The vowel system in Kawesqar:

quantifying category overlap and

the status of glides

Dr. Mauricio A. Figueroa Candia

Dr. Daniela A. Mena Sanhueza
Facultad de Humanidades y Arte
Universidad de Concepción
Concepción, Chile
 Introduction
About the Kawesqar people, about their
language, and aims of this study
About the Kawésqar
Introduction

Kawesqar is a critically endangered language originally

spoken in ample areas of Chilean Patagonia. The
Kawesqar were a nomadic seafaring people.
(Viegas Barros, 1990)

Before contact with Europeans, it is estimated that they

numbered around 6000. By the mid-1950s, only about
100 “pure” [sic] Kawesqars remained.
(Gusinde, 1924; Bird, 1946; Hammerly Dupuy, 1947; Clairis, 1972a;
1972b; Aguilera, 1978; Clairis, 1985; Crevels, 2012)

Currently the language is spoken by less than 5 fluent

speakers, most of them of very advanced age and living
in relative isolation in Puerto Edén (a village).
(Crevels, 2012)
About the Kawésqar
Introduction

from about here

(46° 30' S)

Puerto Edén village

●
to about here ●
(53° 28' S)

●
About the Kawésqar
Introduction

As most specialists agree, this language is bound to

become fully extinct in the next few years.
(Viegas Barros, 1991; Sánchez, 1994)

As yet there is no formal nor systematic attempt by the

Chilean State to document or preserve the language.
 About the Kawésqar
Introduction

Source: https://www.eldesconcierto.cl/2020/03/23/muere-carlos-renchi-uno-de-los-mas-antiguos-kawesqar-y-uno-de-los-ultimos-hablantes-de-su-lengua/
About the language
Introduction

Kawesqar has been characterized as an agglutinative and

polysynthetic language, with free lexical stress, genderless
and without plural marking.
(Aguilera, 1978, 1982b)

It has been proposed that Kawesqar is a language isolate.

(Lehmann-Nitsche, 1919; Lipschutz, 1962; Viegas Barros, 1991;
Aguilera, 1982b; Campbell, 2012)

Others have linked Kawesqar to other languages from the

same general area, such as Chono, Selknam and Yaghan.
(Clairis, 1978; Viegas Barros, 1994; Greenberg, 1960; Key, 1979)
About the language: sounds
Introduction

The sound system of Kawesqar is interesting given the

occurrence of several relatively uncommon phenomena.

For example, the language has a natural class of

ejectives ([pʼ tʼ kʼ qʼ]) and, perhaps more interestingly, it
lacks a natural class of contrastive voiced stops.
(Campbell, 2012; Aguilera & Tonko, 2006)

Other interesting sounds that have been reported for the

language, although not necessarily as contrastive units,
are aspirated and implosive stops.
(Aguilera, 1982a; Clairis, 1997)
About the language: sounds
Introduction

Regarding the vowel system, there is considerable

debate as to the number of underlying contrastive units,
ranging from 3 (/e/, /a/ and /o/ + consonants /j/ and /w/) to
7 (/i/, /e/, /æ/, /a/, /ɑ/, /o/ and /u/) depending on the study.
(i.e., Clairis, 1997; Aguilera, 1983)

From a phonetic standpoint, the following units are

commonly reported: [i j], [e], [æ], [a ɑ], [o], [u w].

Pertaining the status of [j] and [w], they have been

interpreted both as variants of the vowels /i/ and /u/ and
as gliding approximant consonants /j/ and /w/.
(Clairis, 1997; Aguilera, 1983; Aguilera & Tonko, 2006)
Main goals
Introduction

This exploratory study aims at providing first acoustic and

statistical evidence to investigate:

(a) the degree of phonetic overlap between vowel

categories, and

(b) the acoustic nature of close front ([i j]) and close
back ([u w]) vowels.

