Imperfect language learning vs. dynamic sound change

Modern Hebrew provides an idiosyncratic case for historical linguistic study: due to the discontinuity of its use as a spoken language, differences between contemporary structures and classical ones do not necessarily reflect change processes, but may instead result from imperfect language learning by the original L2 speakers of Modern Hebrew at the initial stages of speech revival. This article offers a new research direction for delineating the boundaries between the two types of phenomena based on the recent discovery of two collections of recordings of spontaneous Hebrew speech made in the 1960s. Focusing on one conspicuous sound change in contemporary Modern Hebrew, namely the transition from [i] to [e] in the prefix of the verbal pattern hif’il, we show that the variability in contemporary language between hif’il and hef’il has two distinct sources: (i) an initial state of variability between [i] and [e] in forms derived from weak root verbs (initial-[n] and middle-[w/y] roots, e.g. higi’a-hegi’a ‘arrived’) due to imperfect language learning in the initial phases of the formation of Modern Hebrew; and (ii) a recent change from [i] to [e] in forms derived from regular roots (e.g. hitxil-hetxil ‘started’). In this category, the 1960s recordings attest to a stable realization of [i] amongst all age groups, with deviations from the rules of traditional Hebrew grammar occurring only marginally. Based on this data, the measure of synchronic variation documented in the 1960s recordings is analyzed as a precursor of the sound change that developed in the language at a later stage.