Alzahrani Alaa, Aljuaythin Wafa, Alshumrani Hassan, Saleh Alaa Mamoun, Mostafa Mohamed M
King Salman Global Academy for Arabic, Riyadh, Saudi Arabia.
Department of English, King Saud University, Riyadh, Saudi Arabia.
Behav Res Methods. 2025 Jun 9;57(7):194. doi: 10.3758/s13428-025-02692-5.
Crowdsourced normative ratings have benefited psycholinguistic research considerably. Yet, Modern Standard Arabic (MSA) has scarce word norms. To address this scarcity, the current study developed and validated the Kalimah norms in two experiments. In experiment 1, 803 native Arabic speakers provided age of acquisition (AOA) and concreteness (CNC) ratings for 2,467 MSA words, a subset of which was lexically ambiguous (N = 60). Correlations with 12 Arabic and international norms established the sufficient validity of the Kalimah norms. We also observed variations in the ratings for the distinct meanings of the same word form. In experiment 2, we show that our crowdsourced norms can be used to validate large language model (LLM)-generated norm ratings for MSA words. Based on this, we obtained LLM-generated CNC ratings for an additional set of 30,000 MSA words. We make both the Kalimah norms and LLM-generated CNC ratings freely available for research purposes.
众包规范评级极大地促进了心理语言学研究。然而，现代标准阿拉伯语（MSA）的词汇规范却很稀缺。为了解决这一稀缺问题，本研究通过两个实验开发并验证了卡里玛规范。在实验1中，803名阿拉伯语母语者对2467个现代标准阿拉伯语单词给出了习得年龄（AOA）和具体性（CNC）评级，其中一部分单词存在词汇歧义（N = 60）。与12种阿拉伯语及国际规范的相关性确立了卡里玛规范的充分有效性。我们还观察到同一词形不同含义的评级存在差异。在实验2中，我们表明我们的众包规范可用于验证大语言模型（LLM）生成的现代标准阿拉伯语单词规范评级。基于此，我们获得了另外30000个现代标准阿拉伯语单词的大语言模型生成的具体性评级。我们将卡里玛规范和大语言模型生成的具体性评级都免费提供用于研究目的。