While the first subject matter relates to questions of

vitality and of whether contrasts are being maintained, the
second seeks to shed light on the problem of the
interpretation of the vocoids [j] and [w] in the phonemic
system of Kawesqar.
 Methods
Task and participants, technical standards,
annotation & measurements
Task and participants
Methods

Task: lexical list in which a researcher prompted a word

using Spanish and then the participants translated the
word to Kawesqar, repeating it three times.

Participants: 2 female and 2 male participants, between

35 – 60 years of age at the time of the recording (late 70’s
or early 80’s).
Technical standards
Methods

Recordings: the exact conditions and

materials used to record the signals
are unknown; recordings were
originally made in an electromagnetic
cassette format, with considerable
background noise.

Quality of the signals: the signals,

which showed clear signs of
deterioration due to time, were
digitized in a WAV format (mono), with
a sampling frequency of 44100 Hz and
a bit depth of 16 bit (signal-to-noise
ratio ≈ 21 dB).
Annotation & measurements
Methods

Annotation: annotation was conducted at the word and

segmental levels; vowels were coded as [j], [e], [æ], [a],
[ɑ], [o] or [w].

Measurements: all measurements were made in Praat.

(Boersma & Weenink, 2018)

(a) Mean F1 and F2 were measured at the inner 50%

of the vowels, using Formant objects created
separately for each participant (maximum frequency
value was set at 5500 Hz for females and at 5000 Hz
for males).

(b) Mean intensity was measured in the entire vowel,

from Intensity objects.
 Results
The corpus, quantifying vowel overlap,
differences in intensity
The resulting corpus
Results

A preliminary inspection of the measurements revealed

the presence of a few extreme outliers in F2. In total, 15
outliers were removed using the criteria of 2.5 absolute
deviations around the median.
(Leys et al., 2013)

The final corpus comprised of a total of 2493 vowels,

distributed as follows: [j] = 298, [e] = 628, [æ] = 8, [a] =
979, [ɑ] = 136, [o] = 229, [w] = 215.

F1 and F2 values were normalized using Lobanov’s

procedure in order to make the data from female and
male speakers comparable.
(Lobanov, 1971)
The resulting corpus
Results
Quantifying vowel overlap
Results

Pillai scores –derived from MANOVA analyses– and

Bhattacharyya’s Affinity values were calculated for all
pairs of vowel categories.

Pillai scores measure the overlap between two vowel

distributions; scores range from 0 to 1, with values
closer to 0 indicating more overlap.
(Pillai, 1955; Hay, Warren & Drager, 2006)

Bhattacharyya’s Affinity also measures overlap

between distributions, but in this case values closer to
0 indicate less overlap.
(Fieberg & Kochanny, 2005; Warren, 2018)
Quantifying vowel overlap
Results
Quantifying vowel overlap
Results

Pillai score: closer to 0 means near complete overlap.

Bhattacharya’s Affinity: closer to 1 means near complete overlap
Quantifying vowel overlap
Results
Quantifying vowel overlap
Results

Results revealed considerable overlap between most

vowel categories, in some cases reaching virtually
complete overlap (as in the cases of [j] vs [e], [a] vs [ɑ],
and [o] vs [w]).

All vowels displayed substantial variability, which

sometimes resulted in vowels with abnormally large
distributions in the Cartesian space (e.g., [w]), which in
turn made superposition more likely.

Three groupings emerge from the data: (1) a close to

close-mid front area; (2) an open central-back area; and
(3) a close-mid central-back area.
Differences in intensity
Results

Regarding the status of [j] and [w], the intensity of all

vowels but [æ] was measured and compared using one-
way ANOVA.

In a separate analysis, all instances of [j] and [w] were

collapsed into one category (“glides”) and the rest of the
vowels into another (“non-glides”). Then, these two
macro-categories were compared using a Wilcoxon rank
sum test with continuity correction.
Differences in intensity
Results

Left-side panel:
effect of vowel
category in intensity
F(2,2479) = 11.98,
p < 0.001, η2 =
0.024

Right-side panel:
intensity differences
between groups
W = 414810, p <
0.001, r = 0.126
Differences in intensity
Results

These results showed that [j] and [w] displayed

significantly lower intensity values than the rest of the
vowels, and in a direction compatible with a hypothesis of
a more consonantal nature.

