Table of Contents

1. About Small World of Words project (SWOW) & SWOW-ZH
2. Instructions to the repository
   1. Obtaining the data
   2. Raw data
   3. Preprocessing scripts
   4. Processing scripts
      1. Associative frequencies and graphs
      2. Derived statistics
      3. Centralities and similarities
   5. Collection, preprocessing and processing of SWOW-GPT
   6. Applicability in other SWOW lexicons
3. Download the dataset
4. Publications based on SWOW

About Small World of Words project (SWOW) & SWOW-ZH

The small world of words project is a large-scale scientific study that aims to build a mental dictionary or lexicon in the major languages of the world and make this information widely available ¹.

In contrast to a thesaurus or dictionary, we use word associations to learn about what words mean and which ones are central in the human mind. This enables psychologists, linguists, neuro-scientists and others to test new theories about how we represent and process language. This knowledge could also be applied in a variety of ways, from learning about the difference between cultures, to learning (or forgetting) new words in a first or a second language.

SWOW-ZH is a daughter project of SWOW to map mental lexicon in Chinese, as the suffix ZH stands for Zhongwen (中文, Chinese). It was initiated to provide a comprehensive framework to measure the mental lexicon with regard to the Chinese culture and people, and the bases for comparative studies between Chinese and other languages.

The participant task we used is called multiple response association ². The methodology is based on a continued word association task, in which participants see a cue word and are asked to give three associated responses to this cue word. As the number of participants increases, the lexicon becomes comprehensive and efficient in representing mental lexicon. Therefore, it focuses on the aspects of word meaning that are shared between people without imposing restrictions on what aspects of meaning should be considered.

Chinese is a demographically and culturally complex language, whose dialects and writing systems are difficult to exhaust. In the SWOW-ZH project, we primarily focused on Mandarin Chinese (普通话, Putonghua) and simplified Chinese writing system, which are used in most regions of the Chinese mainland. Additionally, the native dialect of the participants was collected as a complementary information. Alternatively, another SWOW daughter project focusing on Cantonese, SWOW-HK, might be of your interest.

• The study was conducted in Professor CAI Qing's lab at the School of Psychology and Cognitive Science, East China Normal University (华东师范大学心理与认知科学学院，蔡清教授团队), in collaboration with Dr. Simon de Deyne at Melbourne University, who founded the SWOW project when he was under the supervision by Professor Gert Storms at University of Leuven.

• Please address questions and suggestions to:

  • DING Ziyi | 丁子益 | ziyi.ecnu@gmail.com | ZiyiDing7@github
  • LI Bing | 李兵 | lbing314@gmail.com | lib314a@github

• Affiliations:

  • Shanghai Key Laboratory of Brain Functional Genomics (Ministry of Education), Affiliated Mental Health Center (ECNU), Institute of Brain and Education Innovation, School of Psychology and Cognitive Science, East China Normal University, Shanghai, China
  • Shanghai Center for Brain Science and Brain-Inspired Technology, Shanghai, China

• Thanks:

  • This work was supported by the National Natural Science Foundation of China (grant numbers 31970987 to Qing Cai) and the Australian Research Council Early Career Grant (DE140101749 to Simon De Deyne)

• License of the data: See https://smallworldofwords.org/en/project/

• License of the code:
  This work is licensed under a Creative Commons Attribution 4.0 International License.

• Cite us:

  • APA: Li, B., Ding, Z., De Deyne, S., & Cai, Q. (2024). A large-scale database of Mandarin Chinese word associations from the Small World of Words Project. Behavior Research Methods, 57(1), 34. http://dx.doi.org/10.3758/s13428-024-02513-1
  • bibtex:

    @article{li_large-scale_2024,
  title = {A large-scale database of {Mandarin} {Chinese} word associations from the {Small} {World} of {Words} {Project}},
  volume = {57},
  issn = {1554-3528},
  url = {https://link.springer.com/10.3758/s13428-024-02513-1},
  doi = {10.3758/s13428-024-02513-1},
  language = {en},
  number = {1},
  urldate = {2025-01-02},
  journal = {Behavior Research Methods},
  author = {Li, Bing and Ding, Ziyi and De Deyne, Simon and Cai, Qing},
  month = dec,
  year = {2024},
  pages = {34},
  }

Instructions to the repository

Prompt: Instead of exploring the repo in your browser, cloning it onto your local machine may be more convenient.

