Perceptual Learning and Memory Consolidation in Adults With and Without LI

Overview

This study tests a memory-based account of atypical speech perception in adults with language-based learning disability (also known as developmental language impairment \[LI\]). One perspective regarding the its etiology considers impoverished speech sound representations to be central to the linguistic symptoms observed in LI. This project examines a potential abnormality in the process of building speech sound representations in LI. Previous work by the PI has found that sleep is important for learning speech sounds. Furthermore, different measures of speech perception (identification and discrimination), reveal distinct patterns of learning that are consistent with that of declarative and procedural memory consolidation. A division of labor by declarative and procedural memory systems in the building of speech representations may imply that problems with phonology may stem from selective weaknesses in declarative or procedural memory in predictive ways. The first project Aim is to identify the memory substrates of novel phonetic category formation. In Experiment 1, the investigators will obtain behavioral measures of declarative, procedural, and speech sound learning before and after post-training sleep in 40 typical adults and 20 adults with LI. Among typical adults, a double dissociation is predicted in which speech identification will be predicted by individual differences in declarative memory, and speech discrimination will be predicted by individual differences in procedural memory. Moreover, adults with LI are predicted to demonstrate consolidation deficits across memory types. The second project Aim is to identify the neural substrates of phonetic category formation. In Experiment 2, the investigators will obtain functional magnetic resonance (fMRI) recordings of 20 TD and 20 LI adults performing post-training identification and discrimination tasks on a trained speech contrast before and after sleep. In typical adults, a Time by Speech-task interaction is predicted. Speech identification will recruit episodic (hippocampal) information on Day 1 relative to classic regions for phonological processing on Day 2. Speech discrimination will result in a change in magnitude of activation from Days 1 and 2 in the left inferior frontal gyrus (IFG), reflecting an overnight decrease in effort. LI (n=20) is predicted to demonstrate reduced overnight change in neural activation relative to TD in both tasks.

The investigators will recruit 60 adults with typical language (TD) and 40 LI participants through a variety of community and university related contacts. 40 TD adults and 20 LI participants will be enrolled in Experiment 1. 20 TD adults and 20 LI participants will be enrolled in Experiment 2. Both experiments will take place over 3 Sessions. During Session 1, the investigators will administer various measures in order to validate study eligibility, classify participants into groups, and other measures in order to further characterize our samples. To that end, participants will be administered a pure tone audiometric screening, and complete an interview questionnaire asking for background information such as handedness, language background, language and reading development, and development of motor skills. Trained research personnel will then administer an approximately 2-hour testing battery that includes measures of nonverbal cognition, language ability, reading ability, attention, executive function, and phonological processing. Participants will also complete questionnaires to assess their habitual sleep quality, sleepiness, and individual chronotpyes. All testing will be completed by a student clinician, and will be supervised by a certified speech-language pathologist. 100% of administrations will be re-scored by a trained student for accuracy. To account for attrition and non-qualifiers, we anticipate enrolling/screening up to 200 individuals in Session 1 in order to arrive at the enrollment targets for Experiments 1 and 2. In both experiments, Sessions 2 and 3 takes place on two consecutive days. Session 2 will take place in the late evening, and Session 3 will take place in the morning. We have intentionally chosen not to employ a wake-state (AM-PM) control. This is done in order to limit the potential variability in the amount of English exposure between sessions. Between-session sleep duration will be monitored with an Actigraph wGT3X-BT accelerometer (Actigraph, corp) worn on the non-dominant wrist. Actigraphs will be set to collect data in 60-second epochs via a validated algorithm (Cain \& Geremia,2011). This method of recording sleep/wake activity has been independently validated to agree with lab polysomnography at 85% accuracy (Slater et al.,2015). Experiment 1: behavioral measures During Session 2, participants will complete the three learning tasks, declarative learning, procedural learning, nonnative speech learning (in that order), and immediate post-learning assessments. During Session 3, participants will be re-assessed in nonnative speech learning, declarative learning, and procedural learning (in that order). * Declarative memory will be assessed using an object recognition task following incidental encoding. * A version of the Serial Reaction Time task will be used to measure procedural learning. * Perceptual learning of speech will be measured through performance on identification and discrimination tasks on the dental and retroflex contrast. Experiment 2: neuroimaging measures During Session 2, participants will complete the nonnative contrast training in the late evening. Then they will enter a magnetic resonance imagine (MRI) scanner, where their blood oxygen-level dependent signals will be recorded while they complete the post-training identification and discrimination assessments. The participants will complete the same in-scanner assessments of the perception of the nonnative contrast during Session 3.