Skip to Content

Department of Linguistics

SPEECH ACOUSTICS

Consonant Acoustics

The Acoustic Characteristics of Nasals

Felicity Cox

The phonetic symbols used on these pages are know as the ANDOSL machine-readable symbols. Click here for information on how to read these symbols.

Introduction

The acoustical aspects associated with nasality have been researched in two major ways.

1. In order to develop an understanding of the acoustic and perceptual characteristics of the nasal consonants.

2. In order to study the effects that nasalisation has on vowels and other consonants.

Brief Physiology

Different speakers achieve auditorily similar results by physiologically different means. However velum lowering is primarily effected by palatoglossus activity (palatoglossus connects the tongue and the velum) and palatopharyngeus (connecting the velum and the thyroid cartilage).

In nasal voice, when the velum is lowered due to palatoglossus activity there is often the added perception of velarisation due to tongue being raised and retracted.

When the palatopharyngeus contracts it influences the mode of vibration of the vocal folds as the larynx may be pulled up thereby stretching the folds causing an increase in F0. Raising the larynx also has the effect of shortening the vocal tract and therefore changing the resonatory characteristics.

In reality, the velum does not move like a hinged trap door but is only the anterior part of a complex velopharyngeal valve which functions as a circular sphincter.

Palatal elevation occurs but there is also movement of the back wall of the pharynx towards the front and in some speakers a structure known as Passavant's pad, a muscular bar or cushion on the back wall of the nasopharynx at the point where it is touched by the lifted velum.

(Velum closure is primarily effected by the palatal tensor, the palatal levator, the superior pharyngeal constrictor)

Nasality as an Auditory Phenomenon

Nasality is above all else an auditory phenomenon and not primarily an articulatory one able to be specified in terms of the position of the velum.

Nasality is a special condition of resonance. The auditory impression of nasality is the result of resonance in a cul-de-sac resonator.

A cul-de-sac resonator is a chamber opening off from the passageway through which a sound is resonated and sent to the outer air. In the production of nasal consonants, the oral tract becomes the cul-de-sac. In nasalised segments, the nasal cavity is the side chamber.

In normal speech production the velum does not completely close off the pharyngeal passage to the nose, and the degree of velic opening is variable. It has been found that the perception of a sound as nasalised will depend on the ratio of the sizes of the two openings into the nasal cavity and the oral cavity. When the nasal port is large relative to the oral port then nasality will be perceived.

The type of phoneme that is produced during speech will affect the height of the velum. The following is a progression from highest to lowest:

  • Voiceless stops
  • Voiced stops
  • Voiceless fricatives
  • Voiced fricatives
  • Oral close vowels
  • Oral open vowels
  • Nasalised close vowels
  • Nasalised open vowels
  • Nasal segments

Acoustic Characteristics of Nasality

The acoustic result of adding the nasal tract to the oral tract can be discussed under three main headings

  1. The resonatory characteristics of the nasal tract
  2. The effect of nasal resonance on oral tract resonance.
  3. Nasal consonants

1. The Resonatory Characteristics of the Nasal Tract

a. Like the oral tract, the nasal tract has its own resonant frequencies or formants, the nasal formants.

The most commonly reported nasal formants occur at 300Hz, 1kHz, 2.2 kHz, 2.9kHz, 4kHz. Labelled N1,N2,N3 etc. However there is considerable variation from speaker to speaker due to the large variation in side and shape of the nasal cavity.

b. The bandwidths of the nasal formants are about 300Hz for the lowest formant increasing to 1000-Hz for the higher formants thus the are highly damped.

c. Antiresonances enter whenever there is a side branch in the main acoustic pathway. An antiresonance or zero serves to decrease the spectral energy at specific frequencies by absorbing the sound at or near the antiresonant frequencies. These cumulatively have the effect of reducing the total amplitude of the sound generated.

2. The Effect of the Nasal Resonance on the Vocal Tract ie. Nasalisation

a. The most general effect of adding nasal resonance to oral resonance is an overall loss of power. This attenuation is directly due to the introduction of antiresonances by the side chamber resonator which absorbs energy particularly at higher frequencies. When an oral sound becomes nasalised, the effect of adding the nasal side chamber is to introduce nasal resonances and antiresonances which interact with the existing oral resonances.

The antiresonances decrease energy at specific frequencies thus reducing and sometimes eliminating some low intensity formants from the acoustic signal.

b. The general attenuation of the signal is reflected in the broadening of all the formant bandwidths which flattens the spectral peaks.

