An unresolved issue regarding working memory (WM) processes relates to whether domain-general attentional resources are required to form and store bound representations. Recent evidence suggests that visual WM performance during tasks that require binding of face-scene pairs is disrupted by concurrent divided attention to a greater degree than when needing to remember only faces or scenes (Peterson & Naveh-Benjamin, 2017). These findings contrast with associative long-term memory (LTM) studies, which have found no differential impact of divided attention on associative relative to item memory (Naveh-Benjamin, Guez, & Marom, 2003). In the current study, a verbal WM change detection paradigm, incorporating methods typical of LTM paradigms as well, was leveraged to examine memory for items and item-item bindings, for unrelated word pairs sampled with and without replacement across trials within distinct experimental blocks. In Experiment 1, WM performance was measured under within-domain verbal interference. In Experiment 2, a cross-domain auditory interference task was used to parametrically vary concurrent load. Whereas the results of Experiment 1 revealed that within-domain interference was sufficient to elicit a binding deficit by disrupting verbal rehearsal, in Experiment 2, parametric variation of a cross-domain interference task revealed a divided attention-related binding deficit that increased in magnitude with increased task difficulty. In both experiments, an item-item binding deficit was observed, in comparison with item memory, regardless of whether word pairs were sampled with or without replacement. These findings reveal a dissociable influence of within-domain and cross-domain interference tasks on item-item binding processes in verbal WM.