree of proficiency in the second language play a role in numerical cognition tasks? To summarize, the goals of the current study were to 1) investigate t he effects of syllable length on a digit span task and mental calculation by comparing performance from Korean and Japanese speakers, and 2) explore the underlying
21 mechanism of number processing among bilinguals by examining the differences in performance between the first language condition and the second language condition. Based on the literature review, we hypothesized that 1) the number of digits recalled accurately would be larger among Korean participants than Japanese speakers, 2) performance in me ntal calculation between two groups also would depend on time to articulate the digit names, 3) results in the second language condition would support the effects of syllable length as there would be no difference in digit span and mental calculation durin g the second language condition between Korean and Japanese participants.
22 CHAPTER 2 MATERIALS AND METHOD S Participants A total of 60 students at the University of Florida (25 English monolinguals, 20 Korean English bilinguals, and 15 Japanese Engli sh bilinguals) participated in the study. Undergraduate English monolingual students participated in the experiment for class credit and bilingual students participated voluntarily Bilingual participants uch as cultural groups for Japanese and Korean students, the Department of Language Literatures and Cultures and word of mouth. For monolingual participants, the LIN SLHS participant pool was used, along with word of mouth. The average age of participants was 23.47 years; there was, however, some variability in age among the three groups of participants (SD=4.28, range=12). Table 2 1 summarizes the demographic information for each group of participants. Table 2 1. Means per each group English Monoling ual Korean Bilingual Japanese Bilingual Age 20.72 (1.99) 27.65 (3.88) 22.47 (3.23) Age of L2 acquisition 10.55 (3.46) 11.47 (1.92) Years in US 3.10 (3.08) 1.83 (2.16) Self rated L2 proficiency 5.70 (1.69) 6.13 (1.30) Values inside parentheses ar e standard deviations.
23 For English monolingual speakers, only those who spoke English natively and use English in their daily life participated in the study. Bilingual speakers spoke Korean or Japanese as their first language (L1), and were first expose d to English as their second language (L2) no earlier than 7 year of age and completed at least middle school in their home countries. Among the bilingual participants, 5 were currently enrolled in the English Language Institute at the University of Florid a and 9 were exchange students from Kansai Gaidai University in Japan. Both English native speakers and bilinguals speakers filled out a questionnaire, which collected demographic information such as age, gender, and race. Bilingual speakers were provided with more questions in which they identified their language background information, such as age of first exposure to English, number of years staying in the English speaking country, official English test scores such as TOEFL, GRE, or SPEAK (if they had an y), and self rated proficiency of English. Self rated L2 proficiency data including reading, writing, speaking, and listening with a scale from 0 for no knowledge to 10 for native like fluency were collected. None of the participants reported any abnormali ties on memory examinations. The language background data showed that there was no significant difference between Japanese and Korean bilinguals in those four dimensions determining their overall proficiency and familiarity in their L2. Design Instructio ns were presented in English for both monolingual and bilingual participants. Monolingual speakers were tested in English only and bilingual participants were given the tasks in both English and their L1. That is, after completing the tasks in English, bil ingual participants were asked to perform the
24 same tasks in their L1 additionally with the different sets of stimuli. All stimuli were presented in black writing on a white background. Procedure and Materials All the experiments were individually conducte d in a quiet testing room and administered by the researcher. response on a worksheet. When the participant arrived to the site of the experiment, the researcher verbally explained the purpose of the study as well as what the participant was expected to be doing during the course of the experiment. The participants were then given the consent form and asked to read it carefully. Once any questions were answered, the participants were asked to sign the form, and the experimenter counter signed. The experiment started with instructions displayed on a laptop computer running PsyScope X, then followed by a forward digit span task, a backward digit span task, and an operation task. For the bilingual participants, the same three tasks were presented in their L1 in addition to the tasks in the L2. Forward Digit Span Task In the forward digit span task, participants were first presented with the number sequences visually while being asked to read the numbers aloud in th e order presented. Number sequences from three to ten digits were used in digit forward trials with two trials per each sequence length and presentation of number sequences on a screen was automatically manipulated at a rate of one digit per second. Numbe rs were selected randomly with several constraints in constructing the lists: no same digit was repeated successively (e.g., 8 8) and no meaningful
25 numbers were used (e.g., 1 9 8 4). One point was awarded for each list recalled in the correct position. In addition, digit sequences were manipulated for five, six and seven digits. Numbers were selected randomly but based on the number of syllables in Japanese counting words. Due to the aforementioned fact that all Korean number words are monosyllabic, the sy llable structure of Korean number words was not a part of consideration in this manipulation. One set of each length mostly consisted of numbers with monosyllabic words in Japanese, constructing numbers with as few syllable as possible. On the other sets, numbers consisted of disyllabic words and few monosyllabic words, designing the numbers to contain as many syllable as possible. For instance, on five digit sets, a first list would be 3( san ) 9( kyu ) 2( ni ) 5( ko ) 4( yon ) and second list would be 6( roku ) 1( ich i ) 7( nana ) 2( ni ) 8( hachi ). Although both number strings contained the same number of digits, the number of syllables differed from 5 (first list) to 9 (second list). In this way, an increase of digits does not necessarily mean increase of syllable length. In other words, the number of syllables for the second list of five digit was 9, but the number of syllables for the first list of six digit can be 7. The construction became more evident to estimate quantitative aspects of variables by maximizing effects of syllable length; however, longer strings such as eight nine and ten digits were not manipulated, because it cannot be done without using the same numbers repeatedly in the sequence. To minimize developing strategies for the task by participants, pho tographs were inserted between each list of numbers and displayed for ten seconds. Photographs were downloaded from the National Geographic website and portrayed landscape or wildlife. Picture quality of the photographs used in this study was 1280 x 1024 p ixels.
