Relationships Between Hearing and Phonation

A paper by Alfred Tomatis published in the Annales des Télécommunications (volume II, nº 7-8, July–August 1956), Cahiers d’Acoustique nº 74, under the title “Relations entre l’audition et la phonation”. A communication relating to the work of the Groupement des Acousticiens de Langue Française (G.A.L.F.). Manuscript received 21 March 1956. In twenty-five pages with seventeen figures, the author sets out the theoretical framework of his work: the leading ear, transcerebral transfer and stammering, the occupational deafness of singers in four stages, auditory and vocal scotomata, auditory selectivity according to language (Italian, French, Russian), and a conclusion on objective audiometry through hearing–phonation feedback.

by Alfred Tomatis
Doctor of Medicine

Cahiers d’Acoustique nº 74, Annales des Télécommunications, vol. II, nº 7-8, July–August 1956.
A communication relating to the work of the Groupement des Acousticiens de Langue Française (G.A.L.F.). Manuscript received 21 March 1956.

ABSTRACT

Objective audiometry — as we practise it — derives from the relationships that link the hearing–phonation circuit. Any disturbance of this circuit is immediately detectable through the perturbations it provokes, either in the rhythm or in the timbre.

PLAN
Paragraph 1 — INTRODUCTION
Paragraph 2 — DISTURBANCES OF RHYTHM: the leading ear
Paragraph 3 — DISTURBANCES OF TIMBRE:
a) The occupational deafness of singers
b) Auditory and vocal scotomata
Paragraph 4 — AUDITORY SELECTIVITY
Paragraph 5 — CONCLUSION: OBJECTIVE AUDIOMETRY

I. — Introduction

The relationships that link hearing and phonation are so intertwined that the latter could not subsist without the existence of hearing, unless one resorts to the artifice known as re-education.

Certainly — at first glance — this seems self-evident. However, as soon as one moves away from the typical case of the deaf-mute, the elements of this association become less obvious and call for a more precise analysis.

In the course of this exposition, we shall see that these relationships are so closely interwoven in the phonation–hearing direction that they constitute a true circuit — and that any rupture, any tear, any anomaly (however slight) encountered in this circuit is quickly detectable:

either because it entails a disturbance of rhythm — that is, an impediment to the normal flow of discourse;
or because it reveals a modification of timbre — that is, of the manner in which this flow occurs.

II. — Disturbances of rhythm: the leading ear

In an earlier work, we demonstrated the existence of an auricular dominance in the acts of phonation: there exists a “leading ear”, just as there exists a leading eye in every individual.

This logical conclusion proved easy to verify — and it was from the study of disorders of phonation in voice professionals that this suggestion emerged, while we were investigating the characteristics of the musical ear.

This leading ear is always situated on the side of the leading eye — therefore generally on the right in a right-handed person, on the left in a left-handed person.

It can be readily demonstrated with the aid of an apparatus that is simple to build, composed of a microphone, an amplifier and a pair of headphones. The subject sings in front of the microphone and listens to himself in the headphones; one can — at will — suppress the monitoring of one ear or the other by means of a switch that removes one of the earpieces from the circuit (the one that remains in operation stays in parallel with a resistance of the same impedance as the eliminated earpiece).

We then observe that:

if the subject can monitor himself with both earpieces, he sings normally;
if the left ear is suppressed (the right having been identified as the leading ear), practically no change is observed in the emission;
but if the subject’s monitoring is limited to the left ear, an immediate modification of rhythm is observed, a considerable slowing of the sounds — at the same time as the voice changes timbre, becomes flat, white and loses its purity.

An experimental result of the same order is obtained if one disturbs the leading hearing no longer by means of the small electronic setup described above, but simply by inducing — for a few minutes — a dazzling caused by white noise.

The hearing of “musicians” — in the broadest sense of the term, that is, of persons having the faculty of hearing and of reproducing purely — presents an identical trace for all in the recording of auditory thresholds.

[Fig. 1 — Typical curve of the musical ear: a progressive rise between 500 c/s and 2000 c/s, with a level difference of the order of 10-20 dB.]

If this curve becomes disjointed, two phenomena appear:

1° When the disjunction occurs between 1000 and 2000 c/s, the subject hears purely, but sings out of tune. He may sometimes become aware of his defect and correct his lapse from purity (Fig. 2).

2° When the disjunction occurs between 500 and 1000 c/s, the hearing above these frequencies being intact (Fig. 3), the subject has lost his musical ear for listening — that is, he hears effectively, but when another person sings out of tune, he himself continues to sing in tune: a phenomenon that is paradoxical in appearance.

