{ "paper_id": "O12-3004", "header": { "generated_with": "S2ORC 1.0.0", "date_generated": "2023-01-19T08:02:57.579761Z" }, "title": "Using Kohonen Maps of Chinese Morphological Families to Visualize the Interplay of Morphology and Semantics in Chinese", "authors": [ { "first": "Bruno", "middle": [], "last": "Galmar", "suffix": "", "affiliation": { "laboratory": "", "institution": "National Cheng Kung University", "location": { "settlement": "Tainan", "country": "Taiwan" } }, "email": "hsuyueshan@gmail.com" } ], "year": "", "venue": null, "identifiers": {}, "abstract": "A morphological family in Chinese is the set of compound words embedding a common morpheme, and Self-organizing maps (SOM) of these Chinese morphological families can be built. Computation of the unified-distance matrices for the SOMs allows us to perform semantic clustering of the members of the morphological families. Such semantic clustering sheds light on the interplay between morphology and semantics in Chinese. We studied how the word lists used in a lexical decision task (LDT) (Chen, Galmar, & Su, 2009) are mapped onto the clusters of the SOMs. We showed that this mapping is helpful to predict whether repetitive processing of members of a morphological family would elicit a satiation in an LDT-habituation-of both morphological and semantic units of the shared morpheme. In their LDT experiment, Chen, Galmar, and Su (2009) found evidence for morphological satiation but not for semantic satiation. Conclusions drawn from our computational experiments and calculations are in accordance with the behavioral experimental results in Chen et al. (2009). Finally, we showed that our work could be helpful to linguists in preparing adequate word lists for behavioral study of Chinese morphological families.", "pdf_parse": { "paper_id": "O12-3004", "_pdf_hash": "", "abstract": [ { "text": "A morphological family in Chinese is the set of compound words embedding a common morpheme, and Self-organizing maps (SOM) of these Chinese morphological families can be built. Computation of the unified-distance matrices for the SOMs allows us to perform semantic clustering of the members of the morphological families. Such semantic clustering sheds light on the interplay between morphology and semantics in Chinese. We studied how the word lists used in a lexical decision task (LDT) (Chen, Galmar, & Su, 2009) are mapped onto the clusters of the SOMs. We showed that this mapping is helpful to predict whether repetitive processing of members of a morphological family would elicit a satiation in an LDT-habituation-of both morphological and semantic units of the shared morpheme. In their LDT experiment, Chen, Galmar, and Su (2009) found evidence for morphological satiation but not for semantic satiation. Conclusions drawn from our computational experiments and calculations are in accordance with the behavioral experimental results in Chen et al. (2009). Finally, we showed that our work could be helpful to linguists in preparing adequate word lists for behavioral study of Chinese morphological families.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "Abstract", "sec_num": null } ], "body_text": [ { "text": "In this paper, we call a morphological family the set of compound words embedding a common morpheme. Hence, the compound words in Table 1 , which all contain the morpheme '\u660e' (m\u00edng) as a first character, belong to the morphological family of '\u660e'. Table 1 .Some examples of the\u660e morphological family (Chen, Galmar, & Su, 2009 In Chinese, the meaning of a morpheme can be either transparent or opaque to the meaning of the compound word embedding it. For example, the common morpheme in Table 1 \"\u660e\" can mean (clear) or (bright) and is transparent to the meaning of \"\u660e\u661f\" (star) but rather opaque to the meaning of \"\u660e\u5929\" (tomorrow). If some members of a morphological family are semantically similar, one could advance as a reason for such a similarity that these members are transparent to the same meaning of the shared morpheme. Most Chinese morphemes are polysemous (Chen & Chen, 2000) . Hence, in theory, transparent members of a morphological family could belong to different semantic clusters whose centers would be the different meanings of the shared polysemous morpheme.", "cite_spans": [ { "start": 299, "end": 324, "text": "(Chen, Galmar, & Su, 2009", "ref_id": "BIBREF1" }, { "start": 865, "end": 884, "text": "(Chen & Chen, 2000)", "ref_id": "BIBREF2" } ], "ref_spans": [ { "start": 130, "end": 137, "text": "Table 1", "ref_id": null }, { "start": 247, "end": 254, "text": "Table 1", "ref_id": null }, { "start": 485, "end": 492, "text": "Table 1", "ref_id": null } ], "eq_spans": [], "section": "Introduction", "sec_num": "1." }, { "text": "This paper is aimed primarily at using computational linguistics methods to perform semantic clustering of the members of the morphological families. Such a clustering is used to predict the results of a behavioral Lexical Decision Task 1 (LDT) designed by Chen, Galmar, & Su (2009) to study the phenomenon of morphological satiation in Chinese.", "cite_spans": [ { "start": 257, "end": 282, "text": "Chen, Galmar, & Su (2009)", "ref_id": "BIBREF1" } ], "ref_spans": [], "eq_spans": [], "section": "Bruno Galmar", "sec_num": "56" }, { "text": "In visual word recognition, morphological satiation is an impairment of morphological processing induced by repetitive exposure to the same morpheme embedded in different Chinese compound words (Chen et al., 2009; Cheng & Lan, 2009) . Chen, Galmar, and Su (2009) posited that morphological satiation is due to habituation of the morphological unit of the repeated morpheme. This is represented in Figure 1 by Diagram (a).", "cite_spans": [ { "start": 194, "end": 213, "text": "(Chen et al., 2009;", "ref_id": "BIBREF1" }, { "start": 214, "end": 232, "text": "Cheng & Lan, 2009)", "ref_id": null }, { "start": 235, "end": 262, "text": "Chen, Galmar, and Su (2009)", "ref_id": "BIBREF1" } ], "ref_spans": [ { "start": 397, "end": 405, "text": "Figure 1", "ref_id": "FIGREF1" } ], "eq_spans": [], "section": "Bruno Galmar", "sec_num": "56" }, { "text": "As a morpheme is thought to be a meaningful unit, it is logical to consider whether a semantic satiation (Kounios, Kotz, & Holcomb, 2000; Smith & Klein, 1990; Tian & Huber, 2010) , an impairment of semantic processing causing a temporary loss of the meaning of the common morpheme, would occur concomitantly with morphological satiation. 