26 After presentation of the pictures, the participant was instructed to recall the numbers orally in the order presented. There was no time limit required during the recall. If participants failed to recall the digits in the correct order, they proce eded to the next digit sequence. Backward Digit Span Task Once participants had finished recalling 16 sequences of numbers in the order, they proceeded to the backward digit span task in which participants were asked to read aloud the numbers in the order presented, and then recall them in the reverse order. Number sequences from three to nine digits for digit backward with two trials per each sequence length. See Appendix A for complete list of digit span stimuli in the first language trial and Appendix B for complete list of the English trial Operation Task Next, the operation span task designed by Turner and Engle (1989) was used required to solve a series of mathematic al operations while trying to recall words at the same time. Participants were first asked to read aloud the math problems and decide whether the answer was true or false by pressing the designated keys in the keyboard on a computer. True or false answers and response times were collected. On the next screen, a word appeared for three seconds and participants were asked to read the word aloud. The next math problem and word pair was presented immediately. Given the arrangement of 12 trials with set sizes r anging from 2 to 5, participants were asked to recall the words only in the same order presented after each trial. There was no time limit during the recall. The entire testing session for
27 English monolingual speakers took approximately 20 25 minutes, whil e bilingual participants spent 40 50 minutes on average to complete the experiment. Operation s timuli A total of 42 math operations consisting of two arithmetic operations such as [(4x2) 1]= 7 or [(9/3)+2]= 6. Bilingual trials consisted of an additional 4 2 math operations that were created in the same degree of difficulty. Word s timuli A total of 42 English words used for monolingual trials and L2 trials for bilingual speakers. An additional 42 Japanese words were used for Japanese speakers and the same n umber of Korean words used for Korean speakers in their L1 trial. L1 word stimuli were translate equivalent in both languages Although the present study investigated the effect of syllable length, length of word stimuli by each language was not considered in designing the stimuli, but semantic consistency between languages was established. Since previous stud ies such as Baddeley ( 1966a ) suggested semantic similarity affects the long term memory and operation tasks involve relatively longer memory trace s s emantic cues in word span should be consistent between Korean and Japanese trials. H owever, it is also important to note that there are three types of orthography in Japanese; Kanji, Hiragana, and Katakana. Thus, statistical differences in recalling words depending on different types of characters used were examined during the pilot study before the experiment; however, the results showed that there was no effect of script type in the recall of words. Thus, the word lists were created with mixed use of scri pts for Japanese, based on frequency and acceptability judgment by native speakers of Japanese. The entire list of words to be recalled was selected from a Familiarity Rating on MRC Psycholinguistics Database. The Familiarity Rating values in this databas e
28 lie in the range of 100 to 700, and the word stimuli used in this study were in the range of 600 to 700. A lexical category of all three language sets of words was concrete nouns. One point was awarded for each word recalled correctly regardless of the p osition within the list. See Appendix C for complete list of the English stimuli and Appendix D for the first language stimuli.
29 CHAPTER 3 RESULTS First, the descriptive statistics of mean number of digits recalled in the correct order, mean percent correct in word span and mental calculation of operation task, and mean reaction time in mental calculation for each group in the first language and the second language conditions are presented. Next, a one way ANOVA evaluates in the first language condit ion whether numerical cognition performance between three language groups is significantly different. Then, a two tailed t test is carried out to examine significant difference in performance between Japanese English bilinguals and Korean English bilingual s in the second language condition. Finally, a correlation analysis is conducted to measure the strength of linear association between variables such as types of tasks, working memory capacity, L2 proficiency, or demographic information. Descriptive Stati stics The total number of correct recall in digit span, a percentage of accurate answers from the operation task, and reaction times of mental calculations are given in Table 3 1. In general, Korean English bilingual outperformed the Japanese speakers on t he forward digit span, backward digit span, and word recall tasks when the first language was used. Korean English bilingual also showed better performance than English monolingual on backward digit span when they were tested in English. Although English i s not their native language, digit memory in the reverse order can be superior in the monolingual group whose first language is English. Moreover, the mean percent correct on word recall on the operation task also demonstrated memory superiority of Korean English bilinguals. This might suggest that the variation source of digit span is not attributed to the syllable
30 efficiency of Korean digit names; rather the advanced capacity of their working memory plays a role in memory of numbers. Thus, the question of whether it is an advantage in working memory capacity or the effect of syllable length is examined through a correlation analysis in the next section. Table 3 1. Descriptive data for variables of interest (mean data) Values inside parentheses are standard deviations. ** Since tasks were conducted in multiple languages, total means of the first language condition ( where Korean and Japanese were used for bilinguals and English for monoli nguals ) as well as the second language condition in which English was used for two bilingual groups is reported. Korean English Bilingual (n=20) Ja panese English Bilingual (n=15) English Monolingual (n=25) Total Means (n=60) English Korean English Japan ese English L1 L2 Forward Digit Span 9.25 (2.59) 14.80 (1.40) 7.73 (2.19) 12.00 (1.89) 11.12 (1.81) 12.57 (2.34) 8.60 (2.51) Backward Digit Spa n 8.25 (2.63) 10.75 (2.31) 6.20 (2.27) 8.60 (2.69) 8.12 (1.72) 9.12 (2.46) 7.37 (2.66) Operation Task Mean Percent Correct (%) Mental Calculation 93.69 (4.25) 91.55 (4.40) 95.40 (3.18) 91.75 (6.03) 91.24 (7.80) 91.47 (6.30) 94.42 (3.87) Word Recall 69.52 (9.70) 83.45 (9.76) 63.02 (7.68) 76.51 (8.48) 78.19 (9.37) 79.52 (9.58) 66.73 (9.36) Mean RT (ms) Mental Calculation 7504.73 5 (1419.380) 6317.031 (1781.715) 7836.294 (1409.108) 6106.044 (1155.117) 6924.327 (2560.602) 6517.324 (2032.937) 7646.831 (1403.971)