An inspection of the phonotactical distribution of [j] and

[w] showed that 78.2% of all instances occurred in the
immediate vicinity of other syllabic vowels, that is, as non-
syllabic glides.
Discussion and closing
remarks
The state of the vowel system
The state of the vowel system
Discussion and closing remarks

Both impressionistic and quantitative evidence suggests

that the vowel system of the speakers included in the
sample –fluent speakers from 4 decades ago– already
displayed evident signs of degradation.

In particular, very high variability was observed within

each vowel category and significant overlap was found
between vowel categories that the literature considers
(sometimes) to be phonologically contrastive.

This can be interpreted as indicative of the loss of vitality

of the vowel system of Kawesqar, which is a language in
imminent danger of extinction.
(Viegas Barros, 1991; Sánchez, 1994)
The state of the vowel system
Discussion and closing remarks

An alternative interpretation of the data is to see the

amount of overlap as evidence of only three underlying
vocalic categories, perhaps /e/, /a/ and /o/.

This would require interpreting instances of [i j] and [u w]

as the underlying consonants /j/ and /w/. Both
explanations have been suggested previously.
(Clairis, 1997)

The high degree of overlap between [æ], [a], [ɑ] could also
suggest that, indeed, [a] and [ɑ] are allophonic variants
of /a/, the latter emerging more often after uvular
consonants, and that [æ], which is very infrequent, might
be a lexically constrained allophone of /a/, or perhaps a
lexically constrained /æ/.
(Aguilera, 1983)
The state of the vowel system
Discussion and closing remarks

Regarding the status of [j] and [w], the results suggest

that there is room for a hypothesis of /j/ and /w/, as
underlying contrastive units.

As suggested by some authors, [j] and [w] could be

interpreted as the consonants /j/ and /w/, although close
to 20% of instances of closed vowels (front and back)
were syllabic.

