Nidaba : A segment distribution database for measuring language
4.7 Frequency data
4.8.4 Vowel inventory parameters
The vowel inventory parameters capture not fine or even broad phonetic detail, given the inher-ent difficulties of categorising vowels that way (Lass, 1984), but rather the presence or absence of phonological contrasts. They cover vowel height, ATR, backness and rounding (Rice, 2002);
nasality, and phonation types.
(1) Height parameter
Does the vowel system have more than one contrast in height?
“Every phonological system contrasts at least two degrees of aperture” and therefore has at least one contrast in height (Hyman, 2008). The majority of languages have more than two heights (Maddieson, 1984).
All the languages in my sample had contrasts between (at least) three heights.
(2) ATR contrast
Is there at least one ATR or tense/lax contrast?
A language with a tense/lax contrast has an additional contrast in its front or back vowels on top of two existing height contrasts. For the purposes of this parameter, any vowel contrast which includes a quality difference is counted, regardless of whether there is also a correspond-ing length difference. This parameter depends on Parameter 4.8.4.1.
Crothers (1978) categorises [ɛ], [a] and [ɔ] as not (necessarily) contrasting in heiɡht, and hence most lanɡuaɡes in his typoloɡy have no more than three distinct categories. However, I shall follow the common practice of categorising seven-vowel systems such as those of Italian and Yoruba as having a tense/lax or ATR contrast, rather than having a rounding contrast in the back vowels (e.g. Calabrese, 1998, Pulleyblank, 1996).
English, Dutch, German, and Lithuanian have at least four distinct categories. Even if the Dutch tense/lax contrast is instead analysed as a length contrast, French loan items give rise to a four-way contrast. The German /ɛ/ vs /e/ distinction is debated; I am here following Wiese (2000) and Baayen, Piepenbrock and Rijn (1993) in treating them as separate. French, Matbat
and Portuguese have a four-way contrast assuming that [ɛ] and [ɔ] are categorised as differing in height from [a].
The Welsh lexicon used in Nidaba evinces no tense/lax contrast, so this is the analysis I am following. However, there is disagreement on whether a certain category of contrast is more properly described as a length contrast or vowel quality contrast (Hannahs, 2013), with variation between speakers / dialects (Iosad, 2017).
Ambel, Cheke Holo, Greek, Spanish, and Sylheti all have five-vowel systems, with no ATR contrast. Hrusso Aka, Polish and Romanian have three contrasting vowel heights.
(3) Multiple ATR contrasts
Are there two or more ATR or tense/lax contrasts?
Such a language may also be described as having a five-way contrast in vowel height (Croth-ers, 1978, Lass, 1984). This parameter implies that 4.8.4.2 is true.
Dutch, English, German and Lithuanian have a tense/lax contrast in both high and mid vowels.
French and Matbat only have a single tense/lax contrast, between low-mid and low vowels.
Portuguese does too, assuming that [ɐ], unlike [a], is not contrastive in height with [ɛ] or [ɔ].
This parameter is not applicable to Ambel, Cheke Holo, Greek, Hrusso Aka, Polish, Ro-manian, Spanish, Sylheti, or Welsh, which lack any tense/lax contrast.
(4) Back parameter
Does the vowel system have contrastive roundness or contrastive frontness?
This parameter captures the difference between vertical vowel systems, such as Kabardian, which only realise frontness or roundness on consonants or morphemes, and the more typical language with such a contrast inherent to vowels (Hyman, 2008).
All the languages in my sample contrast front unrounded vowels with back rounded vowels.
(5) Front rounded parameter
Is there a rounding contrast in the front vowels?
If so, 4.8.4.4 is true; the language has at least contrastive rounding. To avoid ambiguity in setting this parameter, the language must have at least one back or central vowel at the same height as the contrast, such that the front rounded vowel cannot be alternatively analysed as a central or back rounded vowel.
Dutch, French, and German have a rounding contrast in the front vowels.
Ambel, Cheke Holo, English, Greek, Hrusso Aka, Lithuanian, Matbat, Polish, Portuguese, Romanian, Spanish, Sylheti and Welsh do not.
(6) Back unrounded parameter
Is there a rounding contrast in the non-front vowels?
If so, 4.8.4.4 is true; the language has at least contrastive rounding. A vowel system may be described as having back rounded and back unrounded vowels, or back rounded and central unrounded vowels (e.g. Turkish, Rice, 2002); either of these contrasts sets this parameter as true. As in 4.8.4.5, there must be at least one front unrounded vowel at the same height as this contrast.
Polish and Portuguese have a contrast between high central unrounded and high back roun-ded vowels. Romanian and Welsh have a contrast in both high and mid vowels. Hrusso Aka has contrast in the high vowels, and a marginal contrast in the mid vowels. German has a trast between mid central unrounded and mid back rounded vowels, though prosodically con-ditioned15. Lithuanian also has a contrast between [ʌ] and [o:], with a concomitant length dis-tinction. English and Dutch have a contrast between central and back mid vowels, and rounded and unrounded low vowels.
Ambel, Cheke Holo, French, Greek, Matbat, Spanish and Sylheti do not have a rounding contrast in the non-front vowels.
(7) Nasality parameter
Does the vowel system have an oral / nasal contrast?
15Taking schwa to be a contrastive segment in German, following Féry (1991).
This parameter captures the difference between languages with no or allophonic nasal vow-els (e.g. English) and languages which use vowel nasality contrastively (e.g. French). A language with nasal vowels will always have oral vowels too, giving an oral/nasal contrast.
French and Portuguese have nasal vowels. Polish is variously analysed with and without nasal vowels; I have chosen to categorise it as having an oral/nasal contrast in the vowel system, but the lexicon in Nidaba is transcribed with a nasal archiphoneme, allowing for alternative in-terpretations to be applied to the data. Hrusso Aka contains vowel nasalisation only marginally (see also D’Souza, 2015), with nasalisation present in only seven lexical items out of over 3200;
but without token frequency data, I am not conclusively excluding it.
The English, Dutch and German lexicons from the CELEX database contained items tran-scribed with nasal vowels (i.e. French loanwords). These items are both small in number and infrequent. Furthermore, the loanwords are not (consistently) produced with nasal vowels, re-gardless of their transcription in CELEX (see e.g. Cambridge Dictionary, 2015).
Ambel, Cheke Holo, Greek, Lithuanian, Matbat, Romanian, Spanish, and Sylheti do not have a contrast between oral and nasal vowels.
(8) Breathiness parameter
Does the vowel system have a modal / breathy contrast?
Of the different phonation types, all languages have modal voicing in vowels, so any lan-guage with phonemic breathy vowels will have a contrast between modal and breathy phona-tion.
None of the languages in my sample have a modal / breathy contrast.
(9) Creakiness parameter
Does the vowel system have a modal / creaky contrast?
Breathiness and creakiness may both be used contrastively in vowels, including in the same language (Silverman et al., 1995), though this can only produce a three-way contrast.
None of the languages in my sample have this contrast.
The final phonation type, voicelessness, is only ever found predictably in vowels, in certain contexts. It is not used contrastively (Gordon and Ladefoged, 2001).