31 Analysis of group means The performance on digit span and operation tasks was computed with an analysis of variance (ANOVA). A one way ANO VA was conducted to compare the effect of language on numerical cognition tasks in three language groups: Japanese English bilinguals, Korean English bilinguals, and English monolinguals. A one way ANOVA between subjects on the forward digit span, backwar d digit span, word recall, mental calculation, and reaction times in answering mental calculations in the first language condition is presented. In this analysis, Japanese and Korean data from bilinguals and English data from monolingual speakers were cons idered. First, as anticipated, the groups differed in forward digit span [F(2,57)=26.96, p<.0001]. Results from the backward digit span in the first language condition also showed a significant difference between the three groups [F(2,57)=8.55, p<.001]; ho wever, groups did not differ significantly in word recall, mental calculation, or reaction times. Table 3 2 shows the analysis of variance for mean number of correct digit, mean hoc tests indicated pairwise comparisons and revealed which group contributed to the difference. In the forward digit span task, the significant differences were found between Japanese English and Korean English bilinguals (p<.0001), as well as English monolinguals and Korean English bilinguals (p<.0001). Data from the backward digit span also revealed differences between Japanese English bilinguals and Korea English bilinguals (p<.05) and English monolingual and Korean English bilinguals (p<.01). The Japanese English bilingua ls and English monolinguals did not differ significantly from each other in both forward digit span and backward digit span. These results suggested that Korean English bilinguals might be the important group in
32 comparisons of numerical cognition between t hree language groups. The pairwise comparison results are shown in Table 3 3. Table 3 2. Analysis of variance in the first language condition Effect Mean Square F p Forward Digit Span 78.447 26.962 <.001 Backward Digit Span 41.097 8.550 <.001 OP_Word Recall .024 2.835 .067 OP_Mental Calculation .000 .032 .969 OP_RT 3740439.954 .902 .411 OP=Operation Task RT= Reaction Time Table 3 3. Pairwise comparisons (I) Language group (J) Language group Mean Difference (I J) Std. Error p 95% Confidence Interval Lower Bound Upper Bound Forward Digit Span Japanese Korean 2.800 .583 .000 4.20 1.40 Korean English 3.680 .512 .000 2.45 4.91 English Japanese .880 .557 .263 2.22 .46 Backward Digit Span Japanese Korean 2.150* .749 .016 3. 95 .35 Korean English 2.630* .658 .001 1.05 4.21 English Japanese .480 .716 .782 2.20 1.24 The mean difference is significant at the 0.05 level The data from the second language condition were submitted to a two tailed t test, examining whethe r there is a significant difference in numerical cognition performance between Korean English bilingual and Japanese English bilingual. Since English monolingual participants were only tested in their first language, second language results from bilingual speakers were only subjected to analysis. As
33 shown in Table 3 4, significant differences among the language groups were found on backward digit span, t(33)= 2.41, p<.05, and on word recall, t(33)=2.14, p<.05. There was no difference between bilingual langu age groups on forward digit span, mental calculation, or reaction time. Table 3 4. Digit span and WMC measures from bilingual speakers t df Sig. (2 tailed) Mean Difference Std. Error Difference 95% Confidence Interval of the Difference Lower U pper Forward Digit Span 1.828 33 .077 1.517 .830 3.205 .171 Backward Digit Span 2.413 33 .022 2.050 .850 3.778 .322 OP_Word Recall 2.141 33 .040 .065 .030 .127 .003 OP_Mental Calculation 1.304 33 .201 .017 .0134 .010 .044 OP_RT .686 33 .49 8 331.558 483.326 651.775 1314.892 OP=Operation Task, RT=Reaction Time In addition, overall results in comparisons of monolinguals and bilinguals were examined. Two tailed t tests were computed between monolingual and bilingual group which combined d ata from Korean and Japanese speakers. In both forward backward digit span tasks, the the monolingual group, t(33)= 5.07, p<.001 and t(33)= 2.89, p<.01, respectively. Despite the efforts to more thoroug hly inspect the effects of syllable length by implementing trials comparing Japanese long and short sets of stimuli, the results indicated no robust evidence of the effects that the present study was interested in.
34 Correlational Analysis The correlation al analysis was conducted to understand the relationship between multiple variables that potentially influenced the results. The correlations among categorical variables: types of language (Japanese and Korean), sex (male and female), and continuous variab les: age, self rated L2 proficiency, working memory capacity, and measures of numerical cognition are presented in Table 3 5. The same task in the two language conditions are correlated as shown in forward digit span in L1 and L2 (r=.50, p<.01), backward d igit span in L1 and L2 (r=.72, p<.01), and word recall of operation task in L1 and L2 (r=.58, p<.01). As predicted from the previous analysis of group means, positive correlations between types of language and multiple numerical cognition tasks in both fi rst language and second language trials were found. Correlations between the forward digit span task in Japanese and Korean were significant, r(35) = .66, p<.01 as well as backward digit span task between two languages (r=.40, r=.39, p<.05). Working memory measures (word recall) in the first language correlated with forward digit span in the first language (r=.38, p<.05) and backward digit span in the second language (r=.46, p<.01). Working memory measures in the second language also correlated with both fo rward digit span (r=48, p<.01) and backward digit span (r=45 p<.01) in the second language. Overall, these results indicated a significant relationship between language and memory of number sequences, but also suggested there is a firm correlation betwee n working memory capacity and memory of numbers. This finding is not surprising since digit span tasks are one of the most commonly and widely used measures to quantify working memory capacity.
