Recently I've been trying to make an effort to write up fairly comprehensive notes for topics I teach in statistical NLP. The following notes follow weeks one and two of my spring 2013 Coursera course very closely:

• Week 1: Language models
• Week 2: Tagging problems, and Hidden Markov models
• Week 3: Probabilistic context-free grammars (PCFGs)
• Week 4: Lexicalized PCFGs
• Week 5: IBM Models 1 and 2 for machine translation
• Week 6: Phrase-based translation models
• Week 7: Log-linear models
• Week 8: (see section 3 of this note for some background on log-linear models for tagging, i.e., MEMMs) Log-linear models, MEMMs, and CRFs
• Week 8: more recent notes (created November 2014) on MEMMs (Log-linear taggers)