In this repository you will find a basic analysis pipeline for the Chinese SWOW project which allows you to import a preprocessing the data as well as compute some basic statistics.

Obtaining the data

In addition to the scripts, you will need to retrieve the word association data. Currently word association and participant data is available for 10,192 cues. The data consists of over 2 million responses collected between 2016 and 2023. They are currently submitted for publication. Note that the final version is subject to change. If you want to use these data for your own research, you can obtain them from the Small World of Words research page (https://smallworldofwords.org/zh/project/research).

To start the pipeline, SWOW-ZH_raw.(csv|mat) should be put into the data folder.

While the majority of the data was collected on the SWOW platform (ZH), a subset was collected on another China-based surveying platform NAODAO (脑岛) using the same tasks with the same inclusion standards. This presumably won't detriment the reliability of the data.

If you find any of this useful, please consider sharing the word association study (https://smallworldofwords.org/zh/project).

Raw data

Since this is an ongoing project, data is regularly updated. Hence, all datafiles refer to a release date in its filename.

1. SequenceNumber: A system coding, ascending from 1 to the end.

2. TrialsID: Unique identifiers for trials. Each trial is made up of one cue and three responses.

3. ParticipantID: Unique identifiers for the participants.

4. Created_at: Time and date when trials were finished.

5. Age: Age reported by participants.

6. NativeLanguage: Chinese dialects and Mandarin reported by participants.

   • Tags in the NAODAO platform:
     • PUTON: Putonghua or Mandarin, which is the standards of pronunciation populated officially (普通话);
     • SOUTHE: Southeast dialects, which represents northern and southern Fujian dialects, covering most of Fujian, Chaoshan, Hainan and Taiwan (东南部方言：代表为包括闽北及闽南方言，覆盖福建大部及潮汕、海南及台湾);
     • NORTH: Northern dialects representing the three northeastern provinces and the Inner Mongolian dialect, Hebei-Yulu, Jiaodong, Liaodong and the northern part of the Hanshui River Basin (北方方言：代表为东北三省及内蒙方言、冀豫鲁、胶东、辽东和汉水流域北部);
     • SOUTH: Southern dialects representing Cantonese in Guangxi, Guangdong, Hainan, Hong Kong and Macau (南部方言：代表为包括广西、广东和海南的平话、白话，及香港和澳门的粤语);
     • JIANG: Jianghuai dialects, which represents Jianghuai River Basin, Subei and Lunan (江淮方言：代表为江淮流域及苏北、鲁南);
     • SHAN: Shan-Shaan dialects from Shaanxi and Shanxi (陕、晋方言：代表为陕西及山西各地);
     • HAKKA: Hakka languages scattered all over China (客家话：代表为分布在各地的客家族语);
     • SOUTHW: Southwestern dialects from most of Yunguichuan, Hubei, and Hunan (西南方言：代表为云贵川鄂湘大部);
     • WU: Wu dialects from Jiangxi and eastern Anhui, most of Zhejiang and Shanghai (吴方言：代表为江西和安徽东部、浙江大部及上海);
     • NORTHW: Northwestern dialects from Yinchuan, Lanzhou, and Xining (西北方言：代表为银川、兰州、西宁).
   • Tags in the SWOW platform:
     • PUTON: Which is the same as NAODAO;
     • EASTW: Which is the same as WU on NAODAO;
     • JIANG: Which is the same as NAODAO;
     • SHAN: Which is the same as NAODAO;
     • HAKKA: Which is the same as NAODAO.
     • NORTH: Which combines NORTH and NORTHW in NAODAO;
     • SOUTH: Which combines SOUTHE, SOUTHW, and SOUTH in NAODAO;

7. Gender: Gender of the participants (Female / Male / X), including female, male and non-binary.

8. Education: Level of education participants selected from: 1 = None, 2 = Elementary school, 3 = High School, 4 = College or University Bachelor, 5 = College or University Master.

9. City: City location when tested, might be an approximation.

10. Country: Country location when tested.

11. Section: Identifiers for the data resources and the snowball iterations: set1-10 = Ten sets collected in the SWOW platform, NAODAO = One set collected in the NAODAO platform (https://smallworldofwords.org/zh/project/research)).

