Lingvisticæ Investigationes - Volume 40, Issue 1, 2017

Some long-standing questions surrounding anticausatives in languages like French include whether the morphological marking (presence/absence of se) correlates with interpretational differences and/or different syntax. We examine the three anticausatives classes (optional se, obligatory se, no se) in three aspectual contexts and formulate a generalization whereby a default morphological form (reflexive-/non-reflexive-marked) can be identified for each context, plus an interpretive anti-blocking effect: if the lexicon does not provide the default form then the other form (regardless of morphology) preserves the aspectual interpretation of its transitive source. French anticausative se is tied to lexical aspect (rather than syntax), but the distribution is complex and non-transparent. We argue that the grammar allows bidirectional competition among forms and interpretations and the formalize analysis in Bidirectional OT (Superoptimality).

Il est possible de modéliser la variation et le changement valenciel à l’aide des structures formelles que nous nommons polygraphes. Ces derniers étendent les arbres syntaxiques dépendanciels en autorisant des relations à porter sur d’autres relations. En prenant l’exemple de l’expression prépositionnelle ou non du datif en ancien français et en français moderne (complément des verbes donner, aider, et pardonner), nous montrons comment les polygraphes peuvent être employés pour montrer géométriquement la différence entre une préposition obligatoire et une préposition facultative. Le potentiel de ces structures permet également de visualiser l’évolution et de comparer le comportement des lexèmes.

This article explores the question of why Medieval French does not use se-constructions as exhaustively as other Medieval Romance varieties and why the indefinite pronoun on is clearly the preferred choice. It is claimed that se exhibits affix-like behavior in passive constructions and that it is very likely to be a morphological exponent of absorbing the [+Def] feature on T in IACs. These assumptions are connected to the possibility within a language system to store affixes as independent lexical entries, following Speas, 1995 . It will be shown that Medieval French seems to be a language system in flux in which signs of NSL > non-NSL were present over a longer period of time. The uncertain status of affixes within these radical changes would have prompted learners to go for a multifunctional solution that is readily available and unequivocally identifiable: on.

This paper analyzes transitivity and valency in Northern Akhvakh, a language belonging to the Andic group of languages included in the Northeast Caucasian (or Nakh-Daghestanian) family. Northern Akhvakh clause structure is characterized by an extreme flexibility of constituent order, omissibility of arguments with an either anaphoric or unspecified reading, and fully consistent ergative coding of core NPs. Northern Akhvakh has a very low rate of transitivity prominence, and an extremely strong tendency to derive the causal member of noncausal / causal pairs from its noncausal counterpart. Ambitransitivity is very marginal, and the productivity of morphologically unmarked valency alternations is very limited. Causative derivation is the only valency changing mechanism involving verb morphology, and ingestion verbs are the only transitive verbs for which causative derivation is productive.

We investigate the effect of French clitic construction on verb learning. In French, object pronouns precede the verb, and the canonical direct object position remains empty. We test whether children treat such contexts as input for transitivity (since a direct object is morphologically identified) or optional transitivity (due to the empty direct object position). Forty-eight monolingual French preschoolers heard verb input with clitics and noun phrases as direct objects, in two input conditions: obligatory transitivity, and mixed optional transitivity. Results show that children are sensitive to the input, but produce more sentences with null implicit objects in the clitic conditions. This provides evidence that specific properties of a language (e.g. clitic constructions), affect the acquisition of verbal classes.