There has been some dispute in the literature over the number of languages which are necessary input to produce a true pidgin. D. DeCamp says that any two languages in contact can result in an 'interlingual improvization' but more than two languages in contact are required for the development of a true pidgin. Some definitions include formal characteristics: restricted vocabulary, absence of gender, true tenses, inflectional morphology, or relative clauses. A number of linguists have tried to explain the similarities which pidginized speech varieties show in the expression of grammatical categories and syntactic relationships by appealing to more general principles of linguistic organization motivated by specialization to the referential function. The feature comparative in L. Markey's typology refers to the use of non-indigenous comparatives in creoles. J. Haiman says that there is an inverse correlation between the lexical expansion of a language and the iconicity of its grammar.