12. Cue: Cue word.

13. R1Raw: Raw primary associative response.

14. R2Raw: Raw secondary associative response.

15. R3Raw: Raw tertiary associative response.

16. R1: Primary associative response.

17. R2: Secondary associative response.

18. R3: Tertiary associative response.

Preprocessing scripts

To avoid potential errors when reading Chinese strings in MATLAB, we recommend loading and saving all data in mat format. We also provide data in csv format for users of other programming languages. Although the preprocessing scripts were primarily written in MATLAB, for the convenience of non-MATLAB users, we provided plain-text dictionaries in the data/dictionaries folder and R scripts in the scripts folder.

The preprocessing scripts consist of wordCleaning.(m|R), participantCleaning.(m|R) and dataBalancing.(m|R) scripts.

wordCleaning.(m|R): Problematic cue words and responses are marked or modified according to the dictionaries. The dictionaries could be found in the data/dictionaries folder and they were editable. The input of the script SWOW-ZH_raw.mat should be put in the data folder.

The dictionaries

1. tradCues.(txt|mat) and tradCues.(txt|mat): Traditional Chinese cues and responses were transformed into simplified equivalents based on Open Chinese Convert library, ropencc package was access from (https://github.com/Lchiffon/ropencc).

2. englishRes.(txt|mat): English responses were commonly used in Mandarin, their capitalization has been corrected.

3. unsplitedRes.(txt|mat): Joined responses, whose participants typed two or more responses into a single response box with separators like punctuation or symbols (e.g., comma, space, plus sign, and pause sign), were separated successively to individual responses, and only the first three were processed.

4. longRes.(txt|mat): Long responses (length is over six) were marked with #Long, except for meaningful long words that had appeared at least twice. A meaningless long response was defined as a character string that needs adding or deleting at least one character to become a phrase.

5. symbolRes.(txt|mat): Non-Chinese characters, which contained non-Chinese characters (letters, symbols, numbers, and/or punctuations), were modified and kept if meaningful and appeared more than once. Other kinds of such responses were marked as #Symbol.

6. erRes.(txt|mat): Retroflex final (erhua or erization), which is a pronunciation feature that modifies the final sound of certain syllables, usually in the form -er (儿), was deleted from the responses.

7. SWOWZHwordlist.mat: A Chinese word list merged from SUBTLEX-CH³ and Unigram subset of Chinese Web 5-gram Verson 1⁴. Responses excluded in the word list were considered as non-word responses in the participants cleaning stage.

participantCleaning.(m|R): Problematic participants are deleted.

dataBalancing.(m|R): Remain 55 participants for each cue words. The output of the script is written to data/SWOW-ZH_R55.mat. The participants were selected to favor participants with less missing responses and Mandarin speakers. The preprocessed data could be found in the Small World of Words research page (https://smallworldofwords.org/zh/project/research).

Notes on processing traditional characters

The conversion from traditional to simplified words were applied using the OpenCC library (see Open Chinese Convert 開放中文轉換). R users will find an OpenCC port for R called (ropencc) and test its text conversion like the following example:

CONVERTER <- ropencc::converter(ropencc::T2S)
CONVERTER["詞彙"]
[1] "词汇"

Processing scripts

The processing scripts consist of networkGeneration.m, frequencyCalculating.m, centralityCalculating.m and similarityCalculating.m scripts in MATLAB. The equivalent scripts in R were adapted from SWOW-EN: createSWOWGraph.R, createAssoStrengthTable.R, createResponseStats.R, createCueStats.R and createNetworkStatistics.R.

Additionally, gradientValidation.m valid sample size to achieve a fine prediction to relatedness judgment tasks⁵.

Associative frequencies and graphs

• networkGeneration.m: The preprocessed data is used to derive the associative frequencies (i.e., the conditional probability of a response given a cue) and saved in the output folder named as assocFrequency_R1 or _R123, where the first column contains cue words, the second column contain responses, the third column contains associative frequencies between them. Use associative frequencies to extract the largest strongly connected component for graphs based on the first response (R1) or all responses (R123). The graphs are written to data/ SWOW-ZH_network.mat. And the adjacency matrices are written to output folder named as adjacencyMatrix_R1 or _R123 and consist of directed weighted matrices, where each row labeled by N cue words and each column labeled by N responses. Then, the N×N matrices are filled by normalized associative strengths. In most cases, associative frequencies will need to be converted to associative strengths by dividing with the sum of all strengths for a particular cue. Vertices that are not part of the largest connected component are listed in a report in the output folder named as lostNodes_R1 or _R123.
• createSWOWGraph.R and createAssoStrengthTable.R had the same functions as networkGeneration.m.

