2 Classification and inventory of approximants
2.1 Classification
Two main groups are distinguished: centrals and laterals, although semivowels are not distinguished from other centrals by IPA (2020). Martínez-Celdrán (2004) excludes semivowels and divides centrals into rhotics and spirants, a term adopted from the French spirantes. The term was proposed by Martinet (1956:24) in reference to constrictive consonants articulated with a rather limited narrowing of the expiratory channel so that air flows without friction noise (according to Mounin 1974). Spirants are thus nonrhotic central approximants.
2.2 Inventory
An inventory could be deduced from
.the latest IPA chart (IPA 2020)
.a list of approximants found in different studies.
2.2.1 The latest IPA chart
The 2020 IPA chart’s fricatives and approximants are presented in Table 1. In the central line, ‘Approximant’, there are what IPA (1989a: 69) call “blank spaces”. Some of them which represent impossible places of articulation are filled with shading. Instead of the other blank spaces should figure the symbols of the fricative above and the lowering diacritic together. However, symbols with a diacritic do not appear in the chart.
Table 1. Fricatives and approximants in 2020 IPA Chart
2.2.2 List of approximants included in the category
Voiced approximants
The inventory of voiced approximants varies across authors.
Martínez-Celdrán (2004) proposes four spirants: [ β̞ ð̞ ʝ̞ ɣ̞].
Ball & Rahili (2011) describe four more spirants: [z̞ ʒ̞ ʁ̞ ʕ̞].
Another frictionless continuant should be added:
.a voiced unrounded uvular/pharyngeal approximant [ʌ̯], as in Marphali [kʌ̯e] ‘plank’, described by Mazaudon (2007) as a glide.
Maddieson (1984) transcribes with the same symbol, [ʌ̯], the Vietnamese lower midback approximant described by Thompson (1965: 39).
Voiced approximants are thus composed of eleven central frictionless continuants, five semivowels and four laterals. There are thus twenty voiced approximants.
Voiceless approximants
Five voiceless approximants, [ɹ̥], [j̥], [ʍ], and [χ̞], can be described, even though they seem theoretically impossible to several authors:
Arnold (1963: 4) considers that “since the only auditory component of a frictionless continuant is voice, it follows that voiceless frictionless continuants, though articulatorily feasible, are perceptually impossible since they produce no audible effect”.
O’Connor (1973: 61) does not include in his table of consonants the voiceless counterpart of approximants, the ‘voiceless frictionless continuants’, because “this would imply silence.”
Akamatsu (1992: 30) remarks: “At any rate, a voiceless approximant should be silence”.
However, owing to Catford’s studies, the existence of voiceless approximants is found, even though it is debated.
Maddieson (1984: 91), citing Catford (1977), admits the existence of voiceless approximants produced with a friction noise.
Catford (1977) distinguishes fricatives and approximants by their degree of turbulence.
He measures this degree by the Reynolds number (Re). The Reynolds number is a dimensionless quantity in which, when it is used to measure turbulence in the vocal tract, three elements intervene: the kinematic viscosity of air (noted v; 0.14 cm² /s, according to Catford), the articulatory stricture area (usually d) and the volume velocity of the air in the vocal tract (U̅).
The Reynolds number formula is Re = U̅d/v.
Ohala & Solé (2010: 40) do not find the Reynolds number relevant in the analysis of speech sounds. According to them, “some turbulence, even audible turbulence, can occur in conditions where the Reynolds number is far below the ideal threshold between laminar and turbulent flow”. However, the Reynolds number seems relevant in the way that Catford (1977) uses it. With the Reynolds number, Catford (1977) shows that there is a difference in the degree of turbulence between fricatives and voiceless approximants. Two categories of speech sounds are distinguished according to their degree of turbulence: those with a laminar airflow, (wide open front) vowels and voiced approximants (as we will see below, Catford includes vowels, although not wide open front vowels of types [ɛ] and [a], among approximants), and those with a turbulent airflow, fricatives and voiceless approximants. The values of the Reynolds number for each class of sounds according to Catford (1977) and their distribution in relation to the threshold are represented in Table 2. They are extracted from Catford’s figure 27 (1977: 124). They ensue from hypotheses and not from precise calculation and observation, but Catford (1977: 123) finds them realistic. The threshold between laminar and turbulent speech sounds is 1700. It appears that voiceless approximants have a degree of turbulence intermediate between that of voiced fricatives and that of voiceless fricatives. Therefore, even if they have, according to Catford, the same cross-sectional area, 40 mm² , voiceless and voiced approximants are not adjacent in the classification. This is due to the difference in the estimated volume velocity: 494 cm³ in the production of a voiceless approximant and 129 cm³/s in the production of a voiced approximant.
