This OSF site contains source code and example data for machine learning analyses presented in McMurray, B., Sarrett, M., Chiu, S., Black, A., and Aslin, R.N. (submitted) A paradigm for decoding the neural time-course of spoken word recognition from EEG. **Abstract** The efficiency of spoken word recognition is essential for real-time communication. There is consensus that this efficiency relies on an implicit process of activating multiple word candidates that compete for recognition as the acoustic signal unfolds in real-time. However, few methods capture the neural basis of this dynamic competition on a msec-by-msec basis. This is crucial for understanding the neuroscience of language, and for understanding hearing, language and cognitive disorders in people for whom current behavioral methods are not suitable. We apply machine-learning to standard EEG recording to decode the strength and timing of lexical competition during spoken word recognition. We trained a support vector machine to identify which word was heard and analyzed the patterns of confusion over successive 20 msec increments. Results mirrored psycholinguistic findings: Early on, the decoder was equally likely to report the target (e.g., baggage) or a similar sounding competitor (badger), but by around 500 msec, competitors were suppressed. Follow up analyses show that this is robust across EEG systems, with fewer channels, and with fewer trials. Results are individually interpretable, and show high reliability. These findings confirm the utility of a neural paradigm for assessing the dynamics of word recognition, with potential applications for understanding development a variety of clinical disorders. **CONTENTS** Files included in this package. 1. UIs001db.mat. This is the output of our custom EEG processing pipeline for a single subject. 2. sub1.mat. This mat file contains a single structure that has been processed (with process.m) to build a data structure compatible with ECOGanal. 3. ECOGanal.zip. ECOGanal package with code for data visualization and machine learning. 4. Code.zip. Code specific to this project. 5. Process.m. Builds an EcogAnal compatible from the EEGlab data. 6. Features.m Does feature selection and sets free parameters for the SVM. 7. Timecourse.m Base analysis of the timecourse of decoding. All of those files should be placed in the same directory with ECOGanal and code in their own folders. Each function is fully commented. **process.m** This function converts the EEGlab data in UIs001db.mat to the ECOGanal format. This involves reshaping the EEG data, addin information about the trials and stimuli, conducting a time/ frequency analyses, and implements some visualzations. You can skip this step if you already have sub1.mat **features.m** This function will simultaneously do two things. First, it can search the feature space (e.g., different timewindow lengths, different forms of the EEG, time-frequency features etc). Second, for each feature it will search the space of C and Gamma (the two free parameters of the SVM). It is currently configured to only examine a single feature and just report the free parameters (see comments to explore more of the feature space). This assumes a file containing an ECOGanal datastructure (e.g., sub1.mat, output by process.m). It outputs a text file containng the best C and Gamma as well as the accuracy of the decoder for each subject, and for each feature set that was tested. **timecourse.m** This does the timecourse analysis (e.g., Figure 2 of the paper). It assumes .mat file containign an ECOGanal formatted database (e.g., sub1.mat), and the text file output by features.m