In any model of the mental lexicon, representations of words’ forms are needed. The DLM makes use of numerical vectors. The simplest numerical form vectors are derived from orthographic or phonological representations, which are rerepresented as binary vectors (i.e., as lists of numbers containing only 1s and 0s), which results in a rather abstract form representation (see Chapter 1.4). For a word such as, for instance, aap ‘monkey’, one possible binary vector representation is obtained in the following way. First, the word is split up into overlapping bigrams. Thus, aap is split into the bigrams #a, aa, ap, p#, with # denoting word boundaries. If we split the word into tri-grams, it would look like this: #aa, aap, ap# (see Box 4.1 as to why we use n-gram representations rather than making use of phonemes).