35 Table 3 5. Correlations among measures Significant values are marked in boldface. FD= Forward Digit Span, BD=Backward Digit Span, WORD= Word Recall, MATH=Mental Calculation, RT=Reaction Time, L1= First Language, L2=Second Language *. Correlatio n is significant at the 0.05 level. **. Correlation is significant at the 0.01 level. 1 2 3 4 5 6 7 8 9 1 0 11 12 13 14 1 Language 2 Sex .264 3 Age .590 ** .259 Results in L2 4 FD .303 .296 .231 5 BD .387 .190 .257 618 ** 6 WORD .349 .273 .142 .475 ** .447 ** 7 MATH .221 .042 .057 .045 .323 .241 8 RT .119 .257 .262 .533 ** .354 .064 .215 Results in L1 9 FD .660 ** .376 .491 ** .508 ** .619 ** .332 .049 .278 10 BD .404 .193 .286 .454 ** .725 ** .272 .128 .346 .752 ** 11 WORD .358 .487 ** .203 .321 .459 ** .584 ** .184 .022 .382 .276 12 MATH .020 .129 .277 .251 .201 .223 .126 .313 .212 .190 .170 13 RT .069 .094 .099 .285 .346 .161 .116 .712 ** .287 .343 .017 .212 14 Proficiency .009 .132 .125 .193 135 .120 .093 .157 .061 .071 .232 .233 .025
36 Regression Analysis Although the present study revealed a statistically significant difference between Korean speakers and Japanese Speakers in L1 digit span, whether pe rformance in digit span is entirely due to the efficiency of syllable length in Korean is still unclear. Therefore, a regression analysis was conducted in order to see how much of the variance could be accounted for by language type and WMC. In this analys is three separate analyses were carried out in both first and second language conditions to show digit span results as a function of language, WMC and combination of these two variables. Digit span performance 1 in L1 was reported as a function of languag e and WMC separately. In both analyses, results revealed contributions of both language (p. <.001 ) and WMC (p < .05). H owever, when the two variables were added together, only language was the significant factor (p<.01) that made difference between groups. Interestingly, the analysis in L2 showed the opposite pattern of results : while both variables again independently appeared to contribute to differences in digit span performance, when the two factors are considered together, only WMC showed a significant impact ( p<.01). These results can be seen in Table s 3 6 and 3 7. 1 Analyses on forward digit span and backward digit span were first conducted separately. Both trials showed the same results, hence, we calculated means of forward and backward digi t span and used it as integrated data for more simplified analysis.
37 Table 3 6. Regression results from L1 digit span of bilinguals Predictors Std. Error t value Adjusted R 2 Pr(>|t|) Step 1 Language 0.6504 3.806 0.284 0.000582 *** Step 2 W MC 0.03773 2.110 0.09222 0.0425 Step 3 Language +WMC 0.288 WMC 0.03578 1.090 0.28394 Language 0.69448 3.174 0.00331 ** *. Correlation is significant at the 0.05 level. **. Correlation is significant at the 0.01 level. ***. Correlation is significant at the 0.001 level. Table 3 7. Regression results from L2 digit span of bilinguals Predictors Std. Error t value Adjusted R 2 Pr(>|t|) Step 1 Language 0.7438 2.398 0.1225 0.0223 Step 2 WMC 0.03715 3.424 0.2398 0.00 167 ** Step 3 Language +WMC 0.2676 WMC 0.03892 2.745 0.00984 ** Language 0.72522 1.500 0.14334 *. Correlation is significant at the 0.05 level. **. Correlation is significant at the 0.01 level. ***. Correlation is significant a t the 0.001 level
38 Figure 3 1. L1 digit span performance Figure 2. L2 Digit Span Scatterplot
39 CHAPTER 4 DISCUSSION The goal of the study was to evaluate the effect of syllable length in short term memory of number sequences and mental arithmetic betw een Korean English, Japanese English bilinguals and English monolinguals. Participants from three different language backgrounds completed digit span tasks and an operation task. Bilingual speakers were first tested in the second language (English) and the n the same tasks were repeated in the first language (Korean or Japanese). English monolingual participants were only tested in the first language (English). Measures of digit span and operation task were used as the dependent variables. Research Questio n 1 Research Question 1 asked whether Korean participants outperform Japanese speakers in digit span and mental arithmetic based on the observation that Korean digit words have a shorter syllable length. As predicted, a total number of digit sequences rec alled correctly were greater among Korean English speakers than Japanese English speakers, when the first language was used. The findings are supported by previous studies in which larger digit span was observed as a function of shorter syllable length of digit words (Ellis & Hennelly, 1980; Hoosain & Salili, 1987; Stigler et al, 1986). The examination on whether syllable length drives differences in mental arithmetic revealed that there is no robust interaction between the two variables. Two measures in m ental calculation such as percent correct on math problems and reaction time did not show significant difference with any other variables. It is possible that math problem in the operation task was overly simple for the participants that it was unable to p recisely measure arithmetic ability of the participants. Mean data from the mental calculation may support the assumption
