Unlike Henning Reetz’s Simple UPSID interface, this interface is exclusively text based. This introduces some complexity, but it also makes it possible to go beyond rigid search options any fixed interface can offer.
The query language implemented in the interface contains two basic query types—count queries and comparison queries–and allows the user to combine them in a flexible way through logical operators. (If you are comfortable with reading CFGs, all you need to know can be found in the very short grammar, from which the query parser is constructed on the fly using Lark.)
We will describe the basic queries first and then show how to combine them into more complex ones.
The general form of a count query is operator number features or
operator number /phoneme/, where
operator is one of =, <, >, <=, >=,number is a non-negative integer,features are one or more space-separated IPA features (see the full list at the end of the page),
each optionally preceded by the negator ^, andphoneme is an individual segment described in standard IPA notation (https://ipa.typeit.org/full/
is used as a reference input system).For example, a query asking for languages without lateral segments looks like = 0 lateral and a query
to ask for languages with more than five voiceless affricates is > 5 voiceless affricate. Each
feature can be turned into its complement by using ^. Thus, the query
> 5 ^pre-nasalised voiced plosive will return languages that have more than 5 oral plosives that are
not pre-nasalised.
The order of features does not matter. Some features (e.g., nasalised) are shared between vowels and
consonants, and the words vowel and consonant can be used for disambiguation.
Diphthongs, triphthongs, and click consonants have features diphthong/triphthong and
click respectively assigned to them. Monophthong vowels have a feature monophthong, and
non-click consonants have a feature non-click. Use these features to explicitly query for monophthong
vowels and non-click consonants (the same goal can be achieved using ^diphthong ^triphthong vowel and
^click consonant respectively).
Note that the negative value for lateral is inserted by default for approximants and the negative
value for nasal is inserted by default for plosives. Thus, the query
= 0 voiced bilabial plosive will ignore the presence of /m/ (voiced nasal bilabial plosive) in
an inventory, but will not return languages with /b/. Likewise, the query = 0 approximant will skip
languages with /w/ and /j/, but will ignore the presence of /l/ (voiced lateral alveolar approxumant). The
query = 0 fricative, by contrast, will skip a hypothetical language whose fricative inventory consists
of /ɬ, ɮ/.
To make it easier to look for presence and absence of feature bundles and phonemes, two shortcuts are provided:
+, equivalent to > 0, and -, equivalent to = 0. E.g., a query
for languages without /p/ looks like - /p/.
Note that phoneme-based queries look for a symbol-by-symbol match. Thus, a language with /b, pʰ, pʼ/ will still
satisfy the - /p/ query and + /ɫ/ will not retrieve languages where this segment is
denoted by /lˠ/.
Comparison queries make it possible to search for languages that have more or fewer segments of some kind compared
to some other kind of segments. The basic form of the query is operator features, features, where
operator and features have the same form as above.
For example, languages with more vowels than consonants can be retrieved using > vowel, consonant
and languages with the same number of monophthong vowels and diphthongs can be queried by
= monophthong, diphthong (use = monophthong vowel, ^monophthong vowel if you want to count
all polyphthongs).
Sometimes basic queries are not enough. For example, in order to make basic queries simpler, the query language
does not allow for querying nasal and oral plosives simultaneously. Therefore, in order to look for languages
without both oral and nasal bilabial plosives it is necessary to combine two queries:
- nasal bilabial plosive and - bilabial plosive (or, more explicitly
- nasal bilabial plosive and - ^nasal bilabial plosive). This is also helpful to work around variations
in IPA notation, e.g. + /g/ or + /ɡ/ (both the special /ɡ/ symbol and the basic Latin /g/ are
officially recognised as part of IPA).
Three logical operators are supported (in the order in which they are applied):
not (return the complement of the query)and (return the intersection of two queries)or (return the union of two queries)Parentheses can be used for clarity or to override the operator precedence order. Thus,
- /p/ and - /b/ or - /t/ and - /d/ is equivalent to (- /p/ and - /b/) or (- /t/ and - /d/)
(and binds more tightly than or), but - /p/ or - /b/ and - /t/ or - /d/ is
not equivalent to (- /p/ or - /b/) and (- /t/ or - /d/), but is instead equivalent to
- /p/ or (- /b/ and - /t/) or - /d/.
not binds the most tightly, so it is necessary to use parentheses to invert any compound query.
An example compound query:
not (
(- /p/ or - /t/ or - /k/)
and
(- /b/ or - /d/ or - /g/)
) or
> 10 voiced affricateRedundant whitespace is ignored.
Canonical feature names are listed in the British orthography, but ize-spellings are supported as
well. Note that not all possible features are listed for consonants and vowels but only those found in the database
(see the complete
list in the parser grammar). E.g., all types of additional articulations are theoretically possible with both
consonants and vowels.
advanced, alveolar, apical, atr, back, breathy-voice, central, centralised, close, close-mid, creaky-voice, diphthong, front, half-long, less-rounded, long, lowered, mid, mid-centralised, modal-voice, monophthong, more-rounded, nasalised, near-close, near-open, non-syllabic, open, open-mid, overlong, pharyngealised, postalveolar, raised, retracted, rhotacised, rounded, rtr, short, shortened, triphthong, unrounded, velarised, voiceless
advanced, affricate, affricated, alveolar, alveolo-palatal, apical, approximant, aspirated, bilabial, breathy-voice, dental, devoiced, ejective, epiglottal, fricative, glottal, glottalised, hissing-hushing, implosive, interdental, labial-palatal, labial-velar, labialised, labiodental, laminal, lateral, less-rounded, long, lowered, nasal, nasalised, overlong, palatal, palatal-velar, palatalised, pharyngeal, pharyngealised, plosive, postalveolar, pre-aspirated, pre-glottalised, pre-labialised, pre-nasalised, raised, retracted, retroflex