This paper investigates four different types of pauses in conversational narrative: the filled pauses er and erm, and short and long silent pauses. The study is based on the Narrative Corpus (NC), a recently created corpus of everyday narratives. The texts, which include both the narrative and some context, have been annotated for important textual components. The current analysis reveals that pauses are more frequent in conversational narrative than in general conversation. We suggest three factors that account for this high frequency: (i) the need for narrators, in the opening utterance of the story, to provide specific information to orient listeners to the situation in which the events unfolded, (ii) the need to coordinate narrative clauses to match the story events, and (iii) the preference of narrators to present speech, thought, emotion and gesture using direct-mode discourse presentation, which is more "dramatic" but also more costly in terms of reference resolution.

This paper outlines a modest approach to XPath and XQuery, tools allowing the navigation and exploitation of XML-encoded texts. The paper starts off from where Andrew Hardie’s paper “Modest XML for corpora: Not a standard, but a suggestion” (Hardie 2014) left the reader, namely wondering how one’s corpus can be usefully analyzed once its XML-encoding is finished, a question the paper did not address. Hardie argued persuasively that “there is a clear benefit to be had from a set of recommendations (not a standard) that outlines general best practices in the use of XML in corpora without going into any of the more technical aspects of XML or the full weight of TEI encoding” (Hardie 2014: 73). In a similar vein this paper argues that even a basic understanding of XPath and XQuery can bring great benefits to corpus linguists. To make this point, we present not only a modest introduction to basic structures underlying the XPath and XQuery syntax but demonstrate their analytical potential using Obama’s 2009 Inaugural Address as a test bed. The speech was encoded in XML, automatically PoS-tagged and manually annotated on additional layers that target two rhetorical figures, anaphora and isocola. We refer to this resource as the Inaugural Rhetorical Corpus (IRC). Further, we created a companion website hosting not only the Inaugural Rhetorical Corpus, but also the Inaugural Training Corpus) (a training corpus in the form of an abbreviated version of the IRC to allow manual checks of query results) as well as an extensive list of tried and tested queries for use with either corpus. All of the queries presented in this paper are at beginners to lower-intermediate levels of XPath/XQuery expertise. Nonetheless, they yield fruitful results: they show how Obama uses the inclusive pronouns we and our as a discursive strategy to advance his political strategy to re-focus American politics on economic and domestic matters. Further, they demonstrate how sentence length contributes to the build-up of climactic tension. Finally, they suggest that Obama’s signature rhetorical figure is the isocolon and that the overwhelming majority of isocola in the speech instantiate the crescens type, where the cola gradually increase in length over the sequence.

No abstract