Anti-homophony and rhizotony in the Spanish preterite

Abstract Speakers tend to avoid homophony that is problematic. While Spanish admits some homophony of 1SG and 3SG forms in three arrhizotonic tenses/moods (imperfect, past subjunctive, conditional), in two of the three rhizotonic tenses/moods (present indicative and preterite), the 1SG and 3SG contrast systematically. In the preterite, contrast ( DĪXĪ ∼ DĪXIT ) was lost in Old Spanish ( dix[e] ∼ dix[e] ) but subsequently restored ( dije ∼ dijo ). The development dix(e) > dijo has been attributed to analogical extension of arrhizotonic – o , (cf. amó ), with destressing. This article proposes that said extension is the result of anti-homophony (AH). Like analogical extension, AH improves form-function isomorphism yet applies primarily to higher-frequency forms. It is argued that AH targeted 1SG ∼ 3SG homophony selectively, affecting only verb forms without an explicit tense/mood marker (e.g., dixe ∼ dixe vs tenga ∼ tenga ). This morphological restriction on AH explains why homophony persisted in four of the seven basic tenses (e.g., tenga ∼ tenga , tenía ∼ tenía , tendría ∼ tendría , tuviera [–se] ∼ tuviera [–se] ). The modern pairs fui ∼ fue and di ∼ dio , which are the outcome of heavy competition, are shown to be a straightforward result of AH. In a broader perspective, the drive to distinguish 1SG and 3SG forms has brought about a “1–3 pattern” with a range similar to that of the N-pattern, in which stress is retained in certain paradigmatic cells in the absence of conditioning phonological factors. Destressing – o in forms like dijo and vino served to maintain the integrity of the 1–3 pattern as an organizing principle.