40 (M=91.47%, SD=6.30% in L1 and M=94.42%, SD=3.87% in L2). Due to these relative high scores in mental calculation, effects of language might not be shown in the analysis; however, a two tailed t test revealed that mental calculation scores in the second language condition were significantly higher than those of the first language t(33)=2.66, p<.05. This result might be an indication of la nguage preference due to the effect of syllable length. English digit words are shorter than Japanese; however, the number of syllables in English words are nearly equivalent to Korean readings. Likewise, it is also possible that difficulty of mental calcu lation problems caused the difference. Thus, we should not hastily conclude that the difference would be interpreted as language preference in mental arithmetic among bilinguals. Research Question 1 also asked whether there are any other factors that may affect the performance. A possible source of variation in working memory task can be derived from the distinctive orthography of the languages. During the operation task, participants were required to read the word stimuli aloud while they were displayed on the screen visually. Whereas Korean orthography (Hangul) is based on the spatial placement and stacking of letters consisting of initial (syllable onset), medial (syllable nucleus), and final (syllable coda) phonemes, the Japanese Kana writing systems f ollow a linear arrangement of syllables, writing each character in a row. The processing of visual information in working memory can be interpreted by Baddel e memory (Baddel e y, 1986), unique syllable block organization of letters in Korean may facilitate Korean speakers processing word recall. Similar to Korean, Chinese orthography is also based on spatial construction of elements representing a morpheme. Many researchers have shown th at Chinese speakers rely more heavily
41 on visual information than phonological codes during reading Chinese characters. Hoosain (1991), for example, argued that Chinese speakers might develop advanced visuospatial sketchpad abilities through identifying Chi nese logograph. Assuming this enhanced ability, Chan and Elliot (2011) investigated the visuospatial ability of Chinese speakers, whether superior digit span was due to shorter articulatory duration of Chinese number words or advanced visuospatial processi ng skills. Unfortunately, the results did not support the hypothesis that the Chinese advantage in visuospatial processing did not contribute to differences in digit span between Chinese and Malay. Although the Korean writing system shares some similarity with Chinese in terms of its spatial composition of syllables, Korean has more advantages in visual processing due to the characteristics of phonography, which specifies phonological form, while the Chinese system is based on logographic writing, which doe s not specify a phonological form of each morpheme. Thus, whether Korean speakers may develop enhanced visuospatial skills than Japanese speakers is worth for empirical scrutiny. Although both syllable length and variations in WMC were found to determine differences in digit span across Japanese and Korean speakers, there could be other linguistic factors accounting for the differences. It is tempting to speculate that syllable might not be a meaningful unit for dividing words among Japanese speakers as Ja panese is known to assemble phonology by moraic units, not by syllabic units (Cutler & Otake, 1994; Kubozono, 1989). For example, while speakers of Korean (and many other languages) divide the word yon (4) into one syllable, Japanese speakers might divide into two morae, /yo.n/. Thus, it is possible to assume this different system is related to the verbal memory among Japanese speakers and tasks in their first and second languages. Different psychological representations in
42 phonological processing could aff ect phonological short term memory; however, little effort has been made to investigate how the system of dividing words into different units could influence the memory of words across languages. Research Question 2 Research Question 2 asked a bout the ro le of counting words with respect to numerical information processing among bilinguals and took on three main parts: Do results from the second language condition of bilinguals also support the effects of syllable length? Since the same English stimuli ar e tested in Korean and Japanese speakers, does the comparison show any significant differences? With respect to this question, t here was in fact a significant difference between Korean and Japanese speakers in backward digit span for the second language co ndition, but not in forward digit span. However, considering the regression analysis, it appears that it was the variation in WMC that made difference between groups digit span was not influenced significantly by the encoding of numbers in the L2 (Englis h) Thus, the results confirmed the word length hypothesis : E ffects of syllable length were found only in the first language trials and effects of WMC were observed in the second language condition. T his pattern makes sense and is consistent with the prio r observation s that WMC should be highly correlated with digit span task because it is working memory capability when there is no interference such as language in this case.
