Statistics for Learning interpretable word embeddings via bidirectional alignment of dimensions with semantic concepts