The relation between linguistic categories and cognition: The case of numeral classifiers

Abstract

The classifier grammar system categorises things in the world in a way that is drastically different from the way nouns do. Previous research revealed amplified similarity among objects belonging to the same classifier category in Chinese speakers, but how this amplified classifier similarity effect arises was still an open question. The present research was conducted to address this question. For this purpose, we compared speakers of Chinese, Japanese (classifier languages), and German (nonclassifier language) on a range of cognitive tasks including similarity judgements, property induction, and fast-speed word-picture matching. Although Chinese and Japanese classifier systems are similar in their semantic structures, classifier classes for nouns are marked more systematically in Chinese than in Japanese. The amplified classifier similarity effect was found in Chinese but not in Japanese speakers. We explore the nature of the amplified classifier similarity effect and propose an explanation for how it may arise.

Keywords:

• Concepts
• Linguistic relativity
• Classifiers
• Similarity
• Inductive reasoning

Acknowledgements

This research was supported by fellowships from the Japan Society of the Promotion of Science (JSPS) and the Friedrich-Ebert-Stiftung awarded to Saalbach and a Ministry of Education grant-in-aid for Scientific Research awarded to Imai (#15300088 and # 18300089), as well as Keio Gijuku Academic Development Funds and Keio Gijuku Mori Memorial Research Fund awarded to Imai. We are deeply indebted to Zhou Xiaolin, Elsbeth Stern, Sachiko Saitoh, Lennart Schalk, Shu Hua, Zhu Liqi, and Huangfu Gang for their help with data collection and discussion. We also thank Hiroki Okada for writing the program to conduct Experiments 4 and 5, Miriam Leuchter for her advices and help concerning data analysis, and Hella Beister for their thoughtful comments on earlier versions of this manuscript.

The two authors equally contributed to the paper and the order of the authorship is arbitrary.

Notes

¹Although there are different types of classifier languages (e.g., Aikhenvald, 2000; Allan, 1977; Grinevald, 2000; Senft, 2000 , our focus in this paper is on numeral classifier systems only. Thus, the terms “classifiers” and “numeral classifiers” will be used synonymously in the present article.

²The number of classifier counts does not reflect an exhaustive count of all classifier uses. Because the Japanese-Chinese parallel corpus was not tagged, we had to conduct a manual search for cases of classifier uses. We thus started out from the Chinese translation, as we expected to find a much larger number of classifier counts in Chinese than in Japanese. It was not possible to search for all possible classifier types, as there are over 100 classifiers. We thus searched for the classifiers used in this research, as they are frequent and representative classifiers used with the numerals “one” “two” “three” and the demonstratives zhe (this) and na (that). We then examined the corresponding part of the original Japanese text to see if, in the original, a classifier was used in the same cases. Thus, classifier uses with numerals other than “one”, “two”, and “three” as well as uses with classifiers that were not included among the stimuli in this research were not included in the count.

³However, we do not always agree with Gao and Malt's classifications of the classifiers. What appears “arbitrary,” especially for non-native speakers of the language, could have covert semantic meanings, which native speakers could unconsciously detect. Thus, in our view, it is difficult to clearly distinguish their “prototype” and “arbitrary” classifiers. The three types are better thought of as being on the same continuum.