Derived statistics

• frequencyCalculating.m: The script is used to describe the characteristics of responses, cue words and participants.

1. Response statistics

Currently the script calculates the number of cue words where a response is reported as their types, tokens and hapax legomena responses (responses that only occur once). The results can be found in the output folder named as resStats.

2. Cue statistics

Only words that are part of the strongly connected component are considered. Results are provided for the R1 graph and the graph with all responses (R123). The results can be found in the output folder named as cueStats_R1 or _R123. The file includes the following:

• Coverage: How many of the responses are retained in the graph after removing those words that aren't a cue or aren't part of the strongest largest component.

• Unknown: The number of unknown responses

• R1missing: The number of missing R1 responses

• R2missing: The number of missing R2 responses

• R3missing: The number of missing R3 responses

A histogram of the response coverage for R1 and R123 graphs can be obtained from the frequencyCalculating.m script. Vocabulary growth curves can be obtained with scripts/as-vocabulary-growth.R

• createResponseStats.R and createCueStats.R had the same functions as frequencyCalculating.m.

Centralities and similarities

• centralityCalculating.m: Based on the largest strongly connected component for graphs, the script calculates centrality-related indicators including: types and tokens, in-degree, out-degree, PageRank, clustering coefficient, centrality and betweenness. The scrip inserts some functions from the Brain Connectivity Toolbox (BCT) (http://www.brain-connectivity-toolbox.net). The output is written in the output folder named as centrality_R1 or _R123.

• similarityCalculating.m: Based on the largest strongly connected component for graphs, the script calculates four kinds similarity including: cosine similarity only (AssocStrength), positive pointwise mutual information (PPMI), random walk (RW) and word embedding after random walk (RW-embedding). The script is adapted from SWOW-EN and SWOW-RP. The output is written in the output folder named as similarity_R1 or _R123.

• createNetworkStatistics.R had the same functions as centralityCalculating.m.

Validation of sample size

gradientValidation: Based on the raw data after participant cleaning (SWOW-ZH_partcleaning.m), to finish the validation, behavior data from relatedness judgment tasks should be put into the data folder. The sample size is expanded from 20 to 80 participants per cue word for concrete words, and 20 to 120 participants per cue word for abstract words.

Collection, preprocessing and processing of SWOW-GPT

GPT-3.5-turbo was used through the OpenAI API to conduct the three-response free association task, which we refer to as SWOW-GPT. To ensure comparability with human-generated data, identical preprocessing and processing steps have been implemented.

In the SWOW-GPT folder:

• The script WS_gpt.html contains instructions and parameters applied to GPT-3.5-turbo for generating associations;
• The preprocessing and processing scripts, in MATLAB, are organized to mirror the structure of the prep scripts for SWOW-ZH.

SWOW-GPT raw data

There are four columns in SWOW-GPT_raw.(mat|csv): Cue, R1Raw, R2Raw, and R3Raw, representing cue words sent to GPT-3.5-turbo, and three responses for each cue word answered by GPT-3.5-turbo.

SWOW-GPT preprocessing scripts

The preprocessing scripts consist of wordCleaning.m and dataBalancing.m scripts.

wordCleaning.m: Problematic cue words and responses are marked or modified according to the dictionaries. The dictionaries could be found in the SWOW-GPT/data/dictionaries folder and they were editable. The input of the script, SWOW-GPT_raw.mat, should be put in the SWOW-GPT/data folder.

dataBalancing.m: Embed associations from GPT-3.5-turbo into human SWOW-ZH, and then remain 55 participants for each cue words. The inputs of the script include SWOW-ZH_partcleaning.mat in the SWOW-GPT/data folder. The output of the script is written to data/SWOW-GPT_R55.mat. The participants were selected to favor trials with less missing responses.

SWOW-GPT processing scripts

The processing scripts consist of networkGeneration.m, similarityCalculating.m and gradientValidation.m.

