Quasi-Phonemic Contrasts in Spanish *

The striking thing about phonology is that the infinit e phonetic variety in the utterances of any language can be reduced to a small i nventory of contrastive units or phonemes. The bad news is that oftenti m s phonemization is problematic in some corners of the language. Typically if the same language has been investigated by two linguists, we w ill get slightly different phonemic inventories; and these differ ences of opinion usually do not go away as the language is studied more ext ensively. Both the considerable extent to which we normally find agreement amon g linguists in the phonemic analysis of a given language and the existence of areas of disagreement are remarkable facts. In this paper I consider in detail three specific aspects of th e phonemic inventory of Spanish that remain controversial and argue t hat in all three cases it is useful to speak of quasi-phonemic contrasts (alth ough not exactly in the same sense in every case). The paper ends with some general considerations about the nature of phonological categorization. Let me start with an anecdote. I recently submitted a paper repo rting n experimental data on Basque suprasegmentals to a journal. A reader m the sensible suggestion that the examples, which were in co nventional Basque orthography, should also be given in IPA transcri ption. This seemed reasonable enough, and I replied that I would comply. How ever, it immediately became clear to me that this was easier said than done, as I would like to be objective and accurate in the transcription. Should I provide a phonetic transcription? For each sentence over a hundr ed tokens were examined in the study (several repetitions in different pragmatic contexts by several speakers). Doing a phonetic transcriptio n would thus require making some decisions as to how to solve the variabi lity present in the data. To give just one example, one of the experimental senten c s started with the word mollako ‘of the pier’. In this word the k was usually