Language models (LMs) have been used in cognitive modeling as well as
engineering studies -- they compute information-theoretic complexity metrics
that simulate humans' cognitive load during reading. This study highlights a
limitation of modern neural LMs as the model of choice for this purpose: there
is a discrepancy between their context access capacities and that of humans.
Our results showed that constraining the LMs' context access improved their
simulation of human reading behavior. We also showed that LM-human gaps in
context access were associated with specific syntactic constructions;
incorporating syntactic biases into LMs' context access might enhance their
cognitive plausibility.

本文研究了语言模型在认知建模和工程研究中的应用，提出了现代神经 LM 的局限性，其中与人类阅读过程相比，存在上下文访问能力的差异。限制 LM 上下文访问可以提高其模拟人类阅读行为的能力，同时将句法偏好纳入 LM 的上下文访问中也可以增强其认知合理性。

上下文限制使神经语言模型更接近人类

Context Limitations Make Neural Language Models More Human-Like

Abstract Meaning Representation (AMR) is an annotation framework representing
the semantic structure of a sentence as a whole. From the beginning, AMR was
not intended to act as an interlingua; however, it has made progress towards
the idea of designing a universal meaning representation framework.
Accordingly, developing AMR annotation guidelines for different languages,
based on language divergences, is of significant importance. In this paper, we
elaborate on Persian Abstract Meaning Representation (PAMR) annotation
specifications, based on which we annotated the Persian translation of "The
Little Prince" as the first gold standard for Persian AMR. Moreover, we
describe how some Persian-specific syntactic constructions would result in
different AMR annotations.

本文介绍了波斯语摘要含义表示（PAMR）注释规范，基于该规范我们对《小王子》的波斯语翻译进行了注释，这是波斯语 AMR 的首个黄金标准，并描述了一些波斯语特有的句法结构将导致不同的 AMR 注释。