Table 2. Classification of some speech sounds according to the nature of their airflow by means of the Reynolds number (Re) according to Catford (1977). (Re threshold = 1700)
It is therefore legitimate to complete the list with a few voiceless approximants:
. a voiceless postalveolar approximant [ɹ̥], as in Bengali [ˈäbäɹ̥] ‘again’ described by Khan (2010).
. a voiceless dorso-palatal approximant [j̥], as in Klamath [hɑj̥ɑːj̥ɑ] ‘tracks in front of’, as described by Blevins (1993) and as in Scottish Gaelic [əˈmuj̥] ‘outside’, as described by Bauer (2011).
. a voiceless labial-velar approximant noted [w̥] or [ʍ][1], according to the authors, as in [kuw̥ɑ] ‘swells up’, is described in Klamath by Blevins (1993); it is also described in Slovene by Šuštaršič et al. (1995) in the word [ʍsaːk] ‘every’. It also appears in Yao in the word [kuʍala] ‘to shine’, as described by Sanderson (1922).
. a voiceless alveolar lateral approximant [l̥], as in Klamath [l̥aepʼ] ‘be flap’, according to Blevins (1993), and in the Burmese word [l̥a] ‘beautiful’, according to Ball & Rahili (1999: 91).
. a voiceless dorso-uvular approximant [χ̞], as in [aχ̞tist] ‘artist’; it occurs in French before a voiceless obstruent, according to Ramasse (2017).
[1] The same symbol is used to transcribe a fricative in the IPA chart and an approximant in the descriptions of languages where it appears. The contradiction will be dealt with in 4.2.2 below.
2.2.3 Classification and inventory: summary
Twenty-five approximants, twenty central and five lateral, have therefore been found. It is summarized in Table 3. A voiceless dental lateral approximant [l̪̥], which appears in Gununa-Kena, according to Maddieson (1984), was not retained; it concerned only one language, and its articulation was very near that of the voiceless dental/alveolar lateral approximant described, also by Maddieson (1984), in five languages.
Table 3. List of approximants presently included in the category.
It appears from the chart that the class of approximants is a class of sounds in two major parts: central approximants composed of semivowels and spirants on the one hand and lateral approximants on the other hand. The class is aligned with two classes: fricatives and vowels. It appears from the chart that the class of approximants is a class of sounds in two major parts: central approximants composed of semivowels and spirants on the one hand and lateral approximants on the other hand. The class is aligned with two classes: fricatives and vowels.
The presence of voiceless approximants in the list excludes a description of them as frictionless continuants.
Instead of considering them fricatives, as Ohala and Solé (2010) do, it seems preferable not to classify them in terms of the presence or absence of a constriction noise. Voiceless approximants are produced with a turbulent flow. The degree of turbulence permits to distinguish a voiced and voiceless approximant from a fricative with same characteristics regarding the voicing.
An approximant presents a smaller degree of turbulence than a fricative. Voiceless approximants present a turbulent flow to a lesser degree of turbulence than voiceless fricatives. Similarly, the degree of turbulence of voiced approximants is smaller than that of voiced fricatives. Physical limits between the total class of approximants and their neighbors are not described precisely by Catford. In the second part of the present work, the limits of approximants in phonetic and phonologic areas will be specified more precisely. The presence of voiceless approximants in the list excludes a description of them as frictionless continuants. Instead of considering them fricatives, as Ohala and Solé (2010) do, it seems preferable not to classify them in terms of the presence or absence of a constriction noise. Voiceless approximants are produced with a turbulent flow. The degree of turbulence permits to distinguish a voiced and voiceless approximant from a fricative with same characteristics regarding the voicing. An approximant presents a smaller degree of turbulence than a fricative. Voiceless approximants present a turbulent flow to a lesser degree of turbulence than voiceless fricatives. Similarly, the degree of turbulence of voiced approximants is smaller than that of voiced fricatives.
Physical limits between the total class of approximants and their neighbors are not described precisely by Catford. In the second part of the present work, the limits of approximants in phonetic and phonologic areas will be specified more precisely.