43 The second part of this q uestion was as follows: Is it the first language advantage or efficiency of word length that affects memory of numbers and mental calculation? Does degree of proficiency in the second language play a role in numerical cognition tasks? This question can be answered by evaluating the Japanese data because the number of syllables between Korean and English were nearly equivalent and both shorter than Japanese Thus, a s English digit words take less time to articulate than Japanese, we could assume the efficie ncy of word length if they scored better at English trials. On the other hand, if their performance were higher in Japanese trials despite the longer syllable length, this would be consistent with a first language advantage. The results indicated significa ntly better performance during the first language trials in both forward and backward conditions for Japanese speakers supporting the idea that there is a first language advantage Although the present study did not show any interaction between self rate d proficiency of English and working memory measures, the first language advantage might be evidence of effects of proficiency B ecause Japanese participants are late bilinguals and natively more proficient in their first language than the second language, the first language advantage seen here could be interpreted as an effect of proficiency between first and second language s Recently, a number of studies compared performance of working memory tasks in L1 and L2. Chincotta et al. (1997) tested Finnish and Swedish bilinguals whose native languages were different ( one group had Finnish as the first language and Swedish as the second language, the other group vi c e versa ) The study found larger digit span s for speakers in their native language compared to the ir second languages. In addition, van den Noort et
44 al. (2006) compared performance in working memory tasks among multilinguals and found working memory interacted with proficiency of foreign languages The final part of Research question 2 was: Does overal l performance of bilinguals surpass the results of monolinguals due to superior executive functioning of bilinguals? Or, are there any first language disadvantages because of possible interference of the first language? With respect to this question, c omp aring digit span results between monolinguals and bilinguals revealed that digit memory capacity between the two groups was statistically different. This implication may suggest a bilingual advantage of greater memory storage capacity than monolinguals. Th e tendency can be derived from enhanced executive functioning, which allows bilinguals to more efficiently access the storage system. Since bilinguals are used to inhibiting one language while using the other, the capacity to hold information and the abili ty to retrieve stored information is superior to monolinguals (Bialystok, 1999; Bialystok et al., 2004; Colzato et al., 2008; Costa et al., 2008). In contrast with the aforementioned view, some researchers challenged the assertion of a bilingual advantage. Building upon the interlingual interdependence hypothesis, Magiste (1980) demonstrated that both performance of short term and long term memory among bilinguals is less than those of monolinguals. This phenomenon is explained by a single storage system fo r both languages and hence, interference between languages is inevitable. Assuming the variance in WMC, this hypothesis could explain underlying reasons for larger digits recalled in English f a shorter
45 syllable length in Korean digit names might positively interfere during recall of English digit words, and hence resulting in larger digits recalled than Japanese speakers. Limitations and Future Directions Despite efforts to fill gaps in the previous literatures, the present study is limited in various respects. First, as discussed, control of WMC could provide more accurate results on which factor determines the differences in digit span. Preliminary experiments could be implemented to collec t an equal number of participants in accordance with WMC measure. Second, due to the nature of subjectivity in the self rating, proficiency of the second language was not precisely reported to compare with other objectively calculated variables. In the pre sent study, official test scores were collected as a measure of the proficiency; however, it is impossible to expect that participants have the same test score. Although other studies demonstrated a significant role of L2 proficiency in WMC, the present st udy did not show any significance. In light of this fact, English proficiency tests could be conducted before the experiment and use the results as an objective measure. Third, mental calculations could be reformulated to address the problems in degree of difficulty. In this way, whether there is a language preference among bilinguals as a function of word length could be examined. Fourth, as effects of orthography have been found, building word stimuli in the Japanese condition should be more carefully add ressed. There are two writing systems in Japanese; Kanji adopted Chinese character, which has spatial construction of the elements and Kana is used for native Japanese words and foreign borrowing words. Thus, it could be interesting to see whether there is a significant difference between types of scripts in short term memory of words among native Japanese speakers. In addition, implementation of eye tracking methodology
