The main goal of the present study was to investigate whether schema-based encoding of novel word pairs (novel compound words) supports the formation of unitized representations and thus associative familiarity-based recognition. We report two experiments that both comprise an incidental learning task, in which novel noun-noun compound words were presented in semantically congruent contexts, enabling schema-supported processing of both constituents, contrasted with a schema-neutral condition. In Experiment 1, the effects of schema congruency on memory performance were larger for associative memory performance than for item memory performance in a memory test in which intact, recombined, and new compound words had to be discriminated. This supports the view that congruency boosts associative memory by promoting unitization. When contrasting event-related potentials (ERPs) for hits with correct rejections or associative misses, an N400 attenuation effect (520–676 ms) indicating absolute familiarity was present in the congruent condition, but not in the neutral condition. In line with this, a direct comparison of ERPs on hits across conditions revealed more positive waveforms in the congruent than in the neutral condition. This suggests that absolute familiarity contributes to associative recognition memory when schema-supported processing is established. In Experiment 2, we tested whether schema congruency enables the formation of semantically overlapping representations. Therefore, we included semantically similar lure compound words in the test phase and compared false alarm rates to these lures across conditions. In line with our hypothesis, we found higher false alarm rates in the congruent as compared to the neutral condition. In conclusion, we provide converging evidence for the view that schema congruency enables the formation of unitized representations and supports familiarity-based memory retrieval.

本研究的主要目标是调查新词对（新复合词）的基于模式的编码是否支持单元化表示的形成，从而支持基于联想熟悉度的识别。我们报告了两个实验，它们都包含附带学习任务，其中新颖的名词-名词复合词在语义一致的上下文中呈现，从而与图式中性条件相比，能够对两种成分进行图式支持的处理。在实验1中，在必须区分完整、重组和新复合词的记忆测试中，图式一致性对联想记忆性能的影响比项目记忆性能的影响更大。这支持了一致性通过促进单元化来增强联想记忆的观点。当将命中与正确拒绝或联想未命中的事件相关电位 (ERP) 进行对比时，在一致条件下存在指示绝对熟悉度的 N400 衰减效应 (520–676 ms)，但在中性条件下则不存在。与此相一致的是，对不同条件下命中的 ERP 进行直接比较，发现一致条件下比中性条件下有更多的正波形。这表明，当建立模式支持的处理时，绝对熟悉有助于联想识别记忆。在实验 2 中，我们测试了模式一致性是否能够形成语义重叠的表示。因此，我们在测试阶段包含了语义相似的诱饵复合词，并在不同条件下比较了这些诱饵的误报率。根据我们的假设，我们发现与中性条件相比，全等条件下的误报率更高。总之，我们提供了一致的证据来证明图式一致性能够形成统一的表示并支持基于熟悉度的记忆检索。