More research is needed to clarify the status of [j] and

[w]. One avenue of future inquiry pertains intensity
differences between these units and previous and
following segments, along with other correlates of degree
of constriction.
Financed by FONDECYT Iniciación grant N.° 11180356,
titled “Reanálisis de aspectos controversiales del
kawésqar y nuevos aportes para su conocimiento:
descripción acústica, exploración del dominio perceptivo
y procesamiento estadístico multivariable para la
actualización de su inventario fonético y fonológico”.
References
Aguilera F., Ó. E. (1978). Léxico Kawéskar-Español, Español-Kawésqar (Alacalufe
Septentrional). Boletín de Filología, 29, pp. 7-149.
Aguilera F., Ó. E. (1982a). Fonemas kawesqar. Esbozo de la fonología de la palabra. I Parte.
Boletín Indigenista de Chile, 3, pp. 16-30.
Aguilera F., Ó. E. (1982b). Notas sobre la lengua kawésqar. Boletín Indigenista de Chile, 2,
pp. 21-33.
Aguilera F., Ó. E. (1983). Fonemas kawesqar. Esbozo de la fonología de la palabra. II Parte.
Boletín Indigenista de Chile, 4, pp. 49-67.
Aguilera F., Ó. E., & Tonko P., J. S. (2006). Manual para la enseñanza de la lengua
kawésqar. Nivel básico Ib. Santiago/Puerto Edén/Punta Arenas.
Bird, J. (1946). The Alacaluf, en Julian H. Stewart (ed.), Handbook of South American
Indians. Vol I. The marginal tribes. New York: Cooper Square Pub.
Boersma, P., & Weenink, D. (2018). Praat: doing phonetics by computer [Computational
software], version 6.0.39. Downloaded from http://www.praat.org/
Campbell, L. (2012). Typological characteristics of South American indigenous languages. In
L. Campbell & V. Grondona (Eds.), The indigenous languages of South America: a
comprehensive guide. Volume 2 (pp. 259-330). Germany: Walter de Gruyter.
Clairis, C. (1972a). Les Alakalufs de Puerto Edén (1971). Objets et Mondes, 12(2), pp. 197-
200.
Clairis, C. (1972b). Qawashqar: una investigación etnolingüística en el Pacífico. Revista de
Estudios del Pacífico, 5, pp. 7-26.
Clairis, C. (1978). La lengua qawasqar (alakaluf). VICUS. Cuadernos de Lingüística, 2, pp.
29-44.
References
Clairis, C. (1985). El qawasqar. Lingüística fueguina. Teoría y descripción. Anejo 12.
Estudios Filológicos. Valdivia: Universidad Austral de Chile.
Clairis, C. (1997). Lingüística fueguina 1997. Onomazein, 2, pp. 421-450.
Crevels, M. (2012). Language endangerment in South America: The clock is ticking. In L.
Campbell & V. Grondona (Eds.), The indigenous languages of South America: a
comprehensive guide. Volume 2 (pp. 167-233). Germany: Walter de Gruyter.
Fieberg, J., & Kochanny, C. O. (2005). Quantifying home ‐range overlap: the importance of
the utilization distribution. The Journal of Wildlife Management, 69(4), 1346-1359.
Greenberg, J. H. (1960). The general classification of Central and South American
languages. Selected papers of the 5th International Congress of Anthropological and
Ethnological Sciences. Philadelpia, pp. 791-794.
Gusinde, M. (1924). Cuarta expedición a la Tierra del Fuego. Publicaciones del Museo de
Etnología y Antropología de Chile, 4(1-2), pp. 7-67.
Hammerly Dupuy, D. (1947). Redescubrimiento de una tribu de indios canoeros del sur de
Chile. Revista Geográfica Americana, 28(168), pp. 117-122.
Hay, J., Warren, P., & Drager, K. (2006). Factors influencing speech perception in the context
of a merger-in-progress. Journal of phonetics, 34(4), 458-484.
Key, M. R. (1979). The grouping of South American Indian Languages. Tübingen.
Lehmann-Nitsche, R. (1919). El grupo lingüístico alakaluf de los canales magallánicos.
Revista del Museo de La Plata, 25, pp. 15-69.
Leys, C., Ley, C., Klein, O., Bernard, P., & Licata, L. (2013). Detecting outliers: Do not use
standard deviation around the mean, use absolute deviation around the median. Journal
of Experimental Social Psychology, 49(4), 764-766.
References
Lipschutz, A. (1962). Los últimos fueguinos: transculturación y desculturación, extinción y
exterminación. Genus, 18(1-4), pp. 3-29.
Lobanov, B. M. (1971). Classification of Russian vowels spoken by different speakers. The
Journal of the Acoustical Society of America, 49(2B), 606-608.
Pillai, K. C. S. (1955). Some new test criteria in multivariate analysis. The Annals of
Mathematical Statistics, 26(1), 117-121.
Sánchez, G. (1994). Estado actual de las lenguas aborígenes de Chile. Boletín de la
Academia Chilena de la Lengua.
Viegas Barros, J. P. (1990). Dialectología qawasqar. Amerindia, 15, pp. 43-73.
Viegas Barros, J. P. (1991). Fonología histórica qawasqar. In Temas de lingüística aborigen.
Coord. Ana GERZENSTEIN, Facultad de Filosofía y Letras, Universidad de Buenos
Aires, Buenos Aires, pp. 71-99.
Viegas Barros, J. P. (1994). La clasificación de las lenguas patagónicas. Revisión de la tesis
del grupo lingüístico "andino meridional" de Joseph H. Greenberg. Cuadernos del
Instituto Nacional de Antropología y Pensamiento Latinoamericano, 15, Buenos Aires,
pp. 167-184.
Warren, P. (2018). Quality and quantity in New Zealand English vowel contrasts. Journal of
the International Phonetic Association, 48(3), 305-330.