46 to more precisely investigate how speakers like Korean or Chinese read words and take a dvantage of its visual information would yield some interesting results. Fifth, although the current study did not show any correlation between age and digit span, age is one of the most well known variables that affect memory capacity; however, distributi on of age in the current study showed relatively high variability in age among three groups of participants (SD=4.28, range=12). Lastly, whereas 20 participants in the Korean English bilingual group knew the researcher personally, there is no personal rela tionship with Japanese participants or the monolingual group. This could have had a minor influence on the results; however, since the researcher was seated in the experiment room with the participant during the test and performed grading in the meantime, it is entirely possible that the Korean participants felt more pressure than any of the other groups. Therefore, control for the experimental setting could raise different results. Findings from the present study are limited, but implications could help t o further understand the cognitive architecture of memory among monolinguals and bilinguals.
47 APPENDIX A FIRST LANGUAGE DIGIT SPAN STIMULI Forward Digit Span Backward Digit Span 9 2 4 5 9 3 6 1 7 1 6 8 2 5 3 9 4 2 5 9 8 1 6 3 7 1 8 6 4 9 2 5 3 3 9 2 5 4 1 7 2 6 8 6 4 1 8 7 9 2 5 3 1 4 4 2 9 1 3 5 7 1 3 8 6 2 8 1 6 9 7 2 4 9 2 5 3 1 7 3 1 9 2 5 4 1 6 2 9 1 8 3 7 6 8 1 7 9 4 5 3 1 5 9 2 4 8 6 9 5 7 2 3 1 4 5 8 2 5 7 6 3 9 1 1 7 5 8 9 2 6 4 6 8 1 9 2 5 4 7 3 2 9 1 5 8 4 3 6 7 2 8 3 9 5 1 7 4 3 6 1 4 7 3 8 2 5 9 2 7 4 8 6 2 3 8 1 4 7 1 3 9 4 2 5 6 8 4
48 APPENDIX B ENGLISH DIGIT SPAN STIMULI Forward Digit Span Backward Digit Span 8 1 6 5 2 3 4 2 5 1 7 6 7 1 6 8 4 2 9 3 5 9 2 4 8 6 1 7 3 9 2 5 4 2 5 9 3 4 6 1 7 2 8 8 1 6 7 3 2 7 4 9 5 3 5 2 4 7 3 9 8 1 4 7 6 9 6 4 1 8 7 1 2 8 5 3 1 9 4 4 9 1 5 3 9 2 1 8 7 9 6 3 5 7 1 5 2 8 6 1 5 2 9 3 7 4 1 8 5 3 9 2 4 1 3 9 4 7 2 9 6 3 5 1 7 2 1 7 6 3 1 8 6 3 1 4 8 2 9 5 7 9 3 5 2 3 6 2 7 9 8 2 4 6 1 9 7 3 5 1 5 8 6 2 7 1 9 8 3 1 7 9 6 4 8 5 7 2 5 2 9 4 1 7 6 3 8 1
49 APPENDIX C ENGLISH OPERATION TASK STIMULI Arm Drink Year Body Minute Half Chair Home Person Spoon Sleep Water Night Rain Book Pillow Shirt Bread Door Morning Result Hand Exit Ci ty Truck Window Time Snow Answer Feet Butter Sky Work Glass Dad Student Public Beer Letter Hour Paper Fun (9/3)+2=6 2+(4X2)=10 (6X3)+4=14 (2X3) 1=4 (5/1)+4=9 (7X5)+5=40 (6/2)+3=7 (4X5)+2=22 (8X2) 4=14 (2X5) 2=8 4+(9X 2)=24 (3X9) 5=22 6 (2X1)=3 (8/2)+4=8 (5X3)+3=18 (10/5)+2=5 (9/3)+5=6 5+(4X3)=18 (7X3)+1=22 (7X1)+4=12 9 (3X2)=4 (8X2) 5=11 (6X5) 2=28 (3X7) 1=20 7+(5/1)=11 (4X8) 3=28 (5X8) 7=33 (6X2)+5=17 (8/2)+9=13 5 (3X1)=2 5 (3X1)=2 (8X3)+4=28 (9/3) 2=5 ( 4/2)+9=13 (4X4)+3=18 (8/4) 2=2 7+(3X5)=22 (8X3)+2=26 (1X9) 3=6 (6/3)+4=7 5+(3X2)=11 4+(3X6)=20
50 APPENDIX D FIRST LANGUAGE OPERATION TASK STIMULI Korean Japanese Meaning Afternoon air animal Bag bed Beer breakfast car children Student city clock coffee college Dinner egg face family Friend head home Hour key letter m other newspaper Paper party People potato radio Road salt song street Sugar summer Walk telephone tree weather winter (4/2)+8=1 0 9+(1X4)=12 (7X7)+2=51 (6X3) 4=14 (10/5)+6=10 (8/2)+3=7 (2/1)+8=9 (3X9)+2=25 (8X6) 3=45 (4X5) 9=11 9+(2X6)=21 (5X4) 5=15 8 (3X2)=2 (6/2)+3=5 (9X1)+3=12 (9/3)+8=12 (6/2)+5=10 7+(2X9)=26 (7X6)+4=58 (2X8) 5=11 8 (3X2)=3 (7X2) 5=11 (5X3) 3=13 ( 8X2) 3=13 3+(10/2)=23 (4X8) 3=28 (9X8) 8=66 (3X4)+3=15 (6/2) 1=2 10 (3X3)=1 (9X3) 2=29 (6X9) 6=46 (10/5)+7=8 (8/2)+5=11 (3X7)+6=28 (9/3) 1=2 5+(7X4)=33 (2X9)+3=15 (2X7) 7=8 (8/2)+4=8 6+(5X3)=2 9+(4X9)=48
51 REFERENC E LIST Baddeley, A. D., T homson, N., & Buchanan, M. (1975). Word length and the structure of short term memory. Journal of Verbal Learning and Verbal Behavior, 14 (6), 575 589. Baddeley, A. D., (1966a). Short term memory for word sequences as a function of acoustic, semantic and f ormal similarity. The Quarterly Journal of Experimental Psychology, 18 (4), 362 365. Baddeley, A. D., (1966 b ). The influence of acoustic and semantic similarity on long term memory for word sequences. The Quarterly Journal of Experimental Psychology, 18 (4) 302 309. Baddeley, A. D., (2003). Working memory: Looking back and looking forward. Nature Reviews Neuroscience, 4 (10), 829 839. Baddeley, A. D., & Hitch, G. J. (1974). Working memory. The Psychology of Learning and Motivation, 8 47 89. Baddeley, A. D. (1986). Working memory Oxford Oxfordshire] : New York: Clarendon Press ; Oxford University Press. Baddeley, A. D. (2000). The episodic buffer: a new component of working memory?. Trends in cognitive sciences 4 (11), 417 423. Bialystok, E. (1999). Cogn itive complexity and attentional control in the bilingual mind. Child Development, 70 (3), 636 644. Bialystok, E., Craik, F. I., Klein, R., & Viswanathan, M. (2004). Bilingualism, aging, and cognitive control: Evidence from the simon task. Psychology and A ging, 19 (2), 290. Campbell, J. I., & Clark, J. M. (1988). An encoding complex view of cognitive number processing: Comment on McCloskey, sokol, and goodman (1986). Campbell, J. I. D. (1994). Architectures for numerical cognition. Cognition, 53 (1), 1 44. Campbell, J. I. D., & Epp, L. J. (2004). An encoding complex approach to numerical cognition in chinese english bilinguals. Canadian Journal of Experimental Psychology/Revue Canadienne De Psychologie Exprimentale, 58 (4), 229. Chan, M. E., & Elliott, J. M. (2011). Cross Linguistic differences in digit memory span. Australian Psychologist, 46 (1), 25 30. Chincotta, D., & Hoosain, R. (1995). Reading rate, articulatory suppression and bilingual digit span. European Journal of Cognitive Psychology, 7 (2), 201 211.
