What has been considered to be a key limitation of connectionist models of language acquisition?

Question

BlackTom AI · Accepted Answer

First, consider what each option says about connectionist models of language acquisition and how these claims align with common critiques in the literature.
Option 1: 'They are difficult to understand.' This is a general meta-claim about interpretability. While some researchers have noted that distributed representations can be opaque, calling the models simply 'difficult to understand' is vague and does not pinpoint a core theoretical limitation that researchers emphasize about language learning.
Option 2: 'They cannot learn irregular past tense inflections of verbs.' Historically, some early connectionist simulations struggled with irregular forms (like went vs went), which led to the belief that such patterns require explicit rule-like representations. However, many later models demonstrated that irregulars could be learned through distributed representations and lexical memorization, reducing the universality of this limitation. So as a blanket statement, this is not an enduring, universally agreed limitation of connectionist approaches.
Option 3: 'They cannot account for the fact that children get better, get worse, and then get better at irregular past tense inflections of verbs.' This option refers to a problematic pattern that some early models failed to reproduce; however, the dynamic trajectory described by children—improving, sometimes regressing, then improving again—has been argued to be challenging, but connectionist models have evolved to capture complex learning trajectories with experience-dependent adjustments. It is not universally accepted as a definitive limitation across all connectionist work.
Option 4: 'They only learn stimulus response associations which are insufficient for language learning.' This captures a core, widely discussed critique: purely associative learning mechanisms may not fully account for the structured, rule-governed aspects of language and generalization beyond the training data. It emphasizes that language often requires more than simple co-occurrence statistics to acquire abstract structure and productivity, a position advocated by critics who argue for the need to incorporate inductive biases or constraints beyond simple stimulus–response mappings.
In summary, while individual options touch on aspects of drawbacks or debates surrounding connectionist models, the statement that best highlights a commonly argued foundational limitation is that these models 'only learn stimulus–response associations which are insufficient for language learning.' This frames the broader debate about whether language requires more than associative learning to capture abstract generalization and systematicity.

PSYC20007_2025_SM2 Week 11: Language

What has been considered to be a key limitation of connectionist models of language acquisition?

查看解析

登录即可查看完整答案

类似问题

What has been considered to be a key limitation of connectionist models of language acquisition?

The key difference between the interactive activation model of reading and the TRACE model of speech perception is that only the former contains lateral inhibition at the word level.

Which of the following is a connectionist model proposing that concepts are represented by activity that is spread across a network?

What has been considered to be a key limitation of connectionist models of language acquisition?

The key difference between the interactive activation model of reading and the TRACE model of speech perception is that only the former contains lateral inhibition at the word level.

In a consumer society, many adults channel creativity into buying things

Economic stress and unpredictable times have resulted in a booming industry for self-help products

People born without creativity never can develop it

更多留学生实用工具

考试浏览器助手

风格化写作助手

论文查重助手

文献引用助手

课堂转译助手

课堂笔记助手

Quiz搜索助手

学校历年真题

智能刷题助手

智能匹配练习

希望你的学习变得更简单