2 In other words, the satiation observed by Chen et al. (2009) could have two loci: a morphological locus and a semantic locus, as represented in Figure 1 ", "cite_spans": [ { "start": 105, "end": 137, "text": "(Kounios, Kotz, & Holcomb, 2000;", "ref_id": "BIBREF15" }, { "start": 138, "end": 158, "text": "Smith & Klein, 1990;", "ref_id": "BIBREF22" }, { "start": 159, "end": 178, "text": "Tian & Huber, 2010)", "ref_id": "BIBREF23" }, { "start": 382, "end": 400, "text": "Chen et al. (2009)", "ref_id": "BIBREF1" } ], "ref_spans": [ { "start": 484, "end": 492, "text": "Figure 1", "ref_id": "FIGREF1" } ], "eq_spans": [], "section": "Bruno Galmar", "sec_num": "56" }, { "text": "A morphological satiation could also have its loci of satiation on the links between the morphological, lexical, and semantic units, as represented in Figure 1 by Diagrams (b) and (c). We quickly can rule out the possibility of a locus on the link between morphological and 1 An LDT is a behavioral task for which subjects have to identify whether presented visual stimuli are words or non-words. 2 If most of the members of a morphological family used in an experimental task are transparent to the same meaning of the shared morpheme, the same semantic units of the shared morpheme are repeatedly accessed and finally habituate -satiation diagram (d) -. Therefore, there could be a semantic satiation in addition to morphological satiation. ", "cite_spans": [ { "start": 397, "end": 398, "text": "2", "ref_id": null } ], "ref_spans": [ { "start": 151, "end": 159, "text": "Figure 1", "ref_id": "FIGREF1" } ], "eq_spans": [], "section": "by Diagram (d).", "sec_num": null }, { "text": "As human subject agreement for semantic clustering tasks is low (Jorgensen, 1990) , computational corpus-based semantic clustering was thought to be a valuable and complementary experimental approach compared to a behavioral one with human subjects.", "cite_spans": [ { "start": 64, "end": 81, "text": "(Jorgensen, 1990)", "ref_id": "BIBREF11" } ], "ref_spans": [], "eq_spans": [], "section": "Rationale of our Approach", "sec_num": "2." }, { "text": "A corpus of written texts is a human artifact, its content is relevant to the human reader and, from a cognitive psychology standpoint, a corpus does embed a subset of organized human semantic knowledge and is worthy to be studied in computer simulations as a pure abstract semantic memory stripped out of sensory and motor representations.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "Rationale of our Approach", "sec_num": "2." }, { "text": "In natural language processing, proponents of the \"bag of words\" approach simplify each document internal structure to a set of words and use a whole corpus to build a matrix of co-occurrence of the words corpus (Landauer & Dumais, 1997) . Computational methods, such as Latent Semantic Analysis (LSA), take as input such a high dimensional matrix and reduce its dimensionality to form a vector space of the documents and words (Landauer, McNamara, Dennis, & Kintsch, 2007) . This space embeds only an associative kind of semantic information 3 : words that co-occur in the same documents or that have common co-occurrents are close associates.", "cite_spans": [ { "start": 212, "end": 237, "text": "(Landauer & Dumais, 1997)", "ref_id": "BIBREF16" }, { "start": 428, "end": 473, "text": "(Landauer, McNamara, Dennis, & Kintsch, 2007)", "ref_id": "BIBREF17" } ], "ref_spans": [], "eq_spans": [], "section": "Rationale of our Approach", "sec_num": "2." }, { "text": "For a news corpus, the association can often be of the situational type. For example, \"Father Christmas\" will be a close associate of \"department store,\" as there are many news reports around Christmas about the bustling agitation in department stores full of \"Father Christmas\". 4 In cognitive science and AI, it is said that the two terms \"Father Christmas\" and \"department store\" belong to a common memory frame, a frame being defined by Minsky as \"a data-structure for representing a stereotyped situation\" (Minsky, 1974) .", "cite_spans": [ { "start": 280, "end": 281, "text": "4", "ref_id": null }, { "start": 511, "end": 525, "text": "(Minsky, 1974)", "ref_id": "BIBREF21" } ], "ref_spans": [], "eq_spans": [], "section": "Rationale of our Approach", "sec_num": "2." }, { "text": "In the present work, we follow a \"bag of words\" approach by first building a term document matrix (TDM). Then, Self-Organizing Maps (SOMs) and associated unified-distanced matrices (called U-matrix thereafter) are built from the TDM. The SOMs and U-matrices serve to visualize semantic clusters in a morphological family on a 2D hexagonal grid of bins (Kohonen, 2001 ).", "cite_spans": [ { "start": 352, "end": 366, "text": "(Kohonen, 2001", "ref_id": "BIBREF12" } ], "ref_spans": [], "eq_spans": [], "section": "Rationale of our Approach", "sec_num": "2." }, { "text": "On the SOMs, a semantic cluster is made of members of a morphological family that have been fitted into the same bin of the grid and into contiguous bins which are close neighbors -according to the U-matrix information -in the original high dimensional space. SOMs have been used successfully to capture associative semantic relationships between words in corpora. Closer to the present approach, SOMs have been used to study the developmental aspect of vocabulary acquisition in Chinese (Li, 2001 (Li, , 2009 Li, Farkas, & Visualize the Interplay of Morphology and Semantics in Chinese MacWhinney, 2004; Zhao & Li, 2008) . Zhao, Li, and Kohonen (2010) studied the clustering of subsets of the most common Chinese words along both linguistic and semantic dimensions. Kohonen and Xing (2011) computed