Department of Speech and Hearing Sciences, University of Washington, Seattle, WA 98195, USA; Institute for Learning & Brain Sciences, University of Washington, Seattle, WA 98195, USA.
J Neurosci Methods. 2024 Dec;412:110277. doi: 10.1016/j.jneumeth.2024.110277. Epub 2024 Sep 6.
Speech sounds are processed in the human brain through intricate and interconnected cortical and subcortical structures. Two neural signatures, one largely from cortical sources (mismatch response, MMR) and one largely from subcortical sources (frequency-following response, FFR) are critical for assessing speech processing as they both show sensitivity to high-level linguistic information. However, there are distinct prerequisites for recording MMR and FFR, making them difficult to acquire simultaneously NEW METHOD: Using a new paradigm, our study aims to concurrently capture both signals and test them against the following criteria: (1) replicating the effect that the MMR to a native speech contrast significantly differs from the MMR to a nonnative speech contrast, and (2) demonstrating that FFRs to three speech sounds can be reliably differentiated.
Using EEG from 18 adults, we observed a decoding accuracy of 72.2 % between the MMR to native vs. nonnative speech contrasts. A significantly larger native MMR was shown in the expected time window. Similarly, a significant decoding accuracy of 79.6 % was found for FFR. A high stimulus-to-response cross-correlation with a 9 ms lag suggested that FFR closely tracks speech sounds.
COMPARISON WITH EXISTING METHOD(S): These findings demonstrate that our paradigm reliably captures both MMR and FFR concurrently, replicating and extending past research with much fewer trials (MMR: 50 trials; FFR: 200 trials) and shorter experiment time (12 minutes).
This study paves the way to understanding cortical-subcortical interactions for speech and language processing, with the ultimate goal of developing an assessment tool specific to early development.
人类大脑通过复杂且相互连接的皮质和皮质下结构来处理语音。两种神经特征对于评估语音处理至关重要，一种主要来自皮质源（失配响应，MMR），另一种主要来自皮质下源（频率跟随响应，FFR），因为它们都对高级语言信息敏感。然而，记录 MMR 和 FFR 有明显的前提条件，这使得它们难以同时获取。
使用新的范式，我们的研究旨在同时获取这两种信号，并根据以下标准对其进行测试：（1）复制 MMR 对母语语音对比的效果明显不同于 MMR 对非母语语音对比的效果，以及（2）证明可以可靠地区分三个语音的 FFR。
使用 18 位成年人的 EEG，我们观察到母语与非母语语音对比的 MMR 之间的解码准确性为 72.2%。在预期的时间窗口中显示出明显更大的母语 MMR。同样，FFR 的解码准确性达到了 79.6%。高的刺激-反应互相关和 9ms 的滞后表明 FFR 紧密跟踪语音。
这些发现表明，我们的范式可靠地同时捕获 MMR 和 FFR，复制并扩展了过去使用更少试验（MMR：50 次试验；FFR：200 次试验）和更短实验时间（12 分钟）的研究。
这项研究为理解皮质-皮质下相互作用以进行言语和语言处理铺平了道路，最终目标是开发一种特定于早期发育的评估工具。
