Materials

國

立 政 治 大 學

‧

N a

tio na

l C h engchi U ni ve rs it y

participants all have Taiwan Mandarin as their first and primary language, and are familiar with Mandarin Phonetic Symbols (MPS; Zhuyin Symbols), presented as the format of the visual primed syllabic words in the task, with normal or corrected-to-normal vision¹³. The participants with the fundamental knowledge of the explicit linguistics are excluded in the present study.

3.1.2 Materials

Forty-two Mandarin disyllabic words including nouns (N=21) and verbs (N=21) were primed as MPS intending to get rid of lexicographical effect in the word association tasks of the present study. All Mandarin onsets (N=21), syllable structure (N=4), and tone structure (N=4) of the target words were counterbalanced with the word frequency, syntactic categories, and imageability considered¹⁴. Regarding the influences on lexical access, all the experimental words were selected based on the frequency and the syntactic categories from the Word List with Accumulated Word Frequency (http://elearning.ling.sinica.edu.tw/CWordfreq_index.html) in Academic Sinica Balanced Corpus 3.0 (Sinica Corpus) by Huang et al. (1995) and maintained by Chinese Knowledge and Information Processing Group (CKIP, 1998). The distribution of the word frequency

13 Originally 8 participants were involved in the task; however, one of them could not response well to the MPS stimuli in the practice trail, and another of them could not associate a response to more than ten stimuli, i.e., almost a quarter of them, during the task; therefore, the total responses of two participants were excluded for the further analyses.

14 The Mandarin onsets for the stimuli of the tasks in the present study do not include glides, and the syllable structures applies Duanmu’s CVX Theory (1990, 2000); see Wan, 1999 for

phonology of Mandarin, and Duanmu, 1990, 2000 for syllable structures in Mandarin.

‧

立 政 治 大 學

‧

N a

tio na

l C h engchi U ni ve rs it y

in the task is shown as Table 3-2-1, and the visualized statistics of that is demonstrated as Figure 3-2-1; see Appendix I for the complete data for the word frequency.

Table 3-2-1. The distributions of the word frequency in the task¹⁵.

POS Sum Disyllable Filtered Included Frequency Range

Na 370 285 168 21 101-134

Vc 331 218 177 21 75-170

The common nouns (Na) and the activity transitive verbs (Vc) with the frequency ranged from 101 to 134 (Mean= 114.8; SD= 17.0), and from 75 to 170 (Mean= 111.9; SD= 31.3) respectively were first filtered from the Sinica Corpus¹⁶. Monosyllabic words and words with more than two syllables were excluded from the frequency fitted words, so did those without onsets in one or both of the syllables. Therefore, 168 common nouns and 177 transitive verbs were selected and ready to be arranged as experimental materials with the appropriate phonological encoding for the present study. Finally, twenty-one words for each lexical category which can be arranged into the counterbalanced distribution of onsets, syllable structure, and tone structure were selected as the stimuli of the word association

15 POS: parts-of-speech; Sum: the total number of the nouns and verbs included in the frequency range; Disyllable: the number of disyllabic words included in the frequency range; Filtered: the number of words fits the tone and syllable structure of the present study; Included: the number of words includes in the tasks of the present study

16 There were two exceptions, a noun and a verb with lower frequency, in order to meet the phonological encoding for the purpose in the present study. The means and standard deviations

‧

立 政 治 大 學

‧

N a

tio na

l C h engchi U ni ve rs it y

tasks in the present study. The frequency ranges of the stimuli are visualized and explained in the following.

Figure 3-2-1. The boxplots of the distribution among the frequency of the nouns and verbs

The boxplots further demonstrate the range of the selected target words. Since the high frequency of nouns occurs in Mandarin, more nouns are available than verbs in the same frequency segment of the language. Therefore, in the tasks of the present study, the frequency range of the noun target words distributes more concentrative than those of verbs.

The total data of noun targets still presents a relatively symmetric distribution, i.e., the interquartile data is roughly located in the middle of the whole dataset. In comparison, the verb targets have a larger range than the nouns, and a slightly negatively skewed distribution is observed, but the total dataset is still roughly symmetric because the degree of the skewness is really small. Moreover, a Wilcoxon test was conducted to compare the means of the target words, and the result showed no significant difference in frequency between nouns and verbs (p= 0.5543), which indicates that a minimal variance of the frequency effect is involved in the present study.

Since the great effect of the initial onset in English lexicon discussed in the last chapter, the selected nouns and verbs were therefore arranged to fit all of the possible

‧

twenty-one initial consonants for each syllable beginning with [p], [pʰ], [t], [tʰ], [k], [kʰ], [m], [n], [l], [f], [x], [tɕ], [tɕʰ], [ɕ], [ts], [tsʰ], [s], [tʂ], [tʂʰ], [ʂ], and [ʐ]. Therefore, twenty-one words were selected for both nouns and verbs with the consonants of their onsets counterbalanced as Table 3-2-2 for verbs and Table 3-2-3 for nouns.

