Description:
<jats:title>Abstract</jats:title><jats:p>In this article, I show that in spoken French 1s<jats:italic>moi</jats:italic>, unlike 3s<jats:italic>lui</jats:italic>, has undergone both semantic bleaching and heavy prosodic erosion and has therefore progressed along Lehmann’s parameters of grammaticalization. As will be argued, this person-asymmetry is rooted in the fact that<jats:italic>moi</jats:italic>, unlike<jats:italic>lui</jats:italic>, is largely used as a discourse-structuring device. Based on a phonetic analysis, I show that the main factor underlying the prosodic weakening of left-detached<jats:italic>moi</jats:italic>is string frequency, i.e., the degree of predictability of the element immediately following<jats:italic>moi</jats:italic>. The erosion of<jats:italic>moi</jats:italic>in spoken French is particularly strong when it is followed by the clitic 1s<jats:italic>je</jats:italic>; it is even stronger in those cases where<jats:italic>moi</jats:italic>introduces a stance-verb expression of the type<jats:italic>moi je trouve que</jats:italic>… ‘I think that …’. However, despite the massive effects produced by string frequency, the prosodic, distributional and discourse-pragmatic data presented here do not support the view that the prosodically weakened variants of<jats:italic>moi</jats:italic>instantiate a construction that is distinct from non-clitic<jats:italic>moi</jats:italic>.</jats:p>