the SOMs of different linguistic classes for Chinese and studied how word frequency modulates the clusters on the SOMs. Our study is the first one to use SOMs to study the interplay between morphology and semantics in Chinese compounds words sharing a common morpheme, i.e. to study the semantics of morphological families. Previous studies on the semantics of morphological families (Galmar & Chen, 2010a; Galmar & Chen, 2010b ) followed a supervised approach that relied upon etymological knowledge of Chinese morphemes to identify the meaning of a morpheme in a given compound word. The present work followed an unsupervised approach, and the goal is meaning discrimination through clustering rather than meaning identification. We chose the SOM algorithm to ensure that our work can be replicated thanks to the widespread availability of SOM packages and toolboxes and because it has not yet been applied to our specific research topic.", "cite_spans": [ { "start": 488, "end": 497, "text": "(Li, 2001", "ref_id": "BIBREF18" }, { "start": 498, "end": 509, "text": "(Li, , 2009", "ref_id": "BIBREF19" }, { "start": 510, "end": 523, "text": "Li, Farkas, &", "ref_id": "BIBREF20" }, { "start": 587, "end": 604, "text": "MacWhinney, 2004;", "ref_id": "BIBREF20" }, { "start": 605, "end": 621, "text": "Zhao & Li, 2008)", "ref_id": "BIBREF27" }, { "start": 624, "end": 652, "text": "Zhao, Li, and Kohonen (2010)", "ref_id": "BIBREF28" }, { "start": 767, "end": 790, "text": "Kohonen and Xing (2011)", "ref_id": "BIBREF14" }, { "start": 1184, "end": 1206, "text": "(Galmar & Chen, 2010a;", "ref_id": null }, { "start": 1207, "end": 1227, "text": "Galmar & Chen, 2010b", "ref_id": null } ], "ref_spans": [], "eq_spans": [], "section": "Rationale of our Approach", "sec_num": "2." }, { "text": "We used the Academia Sinica Balanced Corpus (ASBC), a five million word annotated corpus based on Chinese materials from Taiwan, mostly newspapers articles. The corpus is made of roughly 10000 documents of unequal length.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "The Academia Sinica Balanced Corpus", "sec_num": "3.1" }, { "text": "We removed the foreign alphabet words and most of the Chinese functional words from the corpus. We kept POS tag information to allow differentiation between different grammatical instances of the same word 5 .", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "The Academia Sinica Balanced Corpus", "sec_num": "3.1" }, { "text": "The TDM was built using the TermDocumentMatrix function of the R package tm (Feinerer, 2008 ) with a self-customized Chinese tokenizer. The TDM is a 136570 terms * 9179 documents matrix.", "cite_spans": [ { "start": 76, "end": 91, "text": "(Feinerer, 2008", "ref_id": "BIBREF8" } ], "ref_spans": [], "eq_spans": [], "section": "The Term Document Matrix (TDM)", "sec_num": "3.2" }, { "text": "The TDM was weighted:", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "The Term Document Matrix (TDM)", "sec_num": "3.2" }, { "text": "1.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "The Term Document Matrix (TDM)", "sec_num": "3.2" }, { "text": "Using the classical term frequency-inverse document frequency (TfIdf) weighting scheme for both local and global weighting of the terms in the TDM (Landauer & Dumais, 1997) . We used the function weightTfIdf of the package tm (Feinerer, 2008) .", "cite_spans": [ { "start": 147, "end": 172, "text": "(Landauer & Dumais, 1997)", "ref_id": "BIBREF16" }, { "start": 226, "end": 242, "text": "(Feinerer, 2008)", "ref_id": "BIBREF8" } ], "ref_spans": [], "eq_spans": [], "section": "The Term Document Matrix (TDM)", "sec_num": "3.2" }, { "text": "Using a weighting scheme at the document level to reduce the effect of the size difference between documents: Each document of the TDM is a genuine article of the ASBC corpus and is considered as a semantic unit. More weight is given to small documents of the ASBC corpus. A complete justification for such a decision is given in . Briefly, one can say that the gist of a news article is easier to extract from a very short article than from a very long one for a human reader due to attention capacity limitations.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "2.", "sec_num": null }, { "text": "For a given morphological family, the rows corresponding to the members of the family in the TDM were extracted. The extracted rows constitute a submatrix of the TDM. From this submatrix, an SOM is built using the Batch map algorithm (Kohonen, 2001 ). The U-matrix (Ultsch & Siemon, 1990 ) is computed to assess how close members fitted to contiguous bins (bins are hereafter called units) on the SOM are in the original high-dimensional space (hereafter called input data space).", "cite_spans": [ { "start": 234, "end": 248, "text": "(Kohonen, 2001", "ref_id": "BIBREF12" }, { "start": 265, "end": 287, "text": "(Ultsch & Siemon, 1990", "ref_id": "BIBREF24" } ], "ref_spans": [], "eq_spans": [], "section": "The Self-Organizing Maps", "sec_num": "4." }, { "text": "As all of the data -the TDM -can be presented to the SOM algorithm from the beginning of learning, the batch version of the SOM algorithm (called \"Batch Map\") is used instead of the incremental learning SOM algorithm. The batch SOM is very similar to the k-means (Linde-Buzo-Gray) algorithm (Kohonen, 2001 ).", "cite_spans": [ { "start": 291, "end": 305, "text": "(Kohonen, 2001", "ref_id": "BIBREF12" } ], "ref_spans": [], "eq_spans": [], "section": "The batch version of the SOM algorithm", "sec_num": "4.1" }, { "text": "Our SOM defines a mapping from the input data space \u211c n of observation samples onto a hexagonal two-dimensional grid of N u units. Every unit i is associated with a reference vector m i \u03b5 \u211c n . The set of units located inside a given radius from unit i is termed neighborhood set N i .", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "The batch version of the SOM algorithm", "sec_num": "4.1" }, { "text": "The Batch Map algorithm can be described as follows (Kohonen, 2001, p. 139-140; Kohonen, Oja, Visa, & Kangas, 2002 , p. 1360 ).", "cite_spans": [ { "start": 52, "end": 79, "text": "(Kohonen, 2001, p. 139-140;", "ref_id": null }, { "start": 80, "end": 114, "text": "Kohonen, Oja, Visa, & Kangas, 2002", "ref_id": null }, { "start": 115, "end": 124, "text": ", p. 1360", "ref_id": null } ], "ref_spans": [], "eq_spans": [], "section": "The batch version of the SOM algorithm", "sec_num": "4.1" }, { "text": "Initialize the N u reference vectors by taking the first N u observation samples.