3° Finally, if the disjunction affects the whole curve and the latter no longer has an ascending threshold but takes on the shape of a saw (Fig. 4), no character of musicality is found in the person examined. He hears and emits out of tune.

In sum, everything happens as though there existed, audiometrically, a global musical ear — which can be broken down into a receptive musical ear and an expressive musical ear. But the dominant fact: these characteristics of the musical ear have value only when applied to the leading ear.

When one considers the spoken voice — and no longer the sung voice — under identical experimental conditions, we record responses that are even more precise. Thus, upon suppression of the leading ear, one notes — besides the immediate modification of timbre — disturbances of rhythm that are more or less marked and variable according to the person examined, but specific and always identical for the same subject. One can then observe a whole range of rhythmic anomalies — from simple mumbling to the most severe stammering.

This is a considerable source of study and a sound theoretical hypothesis concerning the pathogenesis of disorders of phonation — and in particular of stammering.

It only remains to take one step to confirm this hypothesis — by examining the hearing of persons affected by disorders of phonation, in particular stammerers. This is how we proceeded systematically and we now possess several hundred audiometric observations. We reproduce here a few results that we can divide into three groups (Fig. 5, 6, 7).

The majority — at least 90% — corresponds to subjects who are hypoacousic in the leading ear. As can be observed, this is only a relative hypoacusis, almost always unknown to the subject himself and detectable only with the audiometer.

However, this hypoacusis is sufficient to obtain experimentally — by partially suppressing the leading ear — an identical result. As though the slightest hypoacusis of the leading ear were enough to eliminate it from the circuit; the subject immediately adopts the facilitating solution offered to him by the opposite ear — which then benefits from a slight relative hyperacusis, but does not thereby become the leading ear.

We then thought that we were in the presence of a profound modification of the hearing–phonation circuit — and that this disturbance might give us the explanation for the whole set of rhythmic disorders. We can easily demonstrate this anomaly on two very simple diagrams.

Normally, the hearing–phonation circuit follows the itinerary below (Fig. 8):

the leading ear (which we shall take to be the right one — to simplify the exposition);
the right auditory centre;
the left auditory centre;
the left motor centre;
the musculature of phonation;
and the air pathway from mouth to leading ear.

[Fig. 8 — Normal hearing–phonation circuit in a right-handed subject: right ear → left auditory centre → organs of phonation.]

If — for any reason — the leading ear is suppressed, the opposite ear (the left ear in our example) becomes the input pathway of a new circuit, comprising the following stages (Fig. 9): left ear → right auditory centre → left auditory centre → left motor centre → musculature of phonation → mouth-to-left-ear pathway.

[Fig. 9 — Hearing–phonation circuit in a right-handed subject who has lost his leading ear: the “transcerebral transfer” is noticeable.]

We observe that, in this second pathway — more complex — there appears immediately a very considerable element of delay, which we have called “transcerebral transfer”. We have been able to measure this transfer: it can oscillate between 1/5 and 1/40 of a second depending on the person, but remains specific to each individual.

When the duration of this transfer lies between 1/10 and 1/20 of a second, with a maximum at 1/15 of a second, the subject is always a stammerer.

It can be seen, therefore, that not all persons are necessarily stammerers when their leading hearing is impaired. Two conditions prove indispensable: the loss of the leading hearing and a transcerebral transfer of the order of 1/15 of a second.

Now, 1/15 of a second corresponds roughly to the average duration of the French syllable. One then better understands — on the one hand — the necessity of redoubling the syllable to make up for this delay; and on the other hand — the phenomenon of repetitions, which escapes the control of the left cortex.

This value of 1/15 of a second — almost specific to stammering — explains the disappearance of stammering when a slowing of speech is imposed: either artificially (by imposing a bradylalia), or normally in all forms of language that lengthen the rhythm in duration — as is the case in the sung phrase.

It is also in the constancy of this value of 1/15 of a second that one can perceive why the subject stammers in French and not in English — the average value of the English syllable being of the order of 1/20 of a second.

Clinically, the acute disorders of phonation — encountered in subjects affected by an otitis involving the leading ear — reinforce this hypothesis. We have personally observed two significant stammerings in the course of otitis — disorders that went on to disappear as the function of the leading ear was restored.

The most important element — if not the proof — that led us to this hypothesis is the almost immediate disappearance of all phonatory disorders as soon as the normal circuit is restored. We regularly take advantage of this with success in the treatment of stammering.

