Prior neuropsychological evidence suggests separable, domain-specific working memory (WM) buffers for maintaining phonological (i.e., speech sound) and semantic (i.e., meaning) information. The phonological and semantic WM buffers can be damaged separately from each other, having distinct effects on both language comprehension and production. The phonological WM buffer’s proposed location is the left supramarginal gyrus (SMG), whereas semantic WM has been related to the left inferior frontal gyrus (IFG), middle frontal gyrus (MFG), and angular gyrus (AG). However, less is known about the white matter correlates of phonological and semantic WM and any corresponding relationships to multiword language processing. Here, I report 1) the white matter correlates of phonological and semantic WM, 2) the white matter correlates of multiword language production and comprehension, and 3) evidence that semantic WM can mediate the relation between tract integrity and language processing. These analyses focus on the left hemisphere as there is strong evidence for the involvement of left hemisphere regions in verbal WM and language processing. I also present exploratory analyses of the relations between right hemisphere tracts and both WM and language processing to address the possibility that WM and/or language function reorganizes to the right hemisphere after left hemisphere brain damage. Participants were 45 individuals with left hemisphere brain damage. Each participant was tested on a battery of single word processing, phonological WM, semantic WM, sentence comprehension, and narrative production tasks. T1 and diffusion weighted scans were also obtained for each participant. Virtual dissections were performed for each participants’ arcuate fasciculus (AF), inferior fronto-occipital fasciculus (IFOF), inferior longitudinal fasciculus (ILF), middle longitudinal fasciculus (MLF), and uncinate fasciculus (UF). These tracts were chosen because they have terminations in regions that have previously been identified as potential domain-specific WM buffers: the SMG for phonological WM and the IFG, MFG, and AG for semantic WM. In chapter 2, to test the relation between white matter tract integrity and domain-specific working memory, I took a multiple regression approach which allowed us to control for both single word processing and gray matter damage to the regions where each tract terminates. I predicted that the indirect pathway of the left AF would support phonological WM while the left IFOF, ILF, MLF, UF, and direct pathway of the AF would support semantic WM. Further, because semantic WM is critical for multiword language processing, it was predicted that tracts supporting semantic WM should also support multiword language production and comprehension. In line with the predictions, the left ILF, left MLF, left IFOF, and left direct segment of the AF were related to semantic WM performance. Relationships between semantic WM and right ILF and right IFOF integrity were also observed. Phonological WM was related to both the left ILF and MLF. In chapter 3, I used a multiple regression approach to investigate the relation between tract integrity and language processing while controlling for single word processing and gray matter damage. I found that the left AF, IFOF, ILF, and right IFOF were related to language comprehension while the left IFOF and ILF were related to language production. The right IFOF was also related to language comprehension. Further, I used mediation analysis to investigate whether semantic WM fully or partially mediates the relation between left IFOF and left ILF tract integrity and multiple aspects of multiword language processing. Additionally, phonological WM mediated the relation between left ILF integrity and language comprehension. The results of this work inform our understanding of the white matter correlates of both WM and language processing, especially semantic WM, and the neural basis of the relation between semantic WM and language processing, which no previous work has investigated. From a theoretical perspective, this work is also important for adjudicating between alternative buffer theories of verbal WM and should inform future research on neuroplasticity, particularly how the right hemisphere may support cognitive processing after left hemisphere brain damage