Ellipsis in comparative constructions provides a challenging test of semantic interpretation because comprehenders must reconstruct implicit constituents that are not overtly expressed. This study investigates whether large language models (LLMs) can capture the structured semantic inferences required for understanding Chinese comparative constructions. We introduce a linguistically motivated evaluation framework targeting two essential components of comparative interpretation: (i) identification of comparative arguments and (ii) recovery of elliptical constituents. Based on 2,076 naturally occurring Chinese bi-comparative sentences, we construct an annotated resource with comparative argument boundaries, recovered elliptical constituents, and five comparison dimensions representing distinct semantic bases of comparison. We evaluate several representative LLMs, including DeepSeek-V3.1, GPT-4o, and Gemini-2.0-Flash. While all models reliably identify overt comparative arguments, they exhibit substantial degradation when reconstructing implicit constituents. These results suggest that current LLMs can capture surface-level structural regularities but still face challenges in recovering implicit semantic representations required for human-like interpretation. Our resource provides a controlled linguistic testbed for studying the relationship between linguistic structure, semantic inference, and computational models of language understanding.

This paper investigates the ability of Large language Models (LLMs) in capturing linguistic patterns from unseen languages and applying them to translation between the languages and English within an in-context learning framework. Inspired by the International Linguistics Olympiad (IOL), we create test data consisting of translation puzzles between 40 low resource languages and English. We test the LLMs in two different strategies: direct prompting and step-by-step prompting. In the latter, the puzzles are manually decomposed into intermediate steps to allow LLMs learn and apply linguistic rules incrementally. The results show that this strategy can significantly improve the performance of LLMs, achieving comparable or slightly superior results to humans when translating the unseen languages to English. However, LLMs still struggle with translating English into the unseen languages, typically with complex syntactic rules. We further observe that LLMs cannot deal with languages with object-subject and noun-adjective word order compared to others, reflecting the potential impact imposed by typological features of languages in training data.

In this study, we evaluate the performance of large language models (LLMs) in detecting and correcting grammatical errors made by Chinese language learners. We find that incorporating various linguistic features—such as dependency structures, parts of speech, and pinyin transliteration—into the prompts can potentially enhance model performance. Among these features, parts of speech and pinyin prove to be the most effective across all tested models. Additionally, our findings show that the success of error correction also depends on the severity of the errors. When the intended meaning is preserved, LLMs tend to provide accurate revisions following the principle of minimal editing. However, when the meaning is obscured, LLMs are more likely to produce divergent outputs, both in comparison to reference corrections and to the responses of other models.