The present study explored effects of working memory (WM), structure, and salience on the processing of irregular past participial forms in English relative clauses (RCs) and participial reduced relative clauses (PRRCs) among L1 Thai learners. Based on Bayley's research (1994), the salience in this study was related to phonological changes from the past tense form of English irregular verbs into their past participial form. The research included two groups of past participles with different salience degrees, i.e., an internal vowel change plus an addition of the syllabic [?n] morpheme and an internal vowel change plus an affixation of n. The research participants were seventy Thai undergraduate students with high English proficiency. The research instruments comprised a reading span task and a self-paced reading task. The reading span task was used to classify the learners according to WM degree into two groups: higher and lower WM groups. The self-paced reading task looked into the participants' processing of the two past participial forms. The study hypothesized that the two participant groups with different levels of WM would manifest different fashions for online and offline processing. Moreover, the higher WM learners were predicted to take as similar amounts of reading time for RCs as their lower WM counterparts and spend more time reading PRRCs than the participants with fewer cognitive resources. The research findings partially supported the first hypothesis, but refuted the second hypothesis. That is, the distinction between the cognitive capacity levels of the two participant groups tended to affect their online processing, but not offline processing. This suggested that WM effects could depend on task types, which was consistent with the findings of Hopp (2015) and Zhou et al. (2017). Furthermore, the higher WM participants were more likely to read the two constructions faster than the lower span ones. The finding indicated that a high level of cognitive capacity could increase L2 learners' speed of combining the upcoming with the preceding information�(Just & Carpenter, 1992). This study made a contribution to L2 processing research by substantiating the dependence of WM impact on the task type as well as the relationship between L2 learners' cognitive resources and their processing speed. The research also provided theoretical and pedagogical implications.