University of St Andrews
ML2001, Structure of Language (1998/99).
First Essay: (Professor Press)
"Discuss the
three
principal branches of phonetics, explaining how each of them
analyses/classifies speech sounds, how they differ from each other, and
how they are interrelated."
Christian Asseburg, 29
October 1998
Please state your source when quoting.
Subdivisions of Phonetics
Phonetics is a science that deals with speech sounds. As a spoken
conversation is the most frequent manner of communication between human
beings, an urge to find out more about the exact mechanisms that speech
requires is very natural.
The act of spoken communication always involves at least the following
three stages. After the speaker has decided which sounds to employ in
order to convey his message, he has to produce these sounds using his
vocal organs. Next, the sounds need to travel through the air and cover
the distance between speaker and listener. Finally, the listener receives
sensory input at his ears and has to recover the intended message from
those sensory stimuli.
Source: J. D. O'Connor: Phonetics. p. 10.
Each of these three stages allows for distinct scientific
investigations. It is thus along these lines that phonetics may be broken
up into the principal branches of articulatory, acoustic, and auditory
phonetics.
Characterisation of Speech Sounds
There exists a long tradition of research into the components of speech,
dating as far back as to the Ancient Greeks. It is the articulatory branch
of phonetics that has geared the greatest interest, since every competent
speaker is at least vaguely familiar with the organs involved in speech
production, whereas measuring the physical properties of sound waves
requires sophisticated physical instruments, and the impact of an auditory
stimulus on the ear-brain-system is even more elusive to a scientific
approach.
Articulatory Phonetics
Producing speech means employing one's vocal organs to modify an egressive
airstream created by the lungs. In normal breathing, there are no
constrictions to this air flow. However, the organs that form part of the
vocal tract, most notably the larynx, velum, tongue, and lips, are subject
to conscious muscular activity which modifies their positions and thereby
the shape of the cavities through which the air needs to pass. In
articulatory phonetics, speech sounds are classified by the relative
positions of the vocal organs and by their effects on the air flow.
The major categories into which articulatory phonetics divides speech
sounds are stops, fricatives, and vowels. These terms refer to the degree
of constriction that is imposed on the airstream. In stops, also called
plosives, no air at all may pass out of the vocal tract because the
passage is completely blocked off at some point (eg. [b], [t]). In
fricatives, there is a considerable narrowing of the passage which results
in audible friction (eg. [x], [z]). Finally, if the vocal passage is
almost unrestricted, the resulting sound is called a vowel (eg. [a],
[o]).
Voice
Another important property of the vocal tract is the position and shape of
the vocal chords, which are located inside the larynx. In normal
breathing, these are relaxed and offer a free passage of air. However,
the vocal chords can be tensed to a certain degree, which causes them to
vibrate in the egressive airstream. Those vibrations result in a musical
sound, which is called voice. Vowels are usually voiced, i.e. the vocal
chords vibrate while the air passes through larynx and mouth. Fricatives
may be voiced or voiceless. In stops, the air, by definition, ceases to
flow for a short moment, and naturally, the vocal chords do not vibrate.
However, stops are still considered to be voiced when the vocal chords
vibrate just before or just after the stop, and voiceless otherwise. By
tensing the vocal chords to a higher degree than that required for them to
vibrate, the air passage can also be completely closed off at the larynx.
The resulting speech sound is called a glottal stop.
Stops and Fricatives
Further classifications that are used in articulatory phonetics to
distinguish between various stops and fricatives refer to the place of
articulation. For stops, one particular place in the vocal tract is highly
relevant, namely that location at which the passage is blocked.
The tongue, being the most agile vocal organ, is usually the active agent
in constricting the air flow. To specify the exact place of articulation,
it is helpful to name the other part of the vocal tract which serves in
passive opposition to the tongue. For example, the sound [t] is referred
to as an alveolar stop because the tip of the tongue blocks the air
passage by touching the alveolar ridge. Other common places of
articulation are dental, palatal, and velar, but it may sometimes be found
necessary to specify the place of articulation even more exactly. The
lips, being positioned outside the oral cavity, can also be used to block
the air flow. The phonetician refers to the constriction caused by the
closing of the lips as bilabial, e.g. [b]. Sounds that are formed by the
lower lip touching the upper teeth are called labiodental, e.g. [f].
As fricatives are closely related to stops, varying only in the degree of
constriction imposed on the airstream, the above places of articulation
lend themselves equally well for describing fricatives.
