ASA Lay Language Papers
162nd Acoustical Society of America Meeting

"Impossible" language sounds

Dan Brenner -- dbrenner@email.arizona.edu
Andrea Davis
Natasha Warner
Andrew Carnie
Muriel Fisher
Jessamyn Schertz
Michael Hammond
Diana Archangeli
University of Arizona
Tucson, AZ 85721

Popular version of paper 4aSCb1
Presented Thursday morning, November 3, 2011
162nd ASA Meeting, San Diego, Calif.

What do people do when their language tricks them into pronouncing an “impossible” sound?  Nasalized fricatives ([v~, x~, h~] etc.) involve making a sound like "v" in English "voice," but also letting air out the nose at the same time, somewhat like saying a "v" and an "m" at the same time.

Linguists have claimed that these sounds should be impossible to make in any language, because the aerodynamics of the mouth and nose make it impossible to make a sound like "v" while also having air go out through the nose.  These sounds are said to occur in Scots Gaelic.

Scots Gaelic is an endangered Celtic language spoken in Scotland, and related to Irish, Welsh, and other Celtic languages.  Scots Gaelic uses several sounds that are very uncommon in languages of the world, and completely unfamiliar in languages like English – such as these nasalized fricatives.

Possibilities with “impossible” sounds

This raises the question of what sound Scots Gaelic speakers are actually making when they pronounce the words which are thought to have these sounds:

• Is it possible to make these sounds after all, and Scots Gaelic speakers do so?  
• Or do Scots Gaelic speakers actually pronounce a slightly different sound, for example with air flowing out through the nose during the vowel before the "v" instead of during the "v" sound?  
• Or do they simply not pronounce these sounds in this way at all, and make a normal "v" instead?
  

How do we find out?

In our work, we measured the amount of air flowing out of the nose and mouth while a person talks (using a machine with sensors located in air masks held over the mouth and nose).  This allows us to find out whether the sounds that are said to be nasalized fricatives have air flowing out the nose during the "v" (or other consonant).  

We took this equipment to the Isle of Skye and recorded 14 native Scots Gaelic speakers' pronunciations of words containing these sounds and comparison words with the non-nasalized (regular) "v" sound.  For example, speakers read words like "comhairle" (meaning 'advice,' in which the sound spelled "mh" is expected to be a nasalized "v"), and "cobhair" (meaning 'relief,' in which the sound spelled "bh" is expected to be a regular, non-nasalized "v").

What do people really do?  Five different strategies!

We found that most of the time, most speakers simply do not have air flowing out the nose during or near these sounds at all:  they pronounce a normal "v" instead of a nasalized one.  

However, in some of the consonants where a nasalized "v" is expected, speakers do have at least some air flowing out the nose during at least part of the consonant.  So how do they do this?

• We found that some speakers pronounce some of their consonants with a regular "v" sound during the first part of the sound, then switch to letting air flow out the nose during the later part of the sound, so that they can make the expected sound without having to accomplish both parts of it at the same time.  
• Other times, speakers have air flowing out the nose only very briefly, as they transition from the previous vowel into the "v" or from the "v" into the next vowel, but not during any substantial part of the "v" itself.  
• Another solution speakers use is to change the "v" sound to a "w" sound, which can be combined with air flowing out the nose without an aerodynamic problem.  
• In a very few cases (less than 10 out of more than 1600 words we measured), speakers do manage to have air flow out both the mouth (the "v" sound) and the nose at the same time, but with very little air flowing out the mouth, the nose, or both, so that the sound becomes weaker. 

Dealing flexibly with “impossible” sounds

These results are especially interesting, because they show that when the sounds of a language present a physical problem (it may not be physically possible to make a particular sound) speakers adopt a wide range of solutions, even though they are speaking the same language.

Furthermore, even a single speaker uses a wide variety of pronunciations for different words.  Thus, rather than the language simply shifting to one alternative type of sound, speakers use many different adjustments to the position of the mouth to make various versions of the sound.