Applicability in other SWOW lexicons

Since other SWOWs are mainly processed by R scripts, a MATLAB scrip is provided thus other SWOWs could be processed by MATLAB. The SWOWs.m is used to count associative frequencies and generate graphs, and calculate in-degrees of other SWOWs. The inputs of the script are preprocessed data of other SWOWs put in the data/SWOWs folder. The outputs of the script are the graphs written to data/SWOWs/SWOW-XX_network.mat. While the XX could be substituted by EN (American English), NL (Dutch) and RP (Rioplatense Spanish). The outputs could be loaded as inputs into centralityCalculating.m and similarityCalculating.m

Download the dataset

Access the dataset from the webpage: https://smallworldofwords.org/zh/project/research

• Current: 4 November 2024

Publications based on SWOW

Following is an exhaustive list of the publications based on or used part of the lexicons:

• Journal articles
  • Cox, C. R., & Haebig, E. (2023). Child-oriented word associations improve models of early word learning. Behavior Research Methods, 55(1), 16–37. https://doi.org/10.1037/0033-295X.82.6.407
  • De Deyne, S., Navarro, D. J., Collell, G., & Perfors, A. (2021). Visual and affective multimodal models of word meaning in language and mind. Cognitive Science, 45(1), 12922. https://doi.org/10.1111/cogs.12922
  • De Deyne, S., Navarro, D. J., Perfors, A., & Storms, G. (2016). Structure at every scale: A semantic network account of the similarities between unrelated concepts. Journal of Experimental Psychology: General, 145(9), 1228-1254. http://dx.doi.org/10.1037/xge0000192
  • Jana, A., Haldar, S., & Goyal, P. (2022). Network embeddings from distributional thesauri for improving static word representations. Expert Systems with Applications, 187, e115868. https://doi.org/10.1016/j.eswa.2021.115868
  • Johnson, D. R., & Hass, R. W. (2022). Semantic context search in creative idea generation. The Journal of Creative Behavior, 56(3), 362-381. https://doi.org/10.1002/jocb.534
  • Kumar, A. A., Balota, D. A., & Steyvers, M. (2020). Distant connectivity and multiple-step priming in large-scale semantic networks. Journal of Experimental Psychology: Learning, Memory, and Cognition, 46(12), 2261-2276. https://doi.org/10.1037/xlm0000793
  • Kumar, A. A., Steyvers, M., & Balota, D. A. (2021). Semantic memory search and retrieval in a novel cooperative word game: a comparison of associative and distributional semantic models. Cognitive Science, 45(10), e13053. https://doi.org/10.1111/cogs.13053
  • Maxwell, N. P., & Buchanan, E. M. (2020). Investigating the interaction of direct and indirect relation on memory judgments and retrieval. Cognitive Processing, 21(1), 41-53. https://doi.org/10.1007/s10339-019-00935-w
  • Meersmans, K., Bruffaerts, R., Jamoulle, T., Liuzzi, A. G., De Deyne, S., Storms, G., Dupont, P., & Vandenberghe, R. (2020). Representation of associative and affective semantic similarity of abstract words in the lateral temporal perisylvian language regions. NeuroImage, 217, 116892. https://doi.org/10.1016/j.neuroimage.2020.116892
  • Meersmans, K., Storms, G., De Deyne, S., Bruffaerts, R., Dupont, P., & Vandenberghe, R. (2022). Orienting to different dimensions of word meaning alters the representation of word meaning in early processing regions. Cerebral Cortex, 32(15), 3302-3317.
  • Melvie, T., Taikh, A., Gagn\'e, Christina L, & Spalding, T. L. (2022). Constituent processing in compound and pseudocompound words. Canadian Journal of Experimental Psychology/Revue canadienne de psychologie expérimentale, 77(2), 98–114. https://doi.org/10.1037/cep0000287
  • Richie, R., & Bhatia, S. (2021). Similarity judgment within and across categories: a comprehensive model comparison. Cognitive Science, 45(8), e13030. https://doi.org/10.1111/cogs.13030
  • Sarkar, S., Bhagwat, A., & Mukherjee, A. (2022). A core-periphery structure-based network embedding approach. Social Network Analysis and Mining, 12, 32. https://doi.org/10.1007/s13278-021-00749-9
  • Valba, O., & Gorsky, A. (2022). K-clique percolation in free association networks and the possible mechanism behind the 7±2 law. Scientific Reports, 12, 5540. https://doi.org/10.1038/s41598-022-09499-w
  • Valba, O., Gorsky, A., Nechaev, S., & Tamm, M. (2021). Analysis of english free association network reveals mechanisms of efficient solution of remote association tests. PLOS ONE, 16(4), e248986. https://doi.org/10.1371/journal.pone.0248986
  • Vankrunkelsven, H., Vankelecom, L., Storms, G., De Deyne, S., & Voorspoels, W. (2021). Guessing Words. In G. Kristiansen, K. Franco, S. De Pascale, L. Rosseel, & W. Zhang (Eds.), Cognitive Sociolinguistics Revisited (pp. 572–583). : De Gruyter Mouton.
  • Verheyen, S., De Deyne, S., Linsen, S., & Storms, G. (2020). Lexicosemantic, affective, and distributional norms for 1,000 dutch adjectives. Behavior Research Methods, 52(3), 1108–1121. https://doi.org/10.3758/s13428-019-01303-4
  • Wong, T. Y., Fang, Z., Yu, Y. T., Cheung, C., Hui, C. L., Elvevåg, B., ... & Chen, E. Y.(2022). Discovering the structure and organization of a free cantonese emotion-label word association graph to understand mental lexicons of emotions. Scientific Reports, 12, 19581. https://doi.org/10.1038/s41598-022-23995-z
  • Wulff, D. U., De Deyne, S., Aeschbach, S., & Mata, R. (2022). Using network science to understand the aging lexicon: linking individuals' experience, semantic networks, and cognitive performance. Topics in Cognitive Science, 14(1), 93-110. https://doi.org/10.1111/tops.12586