Alongside these cases linked to a relative hypoacusis — representing 90% of cases — there exist a certain number of cases that simple audiometry does not allow to be detected, but which present a marked disturbance of auditory selectivity (to which we shall return a little further on). Finally, a third group brings together subjects whose right-handedness is not obvious — as in the ambidextrous; the leading ear is then less defined.

III. — Disturbances of timbre

a) The occupational deafness of singers

It is again to voice professionals — and particularly to singers — that we owe the idea of the possibility of a sound auto-traumatism, after having quantitatively analysed the voice of all the singers examined.

The magnitude of the sound energy they are able to develop did not fail to surprise us — all the more so as we had set out from the classical, but false, data limiting the intensity maxima to the order of 80 dB. Yet, at the distance we took as a reference, we readily encountered 100, 110, even 120 dB.

It is logical to think that a person subjected to such an intensity — for several hours a day — may, after a longer or shorter time, undergo the onset of a traumatic deafness.

We report here a few typical cases, which we shall compare with workers labouring beside aircraft engines for an equivalent length of time.

Both — the ones and the others — can illustrate the four stages of occupational deafness (Fig. 10, 11, 12 and 13).

We observe that, in these singers, there sets in a deafness of the occupational type, starting at the frequency of 4000 c/s and spreading towards the high sounds, then the low — exactly as in subjects exposed to noise.

In other words — and we particularly insist on this point — singers destroy their hearing through their own sound intensity; a phenomenon whose consequences are very serious.

b) Auditory scotomata and vocal scotomata

The consequences are serious, because this auditory loss — affecting selectively the high sounds — translates into a V-shaped scotoma, which becomes accentuated, as we are accustomed to observe in occupational deafness — while, moreover, voice disorders appear.

To identify these latter disorders, we carried out a spectral analysis by sweeping a cathode-ray tube, describing on the abscissa the frequencies and on the ordinate their relative intensity. Very quickly, we observed a fundamental phenomenon: the auditory scotoma translates into the appearance of a scotoma in the vocal spectrum.

We can conclude from this that the destruction of the voice is not linked — as is believed — to wear, to fatigue, to destruction of the larynx, but to the diminution of the auditory field. The phenomena of laryngeal suffering are very secondary.

For the singer to be able to obtain the high resonance he ceaselessly seeks, he imperatively needs a perfect hearing of the band extending above 2000 c/s. Without this possibility, his voice “passes to the throat” — and the so-called laryngeal sounds are forced and pushed. At the outset, the singer visits the whole set of resonance possibilities — that is, standing waves easy to feed without considerable muscular energy. Throat sounds — with strong laryngeal supports — require a considerable physical expenditure and prove traumatising for the larynx.

The progressive loss of hearing of high sounds entails disorders of emission that are all the more rapid as the register imposes the use of high ranges. Thus tenors are affected first — as soon as the scotoma rises above 2000 c/s, the singer’s career is seriously compromised. It is known, moreover, that a long voice is usually lower. It is nonetheless less rich in high harmonics — it is whiter.

Without attaching ourselves especially to singers, one can observe clinically that the voice deteriorates as presbyacusis progresses — that is, as ageing advances.

To sum up, one can say that the subject emits only the sounds he is capable of hearing.

IV. — Auditory selectivity

This last conclusion is still too broad and deserves that we dwell on it. While a person no longer reproduces the sounds he does not hear, he does not for all that reproduce all those he hears.

This is why we have studied what we have named auditory selectivity — that is, the faculty the ear has of perceiving the variations of frequency within the sound spectrum and of situating the direction of this variation.

We made use of the following examination procedures:

either by passing sounds going from high to low — and asking the person from what point the sound changes;
or by sending two noises of variable spacing and of different pitches;
better still — by offering the subject (by means of a series of filters) the choice of setting for himself the preferred mode of listening.

We obtained results that are surprising as to their theoretical importance. In the domain of auditory selectivity, there exists a well-defined ear for tenors, for baritones and basses — from which emerges a theory of registers confirming the earlier results.

What is more, there exists an ethnic hearing — we unfortunately cannot expand on this here, but we hope to return to it soon.

A few examples suffice to show the importance of these phenomena.

[Fig. 14 — Selectivity of the Italian ear: the passband falls between 2000 c/s and 4000 c/s.]

The Italian ear is a very “poor” ear. Its selectivity falls between 2000 and 4000 c/s (fig. 14). It is nil between 1000 and 2000 c/s — whereas the French ear is, on the contrary, limited between 1000 and 2000 c/s (fig. 15). Experimentally, we shall see the consequences. For example: the extraordinary appearance of the nasals in relation to the French ear.

