This section reports results from some simple algorithms and goes on to show experiments on our model, investigating how variations in POS tagset size, ngram size and smoothing affect performance.