Poster D40, Friday, August 17, 4:45 – 6:30 pm, Room 2000AB

The neural organization of speech production: A lesion-based study of error patterns in connected speech

Brielle Stark1, Alexandra Basilakos1, Gregory Hickok2, Chris Rorden3, Leonardo Bonilha4, Julius Fridriksson1;1University of South Carolina Department of Communication Sciences and Disorders, 2University of California Irvine Department of Cognitive Science, 3University of South Carolina Department of Psychology, 4Medical University of South Carolina Department of Neurology

Much of what we know about the neural architecture of language production is grounded in studies of single word retrieval. However, when we speak in everyday life, we select two or three words per second from an active vocabulary spanning an estimated 40,000 words (Levelt, 1989) and our word selection is vulnerable to “competition” from multiple sources. While numerous studies have explored single word retrieval, there is a need to evaluate more naturalistic language like the kind of connected speech we produce every day. Indeed, prior behavioral work (Mayer & Murray, 2003; Herbert et al, 2008; Nicholas et al, 1989) suggests that the distribution of paraphasias (word errors) in naming and connected speech differ, likely because connected speech requires dynamic changes in the linguistic system. In the current study, we evaluated the distribution of paraphasias and associated brain damage during a connected speech task and a naming task. 140 participants with left hemisphere stroke and pre-morbid right-handedness were retrospectively included (age: M=60.1±9.75 years; time since stroke: M=49.64±42.92 months). We divided participants into three groups: a group with the connected speech assessment (CS group: N=61, 40 males); a group with the Philadelphia Naming Test (Roach et al, 1996) (PNT group: N=113, 66 males); and a group with both the connected speech assessment and PNT (CSN group: N=34). Connected speech samples were acquired from three picture descriptions (~2 min duration each) and were coded using CHAT/CLAN (MacWhinney, 2000). Presence of aphasia was identified using the Western Aphasia Battery-Revised (Kertesz, 2007). Paraphasias from the PNT and connected speech samples were defined as verbal (semantically related or unrelated), phonemic or neologistic. To evaluate brain damage associated with each paraphasia type, we employed voxelwise lesion-symptom mapping (VLSM; voxelwise p<0.05 and cluster p<0.01; multiple comparison correction of 5000 permutations) for paraphasias in the CS and the PNT groups. To evaluate behavior, we compared the distribution of paraphasias between tasks in the CSN group. While paraphasias in connected speech and naming shared underlying neural substrates, analysis of their distribution suggested that lexical-semantic load was likely reduced during connected speech. We demonstrated that verbal (real word) paraphasias and phonological paraphasias during both connected speech and naming loaded onto the ventral and dorsal stream of language, respectively. Furthermore, our results supported prior evidence localizing semantically related paraphasias to more anterior left hemisphere temporal cortex and unrelated paraphasias to more posterior temporal and temporoparietal cortex, suggesting a gradient of specificity as one translates visual recognition from temporo-occipital cortex to posterior and subsequently anterior temporal cortex. Only VLSM analyses performed in the CS group, and not the PNT group, found significant brain damage associated with phonological paraphasias, suggesting that connected speech may be a particularly salient task on which to further evaluate lexical-phonological processing in the brain. The results presented here demonstrate a common neural substrate for paraphasias as well as characterize the dynamic processes of the language system occurring during connected speech, confirming the need for continued evaluation of the neural substrates of connected speech processes.

Topic Area: Meaning: Lexical Semantics