Syllable judgments

  Search the database

Main interfaceAlternative Interface

  (Please click here for the explanations of variables in each file.)

  About the experiment

    Overall procedure

    All participants performed four tasks, in the following order: a handedness survey (as part of the participant information form), a lexical decision task to estimate vocabulary size, a reading span task, and a wordlikeness judgment task. All tasks were conducted using E-prime 2.0 (Schneider et al., 2002). The whole task was split into two blocks because the pretests along with the main wordlikeness judgment task could take up to three hours to complete. The second session either took place on the same day (after a two-hour lunch break), or on another day (one to ten days later).

    Participants

    114 graduate and undergraduate students between 18 and 27 years old (mean = 21) were recruited from National Chung Cheng University. The results of four participants (2 males and 2 females) were excluded because they failed to return for the second session. All participants self-reported as having no visual or auditory impairment or brain damage. All participants were paid for their participation and gave their informed consent in accordance with internal research ethics standards, among other things acknowledging their right to drop from the study without penalty and consenting to the use of their anonymous data in an online database.

    Handedness survey

      Materials

      A direct Chinese translation of the Edinburgh Handedness Survey (Oldfield, 1971) was used, which asks which hand, foot or eye is preferred in various activities (e.g. writing, throwing, using a toothbrush, kicking) on the equivalent of a five-point scale from -2 (strongly prefer left) to +2 (strongly prefer right). Responses were averaged to give each participant an overall handedness score. Please click here for the content of the participant information and handedness survey.

      Procedure

      Participants saw the name f an activity on the screen in written form, and five selections, each labeled with a number to be entered. The five selections were: (1) you use your right hand, (2) you only use your right hand and never use the left hand, (3) you use your left hand, (4) you only use your left hand and never use the right hand, and (5) you can use both of your hands. There was no time limit to respond. The survey took approximately 5 minutes to complete. Responses were averaged survey items to give each participant an overall handedness score. For the database, this five-point scale was recoded as -2 (strongly prefer left) to +2 (strongly prefer right).

    Lexical decision task

      Materials

      120 pairs of real two-character Chinese words and two-character nonwords were constructed. 120 real words were chosen from the lowest frequency ranking for Accumulated Word Frequency in Sinica Corpus (中央研究院現代漢語語料庫詞頻統計; Chen et al., 1996). The nonwords were constructed to be morphologically plausible so that its intended meaning was clear; participants thus could not guess their nonlexical status from plausibility alone. This was done by matching the word-internal structure of each nonword to that of the corresponding real word. For example, the real word yàolǐ藥理 “medicinal theory” refers to a theory about medicine, while its corresponding nonword was jǐqǔ, 戟曲, where戟 is “halberd” and曲 is “song”, so that the fake word jǐqǔ would have to refer to a song about halberds.

      To limit the burden on the participants, after a pilot test we reduced the original 120 word/nonword pairs to the 21 pairs that had an accuracy below 90% (to avoid possible ceiling effects in the real experiment). The mean character frequencies for the words vs. nonwords showed no significant difference, though there was a trend for the real words to have lower-frequency characters.

      Procedure

      In the beginning of the vocabulary size task, participants were informed that they had to judge the presented vocabulary to be a real word or not by pressing either the 是 “Yes” or 不是 “No” button on the keyboard. Four practice items were provided after the instruction. There was no time limit for the judgments, and reaction times were not recorded. The task took approximately 5-10 minutes to complete. The non-speeded results for a lexical decision task have been shown to correlate with more traditional estimates of vocabulary size, though controversy remains over how best to perform the estimate (Mochida & Harrington, 2006).

    Reading span task

      Materials

      We followed the outlines of Kane et al. (2004) in developing our reading span task. A total of 60 semantically-unrelated sentences were randomly chosen from the Academia Sinica Balanced Corpus (Chen, et al., 1996). Sentence length varied from 13 to 16 characters. 30 among the 60 sentences were changed so that certain word elements made the sentences anomalous, yielding 30 normal sentences and 30 anomalous sentences, as illustrated in (1) and (2) respectively. The 60 sentences were then randomly assigned to 5 levels, from level 2 (two sentences) to level 6 (six sentences). Each level included three sets of trials. A random digit ranging from 0 to 9 followed each of the 60 sentences.

(1) Normal:

王      太太  又   為   他        特別     做了   幾樣     拿手     好   菜

wáng tàitai yòu wèi tā      tèbié     zuò-le jǐ yàng náshǒu hǎo cài

Wang Mrs. again for him especially do-ASP several CL good-at good dish

Mrs. Wang especially prepared some good dishes for him again.

