# Breaking down prefixed words is unaffected by morphological boundary opacity: Evidence from behavioral and MEG experiments

## TL;DR

This study shows that the brain can break down words into their parts even when the boundaries between word parts are unclear.

## Contribution

The research demonstrates that morphological decomposition is robust even with obscured morpheme boundaries, using behavioral and MEG data.

## Key findings

- Behavioral masked priming data show decomposition of prefixed words with obscured boundaries.
- MEG data reveal neural activity modulated by stem-to-whole word transition probability.
- Decomposition remains unaffected by morphophonological changes in Tagalog prefixed words.

## Abstract

Previous experiments support an initial stage of early, form-based visual word recognition, where morphologically complex words like adorable are segmented into morphemes {adore}+{-able}, despite an orthographic change in the stem. However, most experiments have focused on words with clear boundaries between the affix and stem, making decomposition more straightforward. We investigate whether obscured boundaries between the prefix and stem affect morphological decomposition. Using Tagalog as a test case, we compare the processing of prefixed words [1] without morphophonological changes (e.g., {mang}+{hila} becomes manghila “to pull”), [2] with nasal assimilation obscuring prefix identity (e.g., {mang}+{bulag} becomes mambulag “to blind”), and [3] with nasal substitution obscuring both prefix and stem identities and their morphological boundary at orthographic and phonological levels (e.g., {mang}+{tulak} becomes manulak “to push”). Crucially, these morphophonological changes exhibit variability: nasal substitution is more likely than assimilation for voiceless-initial stems, while the opposite holds for voiced-initial stems. Experiment 1 presents behavioral masked priming data that prefixed words are decomposed into morphemes, even with obscured {prefix}+{stem} boundaries. Experiment 2 further supports these results with data from magnetoencephalography showing neural activity is modulated by stem:whole word transition probability, which indicates morphological decomposition. Findings from both experiments unambiguously show that early, form-based decomposition is robust and flexible enough to recognize morphemes, despite morphophonological changes obscuring the {prefix}+{stem} boundary.

The online version contains supplementary material available at 10.3758/s13423-025-02758-7.