Vowels
Sounds that require an unblocked passage of air, most prominently the
vowels, cannot be described in the same way as given above. The
characteristic differences between the vowels are the position and shape
of the tongue, on the one hand, as well as the shape of the lips. The
shape of the tongue is commonly given by the position of its highest
point, which may move in two dimensions. For front vowels, e.g. [i] and
[e], the highest point of the tongue is located close to the hard palate,
and for back vowels, it is closer to the velum, as in [u] and [o]. Also,
the highest point of the tongue may be quite high, as in [i] and [u], or
rather low, as in [a]. These qualities need to be seen on a gradual scale,
however, and many different shades of vowels are found in the languages of
the world. A neat graphical way of demonstrating the position of the
tongue is given by Jones's Trough.
The sound quality of a vowel is greatly affected by the shape of the lips.
A vowel pronounced with rounded lips is called a rounded vowel, whereas
unrounded vowels are those pronounced with spread lips.
The above methods for the classification of speech sounds may be extended
to cover other types of speech sounds, such as affricates, glides,
diphthongs, liquids. Furthermore, I have omitted several features of
speech sounds, for example nasalization and stress. A full coverage of all
elements involved in the production of speech sounds is given by the
alphabet that the International Phonetics Association recommends for the
notation of speech sounds, but explaining or even listing all of them
would take me beyond the scope of this essay.
Acoustic Phonetics
A relatively recent approach to phonetics is the study of the physical
properties of speech sounds. Because acoustic phonetics requires direct
access to the physics of sound waves as these travel through the air (or
any less common medium), it could only be established as a science in its
own right after acoustic physics provided the framework and appropriately
designed measuring instruments for more elaborate studies in the acoustics
of speech sounds.
The most prominent instrument involved in the physical quantification of
speech sounds nowadays is the acoustic spectrograph. This device analyses
the frequencies that overlap to create a uniform sound wave. From earlier
research into the acoustics of musical instruments, it is known that the
frequency we perceive as the pitch of a musical key is not the only
frequency contained within that sound, but many other frequencies, called
harmonics, overlap to make up one sound. It is the distribution and the
intensities of those additional frequencies that give each musical
instrument its characteristic quality.
Vocoids
From an acoustic point of view, it is convenient to begin with describing
sounds that have a musical quality. These sounds are called vocoids in
acoustic phonetics, and they comprise mainly the vowels, but also the
nasal sounds and some of the other sonorants.
Starting with the fundamental frequency of speech as created by the
glottal voice, the cavities of the vocal tract act as resonators or
filters to modify the harmonics of the emitted sound wave. On a
spectrogram, the fundamental frequency is clearly discernible, but one can
also distinguish several bands at higher frequencies which feature higher
intensities than the rest of the frequency spectrum. These bands are
called formants, and in order to facilitate referring to each of those,
they are numbered from the first frequency band above the fundamental
upwards, as in F1, F2, F3... Research has not been concluded yet, but it
seems that the frequencies of the first two formants suffice to give a
rather adequate classification of the vocoids. To quote an example from
Brosnahan, p. 94, the sound [i] may be classified as a vocoid with F1 at
250 Hz and F2 at 2500 Hz, contrasting with [e] with F1 at 400 Hz and F2 at
2000 Hz, and with [y] with F1 at 250 Hz and F2 at 2000 Hz. Care needs to
be taken in comparing speech sounds this way because the frequency levels
vary widely from speaker to speaker and are naturally higher in female
voices. Furthermore, such an analysis ignores the actual intensities of
the formants, focussing on their frequencies only, and for a full
description many other variations need to be taken into account.
Contoids
Acoustic phonetics classifies anything which is not sonorant in nature as
contoids. This group includes fricatives and affricates, which show as
noise on a spectrogram, as well as stops, which, technically speaking, do
not create any sound at all, but affect the preceding and following
sonorant sounds. It seems that research has centered on proper stops,
which may be analysed as silence surrounded by short glides in the
formants of neighbouring sonorant sounds. As such, it becomes obvious that
it is difficult to characterise the acoustic properties of a stop as they
will always depend on the qualities of the surrounding sounds.
Phoneticians have tried to explain exactly what kind of transitions are
required for a stop to be heard as [p], say, or [k]. For [d] as opposed to
[b], for example, O'Connor suggests that in both [d] and [b], the formant
F1 at the beginning of the next vocoid starts out from a little lower, in
order to glide to the frequency required for the following vocoid, and in
[d], F2 performs a transition from some high frequency towards its final
position, whereas for [b] this initial frequency for F2 is somewhat
lower.