[Fig. 15 — Selectivity band of the French-type ear, limited between 1000 and 2000 c/s.]

The Russians, on the other hand, have a very spread-out selectivity — with a greater affinity towards the low frequencies (fig. 16). Their voice is ample and warm. What is more, this very extensive auditory band — unlike that of the French and the Italians — allows them to pass all the consonances and — consequently — foreign sounds. It is known with what ease the Russians learn foreign languages. This phenomenon results simply from their great auditory bandwidth.

[Fig. 16 — Selective field of the Russian ear, extending from low sounds to extremely high ones.]

V. — Conclusion: objective audiometry

From these theoretical and experimental data, one can draw considerable practical elements.

For more than a year, we have been studying objective audiometry — with a real participation of the subject examined, without having to concern ourselves with his responses. It rests exclusively on the experimental findings above. Here is how we proceed:

The subject is placed in front of a microphone (M), as figure 17 shows.

[Fig. 17 — The whole of the setup permitting objective audiometry: M = microphone, An = analyser, Am = amplifier, E = earpiece; low-pass, band-pass and high-pass filters.]

This microphone is connected to an analyser (An) with very rapid automatic sweep, with a very extended band thanks to the superposition of five lines permitting the spectral analysis of sound from 0 to 10,000 c/s or 20,000 c/s over 50 cm.

This same microphone then makes it possible to drive the amplifier (Am) — and several pathways are accessible to us:

1° by pathway I, the person immediately receives his normal voice and can thus monitor himself;
2° by pathway II, the amplified voice can — as chosen — pass either through a low-pass filter variable from zero to infinity; or through a high-pass filter variable from infinity to zero; or through a band-pass filter variable in spread and in pitch;
3° by pathway III, the voice is finally mixed with a background noise — of the white-noise type — that one can dose in intensity (in decibels) and — what is more — limit in its dimension of spread, exactly within the passages of the interplay of the low-pass, high-pass and band-pass filters.

We thus obtain the following results:

1° by pathway I, the person speaks normally in front of the microphone, monitoring himself with the earpieces. We then obtain an envelope spectrum — and we have seen that experimentally this spectrum falls within the envelope curve of the person’s auditory spectrum.

2° by pathway II, using the low-pass filter, we cut off all the high sounds at a variable height and observe a compression of their action within the imposed limits. Likewise with the high-pass filter. In both cases, one observes that, for certain zones, the person can no longer saturate the affected bands. We then know that the zones of passband that one can reduce or open at will — and that one can shift along the trace of the normal auditory spectrum — construct the vocal spectrum, or the same passband imposed on listening. Each time a gap is revealed in the sound spectrum of the cathode-ray tube, we find an auditory gap. The result amply confirms the previous one.

3° by pathway III, one can obtain another test with the aid of a white-noise generator. One sends — into the subject’s listening — a progressive background noise. At a certain moment, one observes that the vocal spectrum decreases in intensity, and this globally for all frequencies. We have just crossed the threshold of hearing. From then on the subject speaks louder, but still presents a vocal spectrum of identical trace — that is, without modification in the spectrum of the injected white noise, concerning for example the 0-1000 c/s band. One then widens the portion of the injected white-noise spectrum, concerning for example the 0-1000 c/s band. The person can then no longer effect a translation towards the high sounds — he cannot speak louder and must change timbre. This is a positive Lombard phenomenon.

We can then progressively widen our injected spectrum towards the high frequencies. Beyond a certain limit — for example 4000 c/s — the person is no longer able to go further. At that moment, we find ourselves at the upper limit of his hearing of high sounds.

One can thus — without the subject’s knowledge — ascertain the extent of his hearing and carry out a true objective audiometry, through hearing–phonation feedback.

One sees, therefore — in these few lines of exposition — the essential role that hearing plays over phonation, as well as the considerable consequences of these relationships, which we could, in this article, only summarise.

Manuscript received 21 March 1956.

Source: Tomatis A., “Relations entre l’audition et la phonation”, Annales des Télécommunications, vol. II, nº 7-8, July–August 1956, Cahiers d’Acoustique nº 74, pp. 133-156. A communication relating to the work of the Groupement des Acousticiens de Langue Française (G.A.L.F.). Manuscript received 21 March 1956. Digitised document from the personal archives of Alfred Tomatis.

Original document — historical PDF facsimile (direct download).

Archive note: text reconstructed from the Polish version in our archives, the 1956 French original not being in our possession. To be checked against the primary source.