  • Wulff, D. U., & Mata, R. (2022). On the semantic representation of risk. Science Advances, 8(27), eabm1883. https://doi.org/10.1126/sciadv.abm1883
  • Yang, Y., Li, L., de Deyne, S., Li, B., Wang, J., & Cai, Q. (2024). Unraveling lexical semantics in the brain: Comparing internal, external, and hybrid language models. Human Brain Mapping, 45(1), e26546. https://doi.org/10.1002/hbm.26546
• Proceedings, pre-prints etc
  • Ashok Kumar, A., Garg, K., & Hawkins, R. (2021). Contextual flexibility guides communication in a cooperative language game. In Proceedings of the 43rd Annual Meeting of the Cognitive Science Society. https://escholarship.org/uc/item/92m138t3
  • Berger, U., Stanovsky, G., Abend, O., & Frermann, L. (2022). A computational acquisition model for multimodal word categorization. arXiv. https://arxiv.org/abs/2205.05974
  • Branco, A., Rodrigues, J., Salawa, M., Branco, R., & Saedi, C. (2020). Comparative probing of lexical semantics theories for cognitive plausibility and technological usefulness. arXiv. http://arxiv.org/abs/2011.07997
  • Du, Y., Wu, Y., & Lan, M. (2019). Exploring human gender stereotypes with word association test. In Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conference on Natural Language Processing (EMNLP-IJCNLP) (pp. 6133-6143).
  • Han, Z., & Truex, R. (2020). Measuring political attitudes with word association. (SSRN Scholarly Paper 3701860). https://doi.org/10.2139/ssrn.3701860
  • Kovacs, C. J., Wilson, J. M., & Kumar, A. A. (2022). Fast and frugal memory search for communication. In Proceedings of the Annual Meeting of the 44th Cognitive Science Society. https://escholarship.org/uc/item/3301p4cj
  • Liu, C., Cohn, T., De Deyne, S., & Frermann, L. (2022). Wax: A new dataset for word association explanations. In Proceedings of the 2nd Conference of the Asia-Pacific Chapter of the Association for Computational Linguistics and the 12th International Joint Conference on Natural Language Processing (Volume 1: Long Papers) (pp. 106-120).
  • Liu, C., Cohn, T., & Frermann, L. (2021). Commonsense knowledge in word associations and ConceptNet. arXiv. https://doi.org/10.48550/arXiv.2109.09309
  • Nedergaard, J., Smith, K., & Smith, K. (2020). Are you thinking what I'm thinking? Perspective-taking in a language game. In S. Denison, M. Mack, Y. Xu, & B. C. Armstrong (Eds.), In Developing a Mind: Learning in Humans, Animals, and Machines: Proceedings for the 42nd Annual Meeting of the Cognitive Science Society (pp. 1001-1007). Cognitive Science Society. https://cognitivesciencesociety.org/wp-content/uploads/2020/07/cogsci20_proceedings_final.pdf
  • Nighojkar, A., Khlyzova, A., & Licato, J. (2022). Cognitive modeling of semantic fluency using transformers. arXiv. http://arxiv.org/abs/2208.09719
  • Petridis, S., Shin, H. V., & Chilton, L. B (2021). Symbolfinder: Brainstorming diverse symbols using local semantic networks. In The 34th Annual ACM Symposium on User Interface Software and Technology (pp. 385-399). https://doi.org/10.1145/3472749.3474757
  • Rodrigues, J., Branco, R., & Branco, A. (2022). Transfer learning of lexical semantic families for argumentative discourse units identification. arXiv. https://doi.org/10.48550/arXiv.2209.02495
  • Rotaru, A. S. (2020). Computational explorations of semantic cognition [Doctoral dissertation, University College London]. https://discovery.ucl.ac.uk/id/eprint/10106344/
  • Salawa, M. (2019). Word embeddings from lexical ontologies: A comparative study [Master's thesis]. http://apohllo.pl/text/mgr/salawa-embeddingi.pdf
  • Sarkar, S., Bhagwat, A., & Mukherjee, A. (2018). Core2vec: A core-preserving feature learning framework for networks. In 2018 IEEE/ACM International Conference on Advances in Social Networks Analysis and Mining (ASONAM) (pp. 487–490). https://doi.org/10.1109/ASONAM.2018.8508693
  • Siow, S., & Plunkett, K. (2021). Exploring the variable effects of frequency and semantic diversity as predictors for a word's ease of acquisition in different word classes. In Proceedings of the 43rd Annual Meeting of the Cognitive Science Society. https://escholarship.org/uc/item/83t6n1rq
  • Thawani, A., Srivastava, B., & Singh, A. (2019).SWOW-8500: word association task for intrinsic evaluation of word embeddings. In A. Rogers, A. Drozd, A. Rumshisky, & Y. Goldberg (Eds.), In Proceedings of the 3rd Workshop on Evaluating Vector Space Representations for NLP (pp. 43–51). Association for Computational Linguistics. https://doi.org/10.18653/v1/W19-2006
  • van Paridon, J., Liu, Q., & Lupyan, G. (2021). How do blind people know that blue is cold? distributional semantics encode color-adjective associations. In Proceedings of the Annual Meeting of the 43rd Cognitive Science Society. https://escholarship.org/uc/item/6sq7h506
  • Wulff, D. U., Aeschbach, S., De Deyne, S., & Mata, R. (2022). Data from the MySWOW proof-of-concept study: linking individual semantic networks and cognitive performance. Journal of Open Psychology Data, 10(1), 1-8. https://doi.org/10.5334/jopd.55
  • Yang, W., & Ma, X. (2022). Building knowledge graphs of experientially related concepts. In Proceedings of the 4th Conference on Automated Knowledge Base Construction (AKBC 2022). https://akbc.ws/2022/papers/13_building_knowledge_graphs_of_e