(2) Anomalous:

*有些   人   一生      的 會議   都   是  混淆不清         的

yǒuxiē rén yìshēng de huìyì dōu shì hùnyáobùqīng de

some people lifetime DE meeting all SHI fuzzy DE

      Following Kane et al. (2004, p. 198), reading span scores were first calculated by taking the number of digits remembered in their correct positions in any given trial (i.e., the complement of the edit distance with target and response digit lists treated as strings), dividing this by the number of items in that trial, and then averaging these ratios across all trials for each participant. However, interpreting these scores as working memory capacity rather than simple short term memory depends on whether participants were truly paying attention to the sentences, we went beyond Kane et al. in weighting the reading span scores by the proportion of sentences correctly classified as meaningful, by multiplying these two ratios together for each participant. Thus if a participant performed perfectly on digit recall but performed at chance (0.5) on the semantic judgments, the final score would be 0.5, not 1.

      Procedure

      In the beginning of the reading span task, participants were informed that they had to judge the sentences and memorize the digits following the sentences. Participants were also informed that they could not revise their responses. Two practice sets of a two-level reading span task were given after the instruction. During the reading span task, each participant saw three sets of sentences and digits from level 2 to level 6. During the task, participants first saw a sentence followed by a digit (0-9). There was no time limit for participants to read the sentence and memorize the digit. Participants hit the space key after they thought they were ready to make judgments. Then they had to judge whether the sentence was acceptable or not. After they finished judging all sentences in a set, they were asked to recall the previous digits in the correct sequential order. For example, participants had to recall two digits after they judged both sentences in level 2. For both judgments and digit recall, there was no time limit for response. The task took approximately 10 minutes to finish.

    Wordlikeness judgment task

      Materials

      Target stimuli were nonlexical syllables written in Zhuyin Fuhao (注音符號). They consisted of all 3,274 prosodically permitted nonlexical Mandarin syllables that can be written in Zhuyin Fuhao, in the following sense. Each target stimulus consisted of a sequence of Zhuyin Fuhao symbols in the prosodically permitted order of (optional) onset consonant, (optional) medial glide, rime, tone. Zhuyin Fuhao distinguishes 22 onsets (including none), four medials (including none), four tones (not counting the mark for tonelessness, not used in this study), and 14 rimes, including none: Zhuyin Fuhao orthographic convention omits the rime symbol for non-palatal sibilants (where the vowel is essentially a sonorant realization of the onset). We excluded symbol strings without any segmental content, syllables where non-palatal non-sibilants appear without a rime symbol (56 = 4 tones × 14 such onsets), and did not include the highly restricted rhotic rime (represented in Pinyin as er), which always appears by itself in a morpheme. This gave 4516 (= 22 × 4 × 4 × (14-1) - 4 - 56) logically possible Mandarin syllables. Lexical syllables were taken to be the 1254 listed in Tsai (2000), excluding 11 toneless and four rhotic rime syllables, along with three recent neologisms (biàng, liāng, biāng in Pinyin). The total number of nonlexical test syllables was thus 3274 (= 4516 - 1254 + 11 + 4 - 3).

      Procedure

      All participants were visually presented with all 3,274 syllables, written in Zhuyin Fuhao, in two blocks (1,599 in the first, 1,675 in the second), each in random order. Each block was preceded by four practice trials to familiarize participants (or refamiliarize them, in the second session) with the task. There were untimed rest breaks every 160 trials. All participants were asked to place their right and left hands, respectively, over the right and left sides of the computer keyboard, and then to judge each stimulus as "like Mandarin" (xiàng Guóyǔ 像國語) by pressing the "L" key (left hand) or as "not like Mandarin" (bú xiàng Guóyǔ不像國語) by pressing the "S" key (right hand), as quickly as possible. Each trial began with the mark '+' for 1,000 milliseconds (i.e. one second) to ensure that their eyes were pointed in the right direction when the stimulus appeared, followed by the target syllable for 4,000 milliseconds. Each trial ended when the participant made a response or until the 4,000 ms limit was passed.
       An untimed break occurred every 80 trials (for a total of 18 breaks per session). Participants had to press SPACE to resume the experiment.

    References

      Chen, K. J., C.R. Huang, L. P. Chang & H.L. Hsu. (1996). SINICA CORPUS: Design Methodology for Balanced Corpora. Proceedings of PACLIC 11th Conference, 167-176.

      Kane, M. J., Hambrick, D. Z., Tuholski, S. W., Wilhelm, O., Payne, T. W., & Engle, R. W. (2004). The generality of working memory capacity: A latent-variable approach to verbal and visuospatial memory span and reasoning. Journal of Experimental Psychology: General, 133(2), 189-217.

      Mochida, A., & Harrington, M. (2006). The yes/no test as a measure of receptive vocabulary knowledge. Language Testing, 23(1), 73-98.

      Oldfield, R. C. (1971). The assessment and analysis of handedness: the Edinburgh inventory. Neuropsychologia, 9(1), 97-113.

      Schneider, W., Eschman, A., & Zuccolotto, A. (2002). E-Prime: User's guide. Psychology Software Incorporated.

      Tsai, C. H. (2000). Mandarin syllable frequency counts for Chinese characters. Chi-Hao Tsai’s Technology Page. Retrieved July, 4, 2009.