Split auxiliary systems: A cross-linguistic perspective (review)

This collection of papers about split auxiliary systems concentrates on the HAVE/BE alternation in Germanic and Romance, but it also contains an account of the use of BE as an irrealis marker in Romanian, a historical study of the HAVE perfect in Greek and Latin, and a couple of papers on other splits, namely particle selection in Korean auxiliary formation and auxiliary selection in Chinese.

Perusing the volume, one is immediately struck by how much the outlook on the problem has changed in the thirty years since David Perlmutter (1978) formulated the unaccusativity hypothesis. At that time, the alternation of HAVE and BE as the auxiliary in the perfect tense was seen as a simple diagnostic for a classification of intransitive verbs that could be modeled in rather straightforward syntactic terms. Now the auxiliary split alone is shown to be conditioned by a complex number of semantic factors, and its relation to unaccusativity is not perceived as unproblematic. The choice of auxiliary has proven to be subject to variation within one and the same language, as well as across languages. This volume gives a good overview of the factors proposed as influencing the alternations and the more recent theoretical proposals to cope with the variation. The methodology used is changing from the traditional introspection and small-scale informant work, still the preferred tools of most researchers in this volume, to the use of questionnaires, experiments, and corpus studies. Unfortunately, most of the studies in this volume are not very explicit about the way the data was collected even when some rather extensive data collection must have taken place, so the reader's understanding of the data-analysis methods used remains impressionistic. Some of the papers discuss not only auxiliary selection but also other unaccusativity tests, but I do not go into the details of those discussions in this review.

The most ambitious paper is GERALDINE LEGENDRE's 'Optimizing auxiliary selection in Romance'. It proposes an optimality-theoretic account of the auxiliary selection hierarchy (ASH) developed in Sorace 2000. The ASH subdivides the intransitive verbs into seven semantic categories. The closer a verb is to the head of this hierarchy, the more likely it is to select BE. At the tail end of the hierarchy one finds HAVE verbs. Legendre translates the ASH into a number of predicate features: telicity, direction, state, inherent displacement, internal motion, inherent homogeneity, and inherent volitionality. This leads to a subdivision of the ASH change-of-state verbs depending on the values they take for homogeneity, telicity, and direction. Homogeneity is a notion adapted from McClure 1995, and it differs from the notion 'incremental theme' (Dowty 1991) in that it proposes that telic events can be homogeneous. The test for homogeneity is whether one can qualify the event as having happened 'a bit': the fact that one can say 'The apple has rotted a bit' when the apple is rotting shows that 'to rot' is homogeneous whereas 'to die' for instance is not.¹ Inherent volitionality is meant to distinguish verbs that refer to volitional processes by virtue of their meaning from verbs that are or are not volitional depending on their argument. Although Legendre makes a valiant effort to establish these categories independently, their vagueness, together with the lack of quality control on the grammaticality/acceptability judgments, remains a weak point of an overall impressive paper.

A second substantial paper based on the ASH is 'Auxiliary selection and split intransitivity in Paduan' by MICHELA CENNAMO and ANTONELLA SORACE. This is a mainly descriptive paper based on a corpus of interviews with twenty-four native speakers. It is disappointing that the authors are rather cavalier in presenting the quantitative underpinnings of their extensive study. They state that their sample included working-class and middle-class speakers as well as different age groups, but no further details about the breakdown are given.

Auxiliary selection in Paduan is stable at...