Table 3-2-2. The proportion of onsets with tones and syllable structures of target verbs

onset 1 onset 2 Gloss IPA Frequency Syllable

‧

Table 3-2-3. The proportion of the onsets, tones, syllable structures, and imageability of target nouns

onset 1 onset 2 Gloss IPA Frequency Syllable Imageability

p tʂ package pau55tʂʷaŋ55 129 CVV.CVN C

Each possible onset presented once in each initial position of the disyllabic words without the duplicated onset displayed. After the common nouns were selected, the imageability was checked on the Extended-HowNet 2.0 (E-HowNet; CKIP, 1998; Huang

‧

立 政 治 大 學

‧

N a

tio na

l C h engchi U ni ve rs it y

et al., 2008; Ma & Huang, 2006; Ma & Shih, 2018) resulting the proportion between concrete nouns (C) and abstract nouns (A) as around 6:4 (62%:38%)¹⁷. If a selected common noun is not included in E-HowNet (e.g., ‘potentiality’ [tɕʰʲɛn35nəŋ35]), the word will be checked on the Ministry of Education Recompiled Mandarin Dictionary by Taiwan Ministry of Education (1994) for the definition to classified its category. As an example,

‘potentiality’ [tɕʰʲɛn35nəŋ35] is defined as a type of ability, and the ability belongs to a subcategory of attribute, which is different from the physical category in E-HowNet;

therefore, in this case, ‘potentiality’ [tɕʰʲɛn35nəŋ35] was classified as an abstract noun.

Tone structure in Taiwan Mandarin were also arranged as a balanced distribution for the experimental stimuli including high-level (55), high-rising (35), low-falling (21), and high-falling (51) tones¹⁸. Disyllabic words with the duplicated low-falling tones do not exist in Mandarin disyllabic words phonetically because of the tone sandhi rule, i.e., when a low-falling tone precedes another low-falling tone, the first low-falling tone will become a high-rising tone as [35.21]. Therefore, neither the surface form nor the underlying form of the duplicated falling-rising tone structure were included in the tone structure for the target words as Table 3-2-5.

17 Extended-HowNet (E-HowNet) is a lexical knowledge base evolved from HowNet and created by the CKIP group for a frame-based entity-relation model to define lexical senses, see CKIP Lab for more information.

18 The conventionally accepted inventory of tones in Mandarin according to Chao’s (1930) tone numbers with ‘5’ indicating the speaker’s highest pitch, ‘3’ middle, and ‘1’ lowest; see Wan, 1999 for more information. Also, due to the dialect difference, the low-falling tone [21] in Taiwan Mandarin is phonetically presented as the fall-rise contour [214] in Beijing Mandarin,

‧

Table 3-2-5. The proportion of the tone arrangement for the target words

Na

Tone structure are also generally arranged as a balanced proportion by each combination of the four tones in Taiwan Mandarin. There are fifteen combinations of the phonetic tone structure for each lexical category involved in the word association tasks; each combination generally distributes at least once and at most twice in the twenty-one stimuli for both nouns and verbs. However, two combinations of the tone structure for verbs do not reach this goal of the tone arrangement (i.e., 55.51: N=3; 51.55: N=0) to maintain other variable control for the phonological encodings (i.e., frequency, onsets, and syllable structures) of the stimuli in the present study.

Syllable structures were taken into account in the present study as well, and the distribution of each syllable structure in each syntactic category demonstrates as Table 3-2-5¹⁹.

19 Duanmu’s CVX Theory was applied to explain the Mandarin syllable structure in this study;

see Duanmu’s (1990, 2000) for more details.

‧

立 政 治 大 學

‧

N a

tio na

l C h engchi U ni ve rs it y

Table 3-2-5. The proportion and arrangement of the syllable structure for the stimuli

Syllable structures Na Vc Sum

CVN.CVN 4 1 5

CVN.CVV 5 6 11

CVV.CVN 8 4 12

CVV.CVV 4 10 14

Sum 21 21 42

Shown as the first left column for the syllable structures of the stimuli in the table, only a nasal or a vowel can present in the syllable-final position in Mandarin. If we compare the probability for one of the nasals (i.e., [n] or [ŋ]) with one of the vowels (i.e., [i], [u], [y], [a], [ɔ], [ɤ], or [ɛ]) to present syllable-finally, the roughly half proportion of the first type of the syllable structure, that is, CVN.CVN, involving in the stimuli of the tasks can be accept. Therefore, with the same reason, the CVV.CVV structure has the highest numbers included in the stimuli, and the remainder (i.e., CVN.CVV and CVV.CVN) has the similar amount of proportion.