52 Chincotta, D., & Underwood, G. (1996). Mother tongue, language of schooling and bilingual digit span. British Journal of Psychology, 87 (2), 193 208. Chincotta, D., Hyona, J., & Underwood, G. (1997). Eye fixations, speech rate, and bilingual digit span: Numeral read ing indexes fluency, not word length. Acta Psychologica, 97 253 275. Chincotta, D., & Underwood, G. (1997). Digit span and articulatory suppression: A cross linguistic comparison. European Journal of Cognitive Psychology, 9 (1), 89 96. Clarke, D. (2003). International comparative research in mathematics education. Second International Handbook of Mathematics Education, 143 184. Colzato, L. S., Bajo, M. T., van den Wildenberg, W., Paolieri, D., Nieuwenhuis, S., La Heij, W., & Hommel, B. (2008). How does bi lingualism improve executive control? A comparison of active and reactive inhibition mechanisms. Journal of Experimental Psychology: Learning, Memory, and Cognition, 34 (2), 302. Conrad, R. (1964). Acoustic confusions in immediate memory. British Journal o f Psychology, 55 (1), 75 84. Conrad, R., & Hull, A. J. (1964). Information, acoustic confusion and memory span. British Journal of Psychology (London, England : 1953), 55 (4), 429. Costa, A., Hernndez, M., & Sebastin Galls, N. (2008). Bilingualism aids conflict resolution: Evidence from the ANT task. Cognition, 106 (1), 59 86. Cowan, N. (2008). What are the differences between long term, short term, and working memory? Progress in Brain Research, 169 323 338. Cutler, A., & Otake, T. (1994). Mora or phon eme? further evidence for language specific listening. Journal of Memory and Language, 33 (6), 824 844. da Costa Pinto, A. (1991). Reading rates and digit span in bilinguals the superiority of mother tongue. International Journal of Psychology, 26 (4), 47 1 483. doi:10.1080/00207599108247135 Dehaene, S. (1992). Varieties of numerical abilities Cognition 44(1). 1 42. Ellis, N. (1992). Linguistic relativity revisited: The bilingual word length effect in working memory during counting, remembering numbers, and mental calculation. Advances in Psychology, 83 137 155. Ellis, N., & Hennelly, R. (1980). A bilingual word length effect: Implications for intelligence testing and the relative ease of mental calculation in welsh and english. British Journal of Psych ology, 71 (1), 43 51.
53 Hess, R. D., Azuma, H., Kashiwagi, K., Dickson, W. P., Nagano, S., Holloway, S., . Hatano, G. (1986). Family influences on school readiness and achievement in japan and the united states: An overview of a longitudinal study. Hoos ain, R. (1982). Correlation between pronunciation speed and digit span size. Perceptual and Motor Skills, 55 (3f), 1128 1128. Hoosain, R. (1991). Psycholinguistic implications for linguistic relativity: A case study of chinese Psychology Press. Hoosain, R ., & Salili, F. (1987). Language differences in pronunciation speed for numbers, digit span, and mathematical ability. Psychologia: An International Journal of Psychology in the Orient, Koechlin, E., Naccache, L., Block, E., & Dehaene, S. (1999). Primed n umbers: Exploring the modularity of numerical representations with masked and unmasked semantic priming. Journal of Experimental Psychology: Human Perception and Performance, 25 (6), 1882. Kolers, P. A. (1968). Bilingualism and information processing. Scie ntific American, 218 (3), 78 86. Kubozono, H. (1989). The mora and syllable structure in japanese: Evidence from speech errors. Language and Speech, 32 (3), 249 278. Longoni, A. M., Richardson, J. T., & Aiello, A. (1993). Articulatory rehearsal and phonolo gical storage in working memory. Memory & Cognition, 21 (1), 11 22. doi:10.3758/BF03211160 Mgiste, E. (1980). Memory for numbers in monolinguals and bilinguals. Acta Psychologica, 46 (1), 63 68. McCloskey, M. (1992). Cognitive mechanisms in numerical pr ocessing: Evidence from acquired dyscalculia. Cognition, 44 (1), 107 157. Miller, G. A. (1956). The magical number seven, plus or minus two: Some limits on our capacity for processing information. Psychological Review, 63 (2), 81. Naveh Benjamin, M., & Ayr es, T. J. (1986). Digit span, reading rate, and linguistic relativity. The Quarterly Journal of Experimental Psychology, 38 (4), 739 751. Otten, M., & Van Berkum, J. J. (2009). Does working memory capacity affect the ability to predict upcoming words in discourse? Brain Research, 1291 92 101. Schweickert, R., & Boruff, B. (1986). Short term memory capacity: Magic number or magic spell? Jou rnal of Experimental Psychology.Learning, Memory, and Cognition, 12 (3), 419 425. doi:10.1037/0278 73184.108.40.2069
54 Schweickert, R., Guentert, L., & Hersberger, L. (1990). Phonological similarity, pronunciation rate, and memory span. Psychological Science, 1 ( 1), 74 77. doi:10.1111/j.1467 9280.1990.tb00071.x Stevenson, H. W., Lee, S. Y., & Stigler, J. W. (1986). Mathematics achievement of chinese, japanese, and american children. Science, 231 (4739), 693 699. Stigler, J. W., Lee, S. Y., & Stevenson, H. W. (198 6). Digit memory in chinese and english: Evidence for a temporally limited store. Cognition, 23 (1), 1 20. Turner, M. L., & Engle, R. W. (1989). Is working memory capacity task dependent? Journal of Memory and Language, 28 (2), 127 154. Vaid, J., & Menon, computations. Spanish Applied Linguistics, 4 (2), 325 342. van den Noort, Maurits WML, Bosch, P., & Hugdahl, K. (2006). Foreign language proficiency and working memory capacity. European Ps ychologist, 11 (4), 289 296.
55 BIOGRAPHICAL SKETCH Aleuna Lee is from Seoul, Korea. She received a Bachelor of Science in m arketing from Bentley University in Waltham, Massachusetts. She earned a Master of Arts in Linguistic s in 2014.