Although some results such as the one above point the way for fruitful
theories, existing explanations have not yet been successful in explaining
satisfactorily the properties characteristic of a certain stop. Fricatives
and other speech sounds require even more research.
To conclude this short description of acoustic phonetics, it is as yet not
very obvious how to understand the spectrograms, which appropriately
reflect the properties of the acoustic waves corresponding to speech
sounds. Some progress has been made, but phoneticians are still quite
remote from being able to understand the incoming physical sounds in as
competent a way as the human ear does.
Auditory Phonetics
Auditory phonetics researches into the way the human ear and brain
perceive and analyse different speech sounds. Quite a lot is known about
the anatomy of the ear, but as it is difficult to obtain any objective
measurements from inside a subject's head, and even more so because the
brain processes involved in analysing speech are very intricate, this
branch of phonetics is arguably the least developed one.
Research into the physiognomy of the ear has shown that it consists of
several cavities which probably serve as resonance chambers as well as
filters. The tympanum, together with the three small bones of the middle
ear, acts as an amplifier in transmitting the incoming sound waves to the
oval window. The oval window forms an interface to the liquid-filled
cavity of the cochlea which contains the Corti organ with many sensitive
nerve cells. Current research suggests that this mechanism maps the
frequencies of incoming sound waves to different areas inside the
cochlea.
The impulses received by the auditory nerve cells travel from the ear to
the brain, where they are subject to complicated neural processing, which
is as yet almost inaccessible to scientific enquiry. Apart from electric
encephalograms, which measure differences in electric potential inside the
brain and which are not particularly helpful to the phonetician, the only
way to obtain information on auditory sensations that human beings
perceive upon hearing certain sound waves is by asking test subjects to
describe their particular sensations. Although such data are obviously
subjective, it seems one can classify speech sounds by using pairs of
contrasting adjectives, such as hissing and buzzing, dull and sharp, heavy
and light (Clark and Yallop, pp. 310), and agreement on whether a given
sound constitutes a vowel or a consonant is usually considerable.
To me, it seems as if auditory phonetics requires a lot more research
before one may properly call it a science, but at present phoneticians
lack accessible ways of measuring auditory qualities of sounds, and it may
be impossible at all to advance the knowledge in auditory phonetics
without breaking moral boundaries.
Relations Between the Three Branches of Phonetics
As noted earlier, the oldest research into phonetics has been in the field
of articulatory phonetics, and therefore many formerly articulatory terms
are also used to refer to related sounds in the other two subdivisions of
phonetics. Even today, it is the articulatory branch of phonetics that has
the greatest influence on phonetics as a whole, especially because no
complicated experimental setup is needed for initial investigation. This
prominence of articulatory phonetics becomes obvious as soon as one reads
any introductory textbook on phonetics.
Articulatory and Acoustic Phonetics
The relation between articulatory and acoustic phonetics is a rather
onesided one. Acoustic phonetics was dependent on articulatory phonetics
in its early days because the physical data collected by scientific
measurements seemed so unrelated to the known speech sounds at first that
known articulatory observations were welcomed as a guiding hand. Finally,
acoustic phonetics has taken off as a science in its own right, and it has
produced results that are in considerable agreement with articulatory
discoveries, particularly concerning the vowels. I would like to quote
from Ladefoged an experiment that highlights the relation between
resonance chambers in articulatory phonetics and the formant F1 in
acoustic phonetics.
"If you place your vocal organs in the positions for making
the vowels in each of heed, hid, head, had and then flick your
finger against your throat while holding a glottal stop, you will produce
a low pitched note for the word heed, and a slightly higher one
for each of the words hid, head, had."
P. Ladefoged: Elements of Acoustic Phonetics, p.
102
The dependency of a contoid on its surrounding vocoids is shown by the
following observation. In articulatory phonetics, it has been noted that
consonants in front of a high front vowel, such as [i], become
palatalized, and experiments by Brosnahan confirms that such contoids are
indeed different. For example, if one were to take the sound waves
corresponding to the sounds [sk] in [ski], and place them in front of the
sounds [ul] from [sku:l], and play the resulting sound pattern to a test
audience, most listeners will identify the heard sounds as [spu:l], and
not as [sku:l] (Brosnahan, pp. 78).
Articulatory and Auditory Phonetics
There is also considerable agreement between articulatory auditory
phonetics, which may seem obvious as verbal communication would be quite
futile otherwise. If asked to group certain speech sounds according to
their articulatory characteristics, and then to analyse the sounds with
respect to their auditory qualities, almost any speaker will produce
highly similar lists. For example, different nasal articulations
frequently result in being classified together as humming sounds.