This flattening is also a function of absorption

c. More specific indicators of nasality are

I. A marked drop in the intensity of the first formant

due to: the first nasal formant boosts the intensity of the lower harmonics, below F1 and another reinforces the harmonics above F1. An antiresonance between serves to lower the intensity of F1.

II. F1 appears to be slightly raised

due to: The presence of a low frequency antiresonance just below F1 tends to make the spectral peak in the region of F1 appear between 50 and 100Hz higher in frequency than it would normally be.

III. There is a drop in intensity of all the higher formants

due to: Antiresonances.

IV. Extra resonances are present in the spectrum at unexpected frequencies.

due to: Nasal cavity resonances

V. The degree of nasalisation that we hear depends on the amount of acoustical impedance in the oral and nasal cavities. Given the same velopharyngeal opening, a greater degree of oral impedance leads to greater nasality perceived.

Low vowels have little oral cavity acoustical impedance therefore the amount of velopharyngeal opening must be large before these sounds are perceived as excessively nasal.

High vowels are produced with larger amounts of acoustical impedance therefore smaller velopharyngeal port openings will result in these sounds being perceived as nasal vowels.

3. Nasal Consonants

Nasal consonants are voiced sounds produced with a complete closure within the oral tract and with the velum lowered to such an extent that the nasal cavity becomes the major resonating tube with the oral cavity as the side branch allowing no escape of air through the mouth.

Nasals are referred to as voiced frictionless continuants and in this sense they resemble vowels. Nasal consonants are characterised by a closure phase, with the resulting sound referred to as the nasal murmur, and transitions to and from adjacent vowels.

Nasals are lower in intensity than vowels due to damping.

The nasal murmur comprises the nasal resonances and the mouth cavity antiresonances. The mouth cavity resonances that are paired with the antiresonances are not considered as important in the spectrum as they are damped by the complete closure in the oral tract. (This is in contrast with the nasal vowels or nasalised sounds which contain important information from the oral resonances as well as the paired nasal resonance/antiresonance. The oral resonances are present as the open mouth allows radiation to the air.)

The murmur is characterised by N1 at 250 -300Hz of high intensity compared with the rest of the spectrum

The frequency location of the oral antiresonances is primarily determined by the length of the side branch. As length increases, the antiresonance moves down in frequency.

The place of articulation of nasals is cued primarily by the characteristics of the murmur and the transitions.

/m/ Bilabial Nasal Stop

tongue anticipates or retains the position of adjacent vowels

voiced except when partially devoiced by a preceding voiceless consonant

main allophone other than labial is the labiodental which occurs when followed by /f,v/ eg "nymph"

low frequency, high intensity N1 250Hz

low frequency antiresonance above 500Hz.

oral resonance F2 about 1000Hz which may coincide with the first antiresonance and so be weakened

The oral resonance may show some continuity with the vowel's second formant. The oral resonance is weak in the murmur itself but as the closure is released it will increase substantially as it moves into the vowel F2 transition.

/n/ Alveolar Nasal Stop

lip shape dependent on adjacent vowels e.g. "coon", "keen"

partial devoicing occurs following voiceless consonants

allophones labiodental when followed by /f,v/ ,"infant"

dental before dentals /T,D/ ,"tenth"

post alveolar contact before /r/ "Henry"

readily assimilates in bilabial or velar contexts

eg "ten people", "ten girls"

low frequency, high intensity N1 250Hz

N2 at about 800Hz which may be weakened by the first antiresonance.

the first antiresonance is higher than that of /m/ as a result of shortened oral tract.

an oral resonance at about 1.4kHz

Lack of energy on spectrograms of /n/ above 250Hz up to at least 2kHz is usual.

The slope of the oral resonance into the vowel transition is an important place cue.

/N/ Velar Nasal Stop

point of closure depends on following vowel. More fronted for "sing" than "sung"

preceding vowel lip position "sing"- spread, "song" - neutral

The velar nasal does not occur in initial position in English

low frequency, high intensity N1 250Hz

the first antiresonance is highest at above 3kHz. Very little side branch

The slope of the oral resonance into the vowel transition is an important place cue.

The problem with analysing spectrograms of nasal consonants is that there is energy in the spectrum from a number of different sources and it is very difficult to interpret where the energy originates: whether nasal oral or the result of shifting from antiresonances.