Recent evidence from visual word recognition points to the important role of embedded words, suggesting that embedded words are activated independently of whether they are accompanied by an affix or a non-affix. The goal of the present research was to more closely examine the mechanisms involved in embedded word activation, particularly with respect to the "edgealignedness" of the embedded word. We conducted two experiments that used masked priming in combination with lexical decision. In Experiment 1, monomorphemic target words were either preceded by a compound word prime (e.g., textbook-BOOK/textbook-TEXT), a compound-nonword prime (e.g., pilebook-BOOK/textpile-TEXT), a non-compound nonword prime (e.g., pimebook-BOOK/textpime-TEXT) or an unrelated prime (e.g., textjail-BOOK/jailbook-TEXT). The results revealed significant priming effects, not only in the compound word and compound-nonword conditions, but also in the non-compound nonword condition, suggesting that embedded words (e.g., book) were activated independently of whether they occurred in combination with a real morpheme (e.g., pilebook) or a non-morphemic constituent (e.g., pimebook). Priming in the compound word condition was greater than in the two nonword conditions, indicating that participants benefited from the whole-word representation of real compound words. Constituent priming occurred independently of whether the target word was the first or the second embedded constituent of the prime (e.g., textbook-BOOK vs. textbook-TEXT). In Experiment 2, significant priming effects were found for edge-aligned embedded constituents (e.g., pimebook-BOOK), but not for mid-embedded (e.g., pibookme-BOOK) or the outer-embedded constituents (e.g., bopimeok-BOOK), suggesting that edge-alignedness is a key factor determining the activation of embedded words.Keywords: compound word processing; embedded words; masked priming; lexical decision For many years, researchers have examined the mechanisms involved in reading morphologically complex words. It is widely agreed upon that complex words are automatically decomposed into morphemic subunits during the initial stages of visual word recognition (as initially suggested by Taft & Forster, 1975). It has been proposed that the segmentation of morphologically complex words is based on a mechanism that identifies and strips off the affix, which then in turn allows the identification of the stem morpheme. However, this hypothesis has recently been challenged by Grainger and Beyersmann (2017) suggesting that embedded words (rather than affixes) represent the primary reading units that initiate morphological segmentation. The focus of our present study was therefore on the role of the stem, rather than the affix, during early visual word recognition. In particular, our goal was to shed more light on the mechanisms involved in activating words embedded in compound words and compound nonwords.Evidence obtained with various paradigms suggests that the processing of the constituent lexemes in compound words, such as text and...