Prosody and musical rising two-tone patterns: sound examples

Ernst Dombrowski, Thurid Holzrichter, Niels Münz, Alexander Nowak, Monika Poschmann

Department of Psychology, Christian-Albrechts-University Kiel

Introduction:

In this document we provide sound material from two experiments on the relation between melody in speech and melody in music. The experiments deal with two patterns occurring in German intonation, rise and rise-fall, which are compared with a particular musical configuration: a set of rising two-tone patterns comprising thirteen melodic intervals from unison to octave. A description of the experiments is given in a paper at the 16th Congress of Phonetic Sciences, Saarbrücken 2007 (Dombrowski, Holzrichter, Münz, Nowak and Poschmann 2007a). We include the poster presented at this conference. The poster was also shown at the Annual Conference 2007 of the German Society for Music Psychology, Gießen (Dombrowski, Holzrichter, Münz, Nowak and Poschmann 2007b).

Background and question:

The work reported here is based on the idea that speech melody and musical melody may be two sides of the same thing. However, a structural comparison of melody in speech and music has to face at least two problems:

1. If a piece of melody in speech and a piece of melody in music show the same con­tour, this does not mean that they are really same. There may only be a surface similarity and no deep parallel. Among others, this is because a compari­son of contour does not include basic aspects of the musical use of pitch, e.g. pitch relationship, scales, tonality, and harmony.

2. Trying to transform intonation contours into musical level pitches inevitably leads to a loss of contour information. E.g., the gliding phrase-final rise often occurring in lin­guistic question patterns is in conflict with level pitch. The same applies to rise-fall contours where a pitch peak goes over to a falling movement.

In particular, losses of contour information apply to the phrase-final features of melodic courses, i.e. phrase-final pitch movements, boundary tones etc. Occurring at the end of a melodic phrase, they can invert its global character.

Losses of contour features in musical melody can be used to study the relation between melody in speech and music, because the part of the “lost” contour information may be taken by other musical features. This is the question of the present experiments. In the experiments we confront two prosodic contour types with a series of melodic intervals. The intervals represent different kinds of musical information: pitch range, consonance vs. dissonance, melodic (or tonal) function. All three may contribute to melodic transformation of speech melody into musical melody vice versa.

Sound examples:

In our congress paper we demonstrate the idea of lost contour information in musical melody with an example by J. S. Bach -- who is one of the most famous transformers of speech to music. Sound example 1 is a section from the bass aria No. 24 Eilt, ihr angefocht­nen Seelen ("Come, ye souls whom care oppres­ses") in Bach’s St John Passion, BWV 245. You hear a set of Wohin ("Where") calls sung by the choir interacting with the soloist (bars 49 to (?), also cf. Fig.1). The Wohin calls may be seen as perfect musical representa­tions of a linguistic question pattern. However, they can stand for different prosodic patterns used in German intonation to express "question", in particular, rise and rise-fall.

The sound example is taken from a recording with Robert Holl, bass, the Arnold Schoenberg Chor, Wien, and the Concentus Musicus (under direction of Nikolaus Harnoncourt; Teldec 9031748622, 3/1995).

(The sound-example is not yet available because of copyrights).

Figure 1: The first triplet of Wohin calls from the bass aria No. 24 in J. S. Bach’s St John Passion BWV245, bars 49 to 51 (Bach 1981).

The contours begin with a rising movement towards the accented vowel they are attached to (i.e. [i] in wohin). But they end with different phrase-final modifications: further rising vs. falling. The rises are high rising early valleys in terms of the Kiel Intonation Model (KIM, Kohler 1991a, b). In an autosegmental-metrical representation they are H* or L* pitch accents with high boundary tones.

The rise-falls are terminal medial peaks in the Kiel Intonation Model. In an autosegmental-metrical representation they are H* pitch accents with low boundary tones (cf. Grice and Baumann 2000).

In our experiments rising musical two-tone pat­terns are compared with four disyllabic utterances:

Für Sie/sie ("for you/her"),

Mal sehn ("do you want to have a look"),

Wieso ("why"),

Warum ("why").

All show an upbeat-downbeat pattern – like the utterance wohin. The four utterances were, again, naturally produced by the speaker ED. They are presented in sound example 4. For each utterance you first hear the rise and then the rise-fall.

Sound example 5 is an item from our experi­ment 1 (cf. Fig. 2). Experiment 1 was an ABX style interval-matching task. Listeners assessed on a five-point scale whether a rising two-tone pattern came closer to a prosodic rise or to a rise-fall. 13 musical inter­vals were examined. For the prosodic patterns the German disyllabic utterance Für Sie/sie ("for you/her") was used.

Figure 2: An item from experiment 1. Noise signal (white noise), utterance 1, utterance 2 (Für S/sie; rise and rise-fall vice versa), melodic interval (piano sound), response pause, noise signal.

Sound example 6 contains a longer stretch from the same experiment. First you hear instructions given to the listeners. They are followed by a set of training items. Then you hear further instructions and the beginning of the test: non-transposed intervals (in this series the A2 is the reference for the two-tone patterns). Short signals of white noise are used as separators between the test items.

Sound example 7 is an item from Experiment 2 (cf. Fig. 3). Experiment 2 was an ABX style contour-matching task. Listeners assessed whether a spoken contour fitted a rising fourth or fifth, i.e. musical patterns with different underlying tonal functions. Four utterances were used: Für Sie/sie ("for you/her"), Mal sehn ("do you want to have a look"), Wieso ("why"), Warum ("why").

Figure 3: An item from experiment 2. Noise signal (white noise), interval 1, interval 2 (rising fourth and rising fifth vice versa), spoken utterance (rise or rise-fall), response pause, noise signal.

Sound example 8 presents the four training items from experiment 2 preceded by a short instruction.

Click on the following link to find our poster presented at the 16^th International Congress of Phonetic Sciences in Saarbrücken, 2007.

REFERENCES

1. Bach, J. S. (1981). St John Passion: BWV 245, Vocal Score (Supplementary edi­tion based on J. S. Bach, Neue Ausgabe sämtlicher Werke, II, 4). Kas­sel: Bärenreiter.

2. Dombrowski, E., Holzrichter, T., Münz, N., Nowak, A., Poschmann, M. (2007). Prosodic rise and rise-fall contours and musical rising two-tone patterns. Proceedings of the 16^th International Congress of Phonetic Sciences. Saarbrücken, 1145-1148.

3. Grice, M., und Baumann, S. (2000). Deutsche Intonation und GToBI. Linguistische Berichte, 191, 267-298.

4. Kohler, K. (1991). A model of German intonation. Arbeits­berichte des Instituts für Phonetik der Universität Kiel (AIPUK), 25. Kiel: IPDS, 295-368.