The Effect of Verbs on the Prosodic Marking of Information Status:
Production and Perception in German

Christine T. Röhr
IfL Phonetik, University of Cologne

In intonation languages like German the marking of information status is an important linguistic function of prosody. Recent annotation systems are able to capture fine-grained differences in an item’s information status (e.g. types of accessible information), which have been shown to be marked by nuclear pitch accent placement and/or type. However, most annotation systems tend to concentrate on the information status of noun phrases (NPs), based on relations between two nominal expressions. Some systems also include verbs and verb phrases as a possible source of a referent’s accessibility, but verbs are usually not assigned an information status themselves. Furthermore, the effect on prosody of verbs has not yet been investigated.
In a production and a follow-up perception experiment we tested the effect of reference relations between nouns and verbs on their prosodic realization. Beside new information, i.e. nouns/verbs that are not derivable from the previous text, we distinguish between three different types of accessible/given information by using different types of noun-verb pairs. The verbs denote an event of intentionally creating an element (e.g. fotografieren ‘to photograph’) and the corresponding nouns either denote an instrument for creating a related element (e.g. Kameras ‘cameras’) or the created element itself, namely the result. The noun denoting the result was either morphologically unrelated to the verb (e.g. Bilder ‘pictures’) or displayed the same word stem (labelled result-stem, e.g. Fotografien ‘photographs’). The target nouns and verbs were part of constructed mini dialogues and occurred in consecutive sentences in both orders.
Results show that nouns denoting a created element (independent of whether morphologically related or not) were less often marked by a nucleus than instrument nouns and new nouns. This mirrors the stronger semantic relatedness of both types of result nouns to the corresponding verb. For the verbs, the differences in prosodic marking are less distinct, but seem to reflect more fine-grained differences in their information status: With increasing discourse-givenness of the verb (from new through instrument and result to result-stem), the nuclear accent was placed increasingly often on the adverb, rather than the verb itself. Acceptability ratings by listeners verify the different preferences in prosodic marking with regard to the investigated semantic relations. Thus, differences in a verb’s informativeness are reflected by some variation in nuclear accent placement, and should thus be integrated into a wider notion of information status.