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "1.", "sec_num": null }, { "text": "For each unit i, collect a list L i of copies of all those observation samples whose nearest reference vector belongs to N i .", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "2.", "sec_num": null }, { "text": "Update the value of each reference vector m i with the mean over L i .", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "3.", "sec_num": null }, { "text": "Step 2 a few times.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "Repeat from", "sec_num": "4." }, { "text": "The Batch Map presents a main advantage over the incremental learning version of the SOM algorithm (Kohonen, 2001; Fort, Letremy, & Cottrell, 2002) in that no learning rate Visualize the Interplay of Morphology and Semantics in Chinese parameter has to be specified. To double-check the computed batch SOM's representativeness of the input data space, we followed the recommendation of both Fort, Letremy, and Cottrell (2002) and Kohonen (2001) to compare organization in the Batch Map and in the incremental learning SOM.", "cite_spans": [ { "start": 99, "end": 114, "text": "(Kohonen, 2001;", "ref_id": "BIBREF12" }, { "start": 115, "end": 147, "text": "Fort, Letremy, & Cottrell, 2002)", "ref_id": null }, { "start": 391, "end": 425, "text": "Fort, Letremy, and Cottrell (2002)", "ref_id": null }, { "start": 430, "end": 444, "text": "Kohonen (2001)", "ref_id": "BIBREF12" } ], "ref_spans": [], "eq_spans": [], "section": "Repeat from", "sec_num": "4." }, { "text": "We used the code in the R package class (Venables & Ripley, 2002) for the batch SOM to build the SOMs. For the morphological family \u8a08 (j\u00ed), used as an illustrative example in Section 5, we built the SOMs on a 7*8 hexagonal grid of 56 bins. The size of the SOM was determined by experimental testing while ensuring a lower quantization error (Kohonen, 2001 ).", "cite_spans": [ { "start": 40, "end": 65, "text": "(Venables & Ripley, 2002)", "ref_id": "BIBREF25" }, { "start": 341, "end": 355, "text": "(Kohonen, 2001", "ref_id": "BIBREF12" } ], "ref_spans": [], "eq_spans": [], "section": "Repeat from", "sec_num": "4." }, { "text": "We reused and modified the code in the R package kohonen (Wehrens & Buydens, 2007) to build the U-matrix for the Batch Map and to plot a grey-level map superimposed on the SOM map. The U-matrix is the distance matrix between the reference vectors of contiguous units. On the grayscale SOMs, contiguous units in a light shade on the SOM are representative of existing clusters in the input data space. Contiguous units in a dark shade draw boundaries between existing clusters in the input data space (Ultsch & Siemon, 1990) .", "cite_spans": [ { "start": 57, "end": 82, "text": "(Wehrens & Buydens, 2007)", "ref_id": "BIBREF26" }, { "start": 500, "end": 523, "text": "(Ultsch & Siemon, 1990)", "ref_id": "BIBREF24" } ], "ref_spans": [], "eq_spans": [], "section": "The Unified-Distance Matrix", "sec_num": "4.2" }, { "text": "We present the results for the study of the \u8a08 (j\u00ed) morphological family. 6 This Chinese morpheme has two main meanings: (1) to count, to calculate or (2) to plan, to scheme. The study was limited to the members in the ASBC corpus embedding \u8a08 as a first character. The SOM map of these members is noted SOM 93 and is shown in Figure 2 .", "cite_spans": [], "ref_spans": [ { "start": 325, "end": 333, "text": "Figure 2", "ref_id": "FIGREF2" } ], "eq_spans": [], "section": "Results", "sec_num": "5." }, { "text": "At the first level, the map is divided in two zones: a dark shaded one -the upper part of the map -and a light shaded one. Most of the words belong to the light shaded zone. Among the diverse existing clusters, we note that:", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "Results", "sec_num": "5." }, { "text": "Cluster C 1 mainly gathers word sharing and other words related to the first meaning of \u8a08.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "Results", "sec_num": "5." }, { "text": "Cluster C 2 gathers three words related to the frame taxi in the same unit.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "Results", "sec_num": "5." }, { "text": "Cluster C 3 includes many words belonging to two contiguous units in a light shade. We decided to recompute a Batch Map SOM for the members in these two units to zoom in and have a clearer map of these members. The map is shown in Figure 3 . Chen et al. (2009) in one block of their LDT experiment. 7 Some of the words have two POS tags, so the total number of data points represented in Figure 4 is 17. Clustering is observed easily with such a few words. Three contiguous units in a light shade form the unique big cluster with a total of six different words. In the latest experimental research on semantic satiation, Tian and Huber (2010) found that after five or seven repetitions of a given word, the word's meaning starts to be satiated. From two to five repetitions, there is semantic priming -behavioral enhancement in semantic tasks -and more repetitions are the realm of semantic satiation.", "cite_spans": [ { "start": 242, "end": 260, "text": "Chen et al. (2009)", "ref_id": "BIBREF1" }, { "start": 621, "end": 642, "text": "Tian and Huber (2010)", "ref_id": "BIBREF23" } ], "ref_spans": [ { "start": 231, "end": 239, "text": "Figure 3", "ref_id": "FIGREF3" }, { "start": 388, "end": 396, "text": "Figure 4", "ref_id": "FIGREF4" } ], "eq_spans": [], "section": "Results", "sec_num": "5." }, { "text": "If, in Chen et al. (2009) 's lexical decision task (LDT), these six 8 words occur successively, there should be semantic satiation. In