Furthermore, when a linguist, or indeed any speaker who is attentive to
actual speech sounds, hears a sound that he cannot classify on its
auditory merits alone, trying to articulate a similar sound himself will
help him to establish the characteristics of that particular sound. And
anyone who has learnt a foreign language will certify that it is a lot
easier to pronounce new sounds, which he has to acquire in order to use
the new language because his native language does not use those sounds, if
he has managed to notice the auditory difference, and vice versa.
Acoustic and Auditory Phonetics
Strange enough, the correlations between auditory and acoustic phonetics
are not very obvious at all. When we listen to a speaker, we are hardly
ever aware of the particular acoustic qualities of his voice, and even if
we are, we would hardly be tempted to discuss any of the technical
properties mentioned above, such as formants. So the relation between
acoustic and auditory phonetics seems to be a very elusive one. Of course
there has to be some kind of relationship, as we purport to know that it
is the acoustic sound wave that transmits the message from the speaker to
the hearer.
One striking correlation between acoustic and auditory phonetics has been
confirmed by experiments. Often, when we hear speech in a noisy
environment, and when misunderstanding a word, the exact order of the
speech sounds gets confused. Brosnahan states that a temporal threshold of
hearing has been shown by experiment, which prevents a listener from
stating exactly which one of two speech sounds given in quick succession
is the first one.
Segmentation of Speech
One question which might have been dealt with at the outset is how do we
know where a speech sound ends within the constant flow of speech and
where the next one begins. The answer may seem quite obvious in the case
of stops, but when one considers the gradual changes that are required for
pronouncing glides, as well as the considerable amount of preparatory
positioning of the vocal organs before any given, well-defined sound may
actually be articulated clearly, it is no longer easy to see any clear
pattern. However, every competent speaker of any language is aware of
clearcut sound boundaries, and these boundaries are a further point of
high overlap between articulatory and auditory phonetics. Acoustic
phonetics, on the other hand, accurately reflects the gradual changes that
take place during speech production, and articulatory phonetics confirms
the gradual movements of the articulatory organs, as seen in realtime
X-ray photography.
Conclusion
Overall, the communicative system of speech seems to allow for a very high
degree of distortion, and speech tolerates quite massive noise overlay,
showing that it is at once very robust and suitable for interhuman
communication as well as highly redundant in its acoustic and semantic properties and
thereby difficult to analyse fully. When I consider the useful
applications of articulatory phonetics in instructing learners of foreign
languages or the impaired of hearing how to pronounce speech correctly,
and of acoustic phonetics in building machines that can produce synthetic
speech, phonetics seems to be a fruitful science, and auditory phonetics
may lead the way to a better understanding of the human brain, one of the
most intricate and least understood structures known to man.
On the other hand, there is considerable overlap between any of the three
major branches, as outlined above, which justifies the classification
of each of them as a subfield within phonetics.
Bibliography
L. F. Brosnahan, B. Malmberg: Introduction to
Phonetics. W. Heffer and Sons, Cambridge, 1970.
J. Clark, C. Yallop: An Introduction to Phonetics and
Phonology. Basil Blackwell, Oxford, 1990.
F. P. Dinneen: An Introduction to General
Linguistics. Holt, Rinehart, and Winston, New York, 1967.
V. Fromkin, R. Rodman: An Introduction to
Language. Harcourt Brace, Fort Worth, 1998.
C. F. Hockett: A Manual of Phonology. Waverly
Press, Baltimore, 1955.
The Principles of the International Phonetics Association
(IPA). University College London, London, 1949.
(ed.) I. Lehiste: Readings in Acoustic
Phonetics. MIT Press, Cambridge (Mass.), 1967.
B. Malmberg: Les Domaines de la
Phonétique. Presses Universitaires de France, Paris, 1971.
(ed.) B. Malmberg: Manual of Phonetics.
North-Holland, Amsterdam, 1968.
B. Malmberg: Phonetics. Dover, New York,
1963.
P. Ladefoged: Elements of Acoustic Phonetics.
University of Chicago Press, Chicago, 1962.
P. Ladefoged: Preliminaries to Linguistic
Phonetics. University of Chicago Press, Chicago, 1971.
J. D. O'Connor: Phonetics. Penguin,
Harmondsworth, 1973.
R. H. Robins: General Linguistics. Longman,
London, 1989.
A. Spencer: Phonology, Theory and Description.
Blackwell, Oxford, 1996.
Christian Asseburg,
last revision 29 October 1998.