Footnotes

¹ De Deyne, S., Navarro, D. J., & Storms, G. (2013). Better explanations of lexical and semantic cognition using networks derived from continued rather than single-word associations. Behavior Research Methods, 45(2), 480-498. http://dx.doi.org/10.3758/s13428-012-0260-7

² Nelson, D. L., McEvoy, C. L., & Dennis, S. (2000). What is free association and what does it measure? Memory & Cognition, 28(6), 887-899. https://doi.org/10.3758/BF03209337

³ Cai, Q., & Brysbaert, M. (2010). SUBTLEX-CH: Chinese word and character frequencies based on film subtitles. PLOS ONE, 5(6), e10729. https://doi.org/10.1371/journal.pone.0010729

⁴ Liu, F., Yang, M., & Lin, D. (2010). Chinese web 5-gram version 1 [dataset]. Linguistic Data Consortium. https://doi.org/10.35111/647p-yt29

⁵ De Deyne, S., Cabana, Á., Li, B., Cai, Q., & McKague, M. (2020). A cross-linguistic study into the contribution of affective connotation in the lexico-semantic representation of concrete and abstract concepts. In Proceedings of the 42nd Annual Meeting of the Cognitive Science Society: Developing a Mind: Learning in Humans, Animals, and Machines (pp. 2776–2782). Cognitive Science Society.