Chen et al.'s LDT, the 13 words in Figure 3 were randomly mixed with 13 non-words. Non-words, being meaningless, should not contribute significantly to satiate the semantic units of the different meanings of the shared morpheme. Therefore, from the analysis of our SOM, we predict that there could be a preliminary sign of semantic satiation only in the case where the 6 members of the big cluster occur successively in the 26-word list -we call this the best case.", "cite_spans": [ { "start": 7, "end": 25, "text": "Chen et al. (2009)", "ref_id": "BIBREF1" } ], "ref_spans": [ { "start": 170, "end": 179, "text": "Figure 3", "ref_id": "FIGREF3" } ], "eq_spans": [], "section": "Visualize the Interplay of Morphology and Semantics in Chinese", "sec_num": null }, { "text": "To compute the probability of this best case, we need to calculate two numbers:", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "Visualize the Interplay of Morphology and Semantics in Chinese", "sec_num": null }, { "text": "1. N a , the number of distinguishable arrangements of n=26 words of which 6belonging to our big cluster -constitute a first set S1 and the 20 remaining ones constitute another set S2. The order of occurrence of the 6 words of S1 does not matter; therefore, the words of S1 are considered to be of a same type T1. For the same reason, words of S2 are of a same type T2, different from type T1.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "Visualize the Interplay of Morphology and Semantics in Chinese", "sec_num": null }, { "text": "26! 230230 6!20! a N = = (2) 2.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "Visualize the Interplay of Morphology and Semantics in Chinese", "sec_num": null }, { "text": "The number of distinguishable arrangements of 6 successive occurrences of S1 words 9 in a 26-word list: 21.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "Visualize the Interplay of Morphology and Semantics in Chinese", "sec_num": null }, { "text": "The probability p of the best case is given by dividing the number of distinguishable arrangements of 6 successive occurrences of S1 words by the number of distinguishable arrangements of n=26 words made of the two types T1 and T2. ", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "Visualize the Interplay of Morphology and Semantics in Chinese", "sec_num": null }, { "text": "This best case has a very low probability, so subjects in Chen et al. would almost always be given a 26-word list that do not warrant -according to our analysis -elicitation of semantic satiation.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "Visualize the Interplay of Morphology and Semantics in Chinese", "sec_num": null }, { "text": "Hence, we agree with Chen et al. that there was no semantic locus of satiation in their experiment. On the other hand, we refine Chen et al.'s conclusions by advancing that one could prepare specific experimental word lists that would maximize the probability of observing semantic satiation.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "Visualize the Interplay of Morphology and Semantics in Chinese", "sec_num": null }, { "text": "By visualizing the SOMs augmented with neighboring distance information from the U-matrix, one can observe whether semantic clusters exist in a morphological family and how the 66 Bruno Galmar experimental data in Chen et al. (2009) is mapped to these clusters.", "cite_spans": [ { "start": 214, "end": 232, "text": "Chen et al. (2009)", "ref_id": "BIBREF1" } ], "ref_spans": [], "eq_spans": [], "section": "General Conclusion", "sec_num": "6." }, { "text": "Conclusions drawn from our computational experimental results are in accordance with Chen et al.'s behavioral experimental results revealing the absence of a semantic satiation while morphological satiation occurs. Nevertheless, we propose that semantic satiation theoretically could be elicited with specifically arranged word lists for Chen et al.'s experiment. Such lists have a very low probability of occurrence when a random assignment of words is used to prepare experimental word lists. Therefore, the present work shows the necessity of preparing adequate experimental word lists based on computational semantic clustering -as shown here -or human norms of semantic similarity if available.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "General Conclusion", "sec_num": "6." }, { "text": "Alternatives to SOMs, such as GTM (Bishop, Svensen, & Williams, 1998) , exist and could be used for comparison purposes with the present results.", "cite_spans": [ { "start": 34, "end": 69, "text": "(Bishop, Svensen, & Williams, 1998)", "ref_id": null } ], "ref_spans": [], "eq_spans": [], "section": "Future Directions", "sec_num": "7." }, { "text": "The source code and R command lines are available upon request in a script file. In order to run the whole script file from the very beginning, one needs the Academia Sinica Balanced Corpus (ASBC). The ASBC has to be purchased. 10 ", "cite_spans": [ { "start": 228, "end": 230, "text": "10", "ref_id": "BIBREF5" } ], "ref_spans": [], "eq_spans": [], "section": "Code to generate the SOMs from the ASBC corpus", "sec_num": "8." }, { "text": "Semantic information also can be, for example, of the categorical or featural types.4 This example is borrowed fromGalmar & Chen (2010b).", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "", "sec_num": null }, { "text": "Some of the Chinese words can have up to 5 different POS tags[10].", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "", "sec_num": null }, { "text": "Others examples are also given in the script file -available upon request -to create and plot the SOMs presented in the present paper.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "", "sec_num": null }, { "text": "One word of the original experiment not existing in the ASBC corpus is missing here.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "", "sec_num": null }, { "text": "Tian and Huber (2010) found satiation effects after six repetitive accesses to a word's meaning.9 Order of occurrence of the S1 words does not matter.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "", "sec_num": null } ], "back_matter": [ { "text": "This work was supported by a doctoral fellowship grant (NSC100-2420-H-006-007-DR) awarded to Bruno Galmar.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "Acknowledgements", "sec_num": null } ], "bib_entries": { "BIBREF0": { "ref_id": "b0", "title": "GTM: The generative topographic mapping", "authors": [ { "first": "C", "middle": [ "M" ], "last": "Bishop", "suffix": "" }, { "first": "M", "middle": [], "last": "Svens\u00e9n", "suffix": "" }, { "first": "C", "middle": [ "K I" ], "last": "Williams", "suffix": "" } ], "year": 1998, "venue": "Neural computation", "volume": "10", "issue": "1", "pages": "215--234", "other_ids": {}, "num": null, "urls": [], "raw_text": "Bishop, C. M., Svens\u00e9n, M., & Williams, C. K. I. (1998). GTM: The generative topographic mapping. Neural computation, 10(1), 215-234.", "links": null }, "BIBREF1": { "ref_id": "b1", "title": "Semantic Satiation of Chinese Characters in a Continuous Lexical Decision Task", "authors": [ { "first": "J", "middle": [ "Y" ], "last": "Chen", "suffix": "" }, { "first": "B", "middle": [], "last": "Galmar", "suffix": "" }, { "first": "H", "middle": [ "J" ], "last": "Su", "suffix": "" } ], "year": 2009, "venue": "", "volume": "", "issue": "", "pages": "", "other_ids": {}, "num": null, "urls": [], "raw_text": "Chen, J. Y., Galmar, B., & Su, H. J. (2009). Semantic Satiation of Chinese Characters in a Continuous Lexical Decision Task.", "links": null }, "BIBREF2": { "ref_id": "b2", "title": "Automatic semantic classification for Chinese unknown compound nouns", "authors": [ { "first": "K", "middle": [ "J" ], "last": "Chen", "suffix": "" }, { "first": "C", "middle": [], "last": "Chen", "suffix": "" } ], "year": 2000, "venue": "", "volume": "", "issue": "", "pages": "", "other_ids": {}, "num": null, "urls": [], "raw_text": "Chen, K. J., & Chen, C. (2000). Automatic semantic classification for Chinese unknown compound nouns.", "links": null }, "BIBREF3": { "ref_id": "b3", "title": "An implicit test of Chinese orthographic satiation", "authors": [ { "first": "C", "middle": [ "M" ], "last": "Cheng", "suffix": "" }, { "first": "Y", "middle": [ "H" ], "last": "Lan", "suffix": "" } ], "year": 2011, "venue": "Reading and Writing", "volume": "24", "issue": "1", "pages": "55--90", "other_ids": {}, "num": null, "urls": [], "raw_text": "Cheng, C. M., & Lan, Y. H. (2011). An implicit test of Chinese orthographic satiation. Reading and Writing, 24(1), 55-90.", "links": null }, "BIBREF4": { "ref_id": "b4", "title": "Advantages and drawbacks of the batch Kohonen Algorithm", "authors": [ { "first": "M", "middle": [], "last": "Cottrel", "suffix": "" }, { "first": "J", "middle": [], "last": "Fort", "suffix": "" }, { "first": "P", "middle": [], "last": "Letremy", "suffix": "" } ], "year": 2005, "venue": "", "volume": "", "issue": "", "pages": "", "other_ids": {}, "num": null, "urls": [], "raw_text": "Cottrel, M., Fort, J., & Letremy, P. (2005). Advantages and drawbacks of the batch Kohonen Algorithm.", "links": null }, "BIBREF5": { "ref_id": "b5", "title": "Contact the Academia Sinica (\u4e2d\u592e\u7814\u7a76\u9662\u8a9e\u8a00\u6240)", "authors": [], "year": null, "venue": "", "volume": "", "issue": "", "pages": "", "other_ids": {}, "num": null, "urls": [], "raw_text": "Contact the Academia Sinica (\u4e2d\u592e\u7814\u7a76\u9662\u8a9e\u8a00\u6240).", "links": null }, "BIBREF6": { "ref_id": "b6", "title": "Using Kohonen Maps of Chinese Morphological Families to 67", "authors": [], "year": null, "venue": "", "volume": "", "issue": "", "pages": "", "other_ids": {}, "num": null, "urls": [], "raw_text": "Using Kohonen Maps of Chinese Morphological Families to 67", "links": null }, "BIBREF7": { "ref_id": "b7", "title": "Visualize the Interplay of Morphology and Semantics in Chinese", "authors": [], "year": null, "venue": "", "volume": "", "issue": "", "pages": "", "other_ids": {}, "num": null, "urls": [], "raw_text": "Visualize the Interplay of Morphology and Semantics in Chinese", "links": null }, "BIBREF8": { "ref_id": "b8", "title": "Text mining infrastructure in R", "authors": [ { "first": "I", "middle": [], "last": "Feinerer", "suffix": "" }, { "first": "K", "middle": [], "last": "Hornik", "suffix": "" }, { "first": "D", "middle": [], "last": "Meyer", "suffix": "" } ], "year": 2008, "venue": "Journal of Statistical Software", "volume": "25", "issue": "5", "pages": "1--54", "other_ids": {}, "num": null, "urls": [], "raw_text": "Feinerer, I., Hornik, K., & Meyer, D. (2008). Text mining infrastructure in R. Journal of Statistical Software, 25(5), 1-54.", "links": null }, "BIBREF9": { "ref_id": "b9", "title": "Identifying Different Meanings of a Chinese Morpheme through Latent Semantic Analysis and Minimum Spanning Tree Analysis", "authors": [ { "first": "B", "middle": [], "last": "Galmar", "suffix": "" }, { "first": "J", "middle": [], "last": "Chen", "suffix": "" } ], "year": 2010, "venue": "International Journal of Computational Linguistics and Applications", "volume": "1", "issue": "1-2", "pages": "153--168", "other_ids": {}, "num": null, "urls": [], "raw_text": "Galmar, B., & Chen, J. (2010). Identifying Different Meanings of a Chinese Morpheme through Latent Semantic Analysis and Minimum Spanning Tree Analysis. International Journal of Computational Linguistics and Applications, 1(1-2), 153-168.", "links": null }, "BIBREF10": { "ref_id": "b10", "title": "Identifying different meanings of a Chinese morpheme through semantic pattern matching in augmented minimum spanning trees", "authors": [ { "first": "B", "middle": [], "last": "Galmar", "suffix": "" }, { "first": "J", "middle": [ "Y" ], "last": "Chen", "suffix": "" } ], "year": 2010, "venue": "The Prague Bulletin of Mathematical Linguistics", "volume": "94", "issue": "1", "pages": "15--34", "other_ids": {}, "num": null, "urls": [], "raw_text": "Galmar, B., & Chen, J. Y. (2010). Identifying different meanings of a Chinese morpheme through semantic pattern matching in augmented minimum spanning trees. The Prague Bulletin of Mathematical Linguistics, 94(-1), 15-34.", "links": null }, "BIBREF11": { "ref_id": "b11", "title": "The psychological reality of word senses", "authors": [ { "first": "J", "middle": [ "C" ], "last": "Jorgensen", "suffix": "" } ], "year": 1990, "venue": "Journal of Psycholinguistic Research", "volume": "19", "issue": "3", "pages": "167--190", "other_ids": {}, "num": null, "urls": [], "raw_text": "Jorgensen, J. C. (1990). The psychological reality of word senses. Journal of Psycholinguistic Research, 19(3), 167-190.", "links": null }, "BIBREF12": { "ref_id": "b12", "title": "Self-Organizing Maps", "authors": [ { "first": "T", "middle": [], "last": "Kohonen", "suffix": "" } ], "year": 2001, "venue": "", "volume": "", "issue": "", "pages": "", "other_ids": {}, "num": null, "urls": [], "raw_text": "Kohonen, T. (2001). Self-Organizing Maps: Springer.", "links": null }, "BIBREF13": { "ref_id": "b13", "title": "Engineering applications of the self-organizing map", "authors": [ { "first": "T", "middle": [], "last": "Kohonen", "suffix": "" }, { "first": "E", "middle": [], "last": "Oja", "suffix": "" }, { "first": "O", "middle": [], "last": "Simula", "suffix": "" }, { "first": "A", "middle": [], "last": "Visa", "suffix": "" }, { "first": "J", "middle": [], "last": "Kangas", "suffix": "" } ], "year": 1996, "venue": "Proceedings of the IEEE", "volume": "84", "issue": "10", "pages": "1358--1384", "other_ids": {}, "num": null, "urls": [], "raw_text": "Kohonen, T., Oja, E., Simula, O., Visa, A., & Kangas, J. (1996). Engineering applications of the self-organizing map. Proceedings of the IEEE, 84(10), 1358-1384.", "links": null }, "BIBREF14": { "ref_id": "b14", "title": "Contextually self-organized maps of chinese words", "authors": [ { "first": "T", "middle": [], "last": "Kohonen", "suffix": "" }, { "first": "H", "middle": [], "last": "Xing", "suffix": "" } ], "year": 2011, "venue": "Advances in Self-Organizing Maps", "volume": "", "issue": "", "pages": "16--29", "other_ids": {}, "num": null, "urls": [], "raw_text": "Kohonen, T., & Xing, H. (2011). Contextually self-organized maps of chinese words. Advances in Self-Organizing Maps, 16-29.", "links": null }, "BIBREF15": { "ref_id": "b15", "title": "On the locus of the semantic satiation effect: Evidence from event-related brain potentials", "authors": [ { "first": "J", "middle": [], "last": "Kounios", "suffix": "" }, { "first": "S", "middle": [ "A" ], "last": "Kotz", "suffix": "" }, { "first": "P", "middle": [ "J" ], "last": "Holcomb", "suffix": "" } ], "year": 2000, "venue": "Memory & cognition", "volume": "28", "issue": "8", "pages": "1366--1377", "other_ids": {}, "num": null, "urls": [], "raw_text": "Kounios, J., Kotz, S. A., & Holcomb, P. J. (2000). On the locus of the semantic satiation effect: Evidence from event-related brain potentials. Memory & cognition, 28(8), 1366-1377.", "links": null }, "BIBREF16": { "ref_id": "b16", "title": "A solution to Plato's problem: The latent semantic analysis theory of acquisition, induction, and representation of knowledge", "authors": [ { "first": "T", "middle": [ "K" ], "last": "Landauer", "suffix": "" }, { "first": "S", "middle": [ "T" ], "last": "Dumais", "suffix": "" } ], "year": 1997, "venue": "Psychological review", "volume": "104", "issue": "2", "pages": "", "other_ids": {}, "num": null, "urls": [], "raw_text": "Landauer, T. K., & Dumais, S. T. (1997). A solution to Plato's problem: The latent semantic analysis theory of acquisition, induction, and representation of knowledge. Psychological review, 104(2), 211.", "links": null }, "BIBREF17": { "ref_id": "b17", "title": "Handbook of latent semantic analysis", "authors": [ { "first": "T", "middle": [ "K" ], "last": "Landauer", "suffix": "" }, { "first": "D", "middle": [ "S" ], "last": "Mcnamara", "suffix": "" }, { "first": "S", "middle": [ "E" ], "last": "Dennis", "suffix": "" }, { "first": "W", "middle": [ "E" ], "last": "Kintsch", "suffix": "" } ], "year": 2007, "venue": "", "volume": "", "issue": "", "pages": "", "other_ids": {}, "num": null, "urls": [], "raw_text": "Landauer, T. K., McNamara, D. S., Dennis, S. E., & Kintsch, W. E. (2007). Handbook of latent semantic analysis: Lawrence Erlbaum Associates Publishers.", "links": null }, "BIBREF18": { "ref_id": "b18", "title": "A self-organizing neural network model of the acquisition of word meaning", "authors": [ { "first": "P", "middle": [], "last": "Li", "suffix": "" } ], "year": 2001, "venue": "", "volume": "", "issue": "", "pages": "", "other_ids": {}, "num": null, "urls": [], "raw_text": "Li, P. (2001). A self-organizing neural network model of the acquisition of word meaning.", "links": null }, "BIBREF19": { "ref_id": "b19", "title": "Lexical organization and competition in first and second languages: Computational and neural mechanisms", "authors": [ { "first": "P", "middle": [], "last": "Li", "suffix": "" } ], "year": 2009, "venue": "Cognitive science", "volume": "33", "issue": "4", "pages": "629--664", "other_ids": {}, "num": null, "urls": [], "raw_text": "Li, P. (2009). Lexical organization and competition in first and second languages: Computational and neural mechanisms. Cognitive science, 33(4), 629-664.", "links": null }, "BIBREF20": { "ref_id": "b20", "title": "Early lexical development in a self-organizing neural network", "authors": [ { "first": "P", "middle": [], "last": "Li", "suffix": "" }, { "first": "I", "middle": [], "last": "Farkas", "suffix": "" }, { "first": "B", "middle": [], "last": "Macwhinney", "suffix": "" } ], "year": 2004, "venue": "Neural Networks", "volume": "17", "issue": "8-9", "pages": "1345--1362", "other_ids": {}, "num": null, "urls": [], "raw_text": "Li, P., Farkas, I., & MacWhinney, B. (2004). Early lexical development in a self-organizing neural network. Neural Networks, 17(8-9), 1345-1362.", "links": null }, "BIBREF21": { "ref_id": "b21", "title": "A framework for representing knowledge", "authors": [ { "first": "M", "middle": [], "last": "Minsky", "suffix": "" } ], "year": 1974, "venue": "", "volume": "", "issue": "", "pages": "", "other_ids": {}, "num": null, "urls": [], "raw_text": "Minsky, M. (1974). A framework for representing knowledge. AIM-306.", "links": null }, "BIBREF22": { "ref_id": "b22", "title": "Evidence for semantic satiation: Repeating a category slows subsequent semantic processing", "authors": [ { "first": "L", "middle": [], "last": "Smith", "suffix": "" }, { "first": "R", "middle": [], "last": "Klein", "suffix": "" } ], "year": 1990, "venue": "Journal of Experimental Psychology: Learning, Memory, and Cognition", "volume": "16", "issue": "5", "pages": "", "other_ids": {}, "num": null, "urls": [], "raw_text": "Smith, L., & Klein, R. (1990). Evidence for semantic satiation: Repeating a category slows subsequent semantic processing. Journal of Experimental Psychology: Learning, Memory, and Cognition, 16(5), 852.", "links": null }, "BIBREF23": { "ref_id": "b23", "title": "Testing an associative account of semantic satiation", "authors": [ { "first": "X", "middle": [], "last": "Tian", "suffix": "" }, { "first": "D", "middle": [ "E" ], "last": "Huber", "suffix": "" } ], "year": 2010, "venue": "Cognitive psychology", "volume": "60", "issue": "4", "pages": "267--290", "other_ids": {}, "num": null, "urls": [], "raw_text": "Tian, X., & Huber, D. E. (2010). Testing an associative account of semantic satiation. Cognitive psychology, 60(4), 267-290.", "links": null }, "BIBREF24": { "ref_id": "b24", "title": "Kohonen's Self Organizing Feature Maps for Exploratory Data Analysis", "authors": [ { "first": "A", "middle": [], "last": "Ultsch", "suffix": "" }, { "first": "H", "middle": [ "P" ], "last": "Siemon", "suffix": "" } ], "year": 1990, "venue": "Proceedings of the International Neural Network Conference (INNC' 90)", "volume": "", "issue": "", "pages": "", "other_ids": {}, "num": null, "urls": [], "raw_text": "Ultsch, A., & Siemon, H. P. (1990). Kohonen's Self Organizing Feature Maps for Exploratory Data Analysis. Paper presented at the Proceedings of the International Neural Network Conference (INNC' 90).", "links": null }, "BIBREF25": { "ref_id": "b25", "title": "Modern applied statistics with S: Springer verlag", "authors": [ { "first": "W", "middle": [ "N" ], "last": "Venables", "suffix": "" }, { "first": "B", "middle": [ "D" ], "last": "Ripley", "suffix": "" } ], "year": 2002, "venue": "", "volume": "", "issue": "", "pages": "", "other_ids": {}, "num": null, "urls": [], "raw_text": "Venables, W. N., & Ripley, B. D. (2002). Modern applied statistics with S: Springer verlag.", "links": null }, "BIBREF26": { "ref_id": "b26", "title": "Self-and super-organizing maps in R: the Kohonen package", "authors": [ { "first": "R", "middle": [], "last": "Wehrens", "suffix": "" }, { "first": "L", "middle": [ "M C" ], "last": "Buydens", "suffix": "" } ], "year": 2007, "venue": "Journal of Statistical Software", "volume": "21", "issue": "5", "pages": "", "other_ids": {}, "num": null, "urls": [], "raw_text": "Wehrens, R., & Buydens, L. M. C. (2007). Self-and super-organizing maps in R: the Kohonen package. Journal of Statistical Software, 21(5), 19.", "links": null }, "BIBREF27": { "ref_id": "b27", "title": "Vocabulary development in English and Chinese: A comparative study with self-organizing neural networks", "authors": [ { "first": "X", "middle": [], "last": "Zhao", "suffix": "" }, { "first": "P", "middle": [], "last": "Li", "suffix": "" } ], "year": 2008, "venue": "", "volume": "", "issue": "", "pages": "", "other_ids": {}, "num": null, "urls": [], "raw_text": "Zhao, X., & Li, P. (2008). Vocabulary development in English and Chinese: A comparative study with self-organizing neural networks.", "links": null }, "BIBREF28": { "ref_id": "b28", "title": "Contextual self-organizing map: software for constructing semantic representations", "authors": [ { "first": "X", "middle": [], "last": "Zhao", "suffix": "" }, { "first": "P", "middle": [], "last": "Li", "suffix": "" }, { "first": "T", "middle": [], "last": "Kohonen", "suffix": "" } ], "year": 2010, "venue": "Behavior Research Methods", "volume": "", "issue": "", "pages": "1--12", "other_ids": {}, "num": null, "urls": [], "raw_text": "Zhao, X., Li, P., & Kohonen, T. (2010). Contextual self-organizing map: software for constructing semantic representations. Behavior Research Methods, 1-12.", "links": null } }, "ref_entries": { "FIGREF0": { "uris": null, "type_str": "figure", "num": null, "text": "Kohonen Maps of Chinese Morphological Families to 57 Visualize the Interplay of Morphology and Semantics in Chinese lexical units, as represented by Diagram (b). The reason is that, in a LDT, this link is changing at each presentation of a new two-character word. The morphological unit of the repeated morpheme constitutes one fixed endpoint of the morphological/lexical link but the over endpoint is always changing.The present work of semantic clustering focuses on clarifying by computational means whether morphological satiation would probably have a sole morphological locus (Diagram (a)) or whether it would have both a morphological and semantic locus(Diagram (d)). The behavioral LDT experiment results inChen et al. (2009) point to the existence of a sole morphological locus." }, "FIGREF1": { "uris": null, "type_str": "figure", "num": null, "text": "Different possible loci of satiation for morphological satiation." }, "FIGREF2": { "uris": null, "type_str": "figure", "num": null, "text": "SOM 93 of the \u8a08 (j\u00ed) morphological family. Using Kohonen Maps of Chinese Morphological Families to 63 Visualize the Interplay of Morphology and Semantics in Chinese" }, "FIGREF3": { "uris": null, "type_str": "figure", "num": null, "text": "SOM for Cluster C3 inFigure 2." }, "FIGREF4": { "uris": null, "type_str": "figure", "num": null, "text": "shows only the 13 words used by" }, "FIGREF5": { "uris": null, "type_str": "figure", "num": null, "text": "SOM of the members of the \u8a08 morphological family used inChen et al. (2009)." }, "TABREF0": { "num": null, "content": "
\u660e\u671d\u660e\u5929\u660e\u767d\u660e\u78ba\u660e\u661f\u660e\uf977
Ming Dynastytomorrowto understand / clearexplicitstarbright
", "type_str": "table", "html": null, "text": "" } } } }