A Field Manual for Acoustic Phonetics

Joan L. G. Baart

• Publisher: SIL International
• Year: 2018
• ISBN:
  9781556714290
• Topic: Filología
• Subtopic: Fonética y fonología

This book provides a practical and easy-to-understand introduction to acoustic speech analysis, primarily aimed at those involved in linguistic analysis and description in the field and at those preparing for such fieldwork. It explains commonly used methods for displaying aspects of a speech wave, such as waveform graphs, spectra, spectrograms, fundamental frequency graphs (pitch graphs), and intensity graphs. It illustrates how the results of acoustic analysis can be interpreted and used to improve the objectivity, accuracy and precision of phonetic descriptions of speech sounds. The book assumes basic knowledge of articulatory phonetics. It can be used to teach introductory courses in acoustic phonetics at the undergraduate level.

5

Prosody

5.1 Introduction

5.1.1 Prosody

In the previous chapters we have been concerned mainly with the properties of vowels and consonants. In this chapter we will look at acoustic aspects of so-called prosodic features. Prosodic features are features that cannot be derived from the intrinsic properties of the vowels and consonants that make up an utterance. They typically stretch out over domains that are wider than a single segment. Examples of prosodic features are stress, tone, and intonation.

The word prosody is derived from Greek prosôidía ‘song sung to instrumental music’ (ôidê = ‘song, ode’, pros = ‘with’). Early on in Greek, the word came to be used to refer to unwritten features of the pronunciation of words, including pitch and length, and when written marks were introduced to indicate such features, these marks were also called “prosodies.” By the second century A.D., the word prosody was already being used to refer quite generally to features of pronunciation that were not expressed in the segmental succession of vowels and consonants (Couper-Kuhlen 1986:1).

Our use of prosody will be more or less synonymous with the term suprasegmentals, which used to be more customary among linguists working in the American structuralist tradition. Couper-Kuhlen (1986:2) is careful to point out that the terms prosody and suprasegmentals are not wholly synonymous. Use of one rather than the other carries with it special connotations and often serves to indicate where one’s linguistic loyalties are. For a good discussion of these differences, the reader is referred to Couper-Kuhlen (1986).

5.1.2 Phonetic domains of prosody

Strictly speaking, terms such as stress, tone, and intonation refer to linguistic (emic) concepts, not to physical phenomena that can be directly observed and measured. High tones and low tones, final stress and initial stress, rising and falling intonation contours, etc., are elements that play a role in the description of a language system. A language system is something that is intimately known by the speakers of that language, but it is not something that can be directly observed.

Speech, on the other hand, is something that occurs in the observable, physical world: somebody’s vocal folds vibrate, a sound wave is propagated through the air, and registered by somebody else’s ear. The question to ask, therefore, is: how are the distinctions that are made by a language system reflected in speech? Conversely: what do the observable properties of speech tell us about the system of a language?

Phonetics, the scientific study of speech, is often described as consisting of three domains: articulatory phonetics studies speech from the perspective of the speaker (how is it produced?); auditory phonetics studies speech from the perspective of the listener (how is it perceived?); acoustic phonetics studies speech as it travels through the air in the shape of sound waves.

Linguistic distinctions of stress, tone, and intonation are reflected in speech in different configurations of pitch, length, and loudness. The latter three terms actually refer to the domain of auditory phonetics. Pitch, for instance, is a term that refers to an auditory (perceptual) phenomenon. It is associated in the articulatory domain with the rate of vibration of the vocal folds. Both perceived pitch and vocal fold vibration are associated in the acoustic domain with fundamental frequency. The vocal folds vibrate at a certain rate; they produce a sound wave that has a certain fundamental frequency; this wave is registered by a human ear and perceived as a sound with a certain pitch.

In the same way, loudness is associated with the amount of energy (physical effort) expended in articulation, and with intensity in the acoustic domain.

Perceived length correlates with the timing of articulatory movements, and with acoustic duration. This state of affairs is represented in table 5.1 (adapted from Couper-Kuhlen 1986:7):

Table 5.1 Phonetic domains of prosody

It is good to note that relations between parameters occurring on the same row in the chart are not entirely simple. It is roughly true, to give an example, that an increase of the rate of vibration of the vocal folds leads to an increased fundamental frequency and to a higher perceived pitch. However, there are a number of distorting factors that we have to take into account. First, there is a perceptual threshold for fundamental frequency differences, such that differences smaller than the threshold are not noticeable to the human ear. Small changes in fundamental frequency may thus not always result in changes in perceived pitch.

Second, we have to take into account that scaling may be different in the different domains. A fundamental frequency difference of, say, 30 Hz is perceived as a large pitch difference if the frequencies involved are low (such as the difference between 90 and 120 Hz), and as a small pitch difference if the frequencies involved are much higher (such as the difference between 360 and 390 Hz).

Third, there may be interactions with other parameters. For instance, an increase in intensity may have an effect on the perception of pitch (resulting e.g. in higher perceived pitch).

In the remainder of this chapter, we will first look in more detail at the acoustic parameters of prosody, which are: fundamental frequency, duration, and intensity. Following that, we will discuss the linguistic features of stress, tone, and intonation, and how they may be studied from an acoustic point of view.

5.2 Fundamental frequency (F0)

A periodic speech wave consists of a sequence of more-or-less similar oscillation patterns (wave cycles). The rate at which these wave cycles are repeated is called the fundamental frequency of the speech wave, which is abbreviated as F0. Fundamental frequency can be expressed in cycles per second, for which hertz is commonly used as a language-independent term (abbreviated Hz).

Periodic speech waves are generated by means of vocal fold vibration (the repetitive opening and closing of the vocal folds), and the fundamental frequency of speech waves is associated with the rate of vibration of the vocal folds.

5.2.1 Extracting fundamental frequency

The fundamental frequency of a speech wave is reflected in several acoustic parameters. In a speech wave graph, for example, we can identify cycles of a wave and determine how much time it takes to complete one cycle. When we know the duration of a cycle of the speech wave (its period), we can calculate how many such cycles would fit into a second, and in this way we can determine the fundamental frequency of the wave at that point.

In a spectrum or a narrow-band spectrogram one can often identify separate harmonics (sine wave components into which a complex periodic wave can be analyzed; see section 3.2.4). Th...

Table of contents

• Introduction
• Speech Waves
• Sound Spectrograms and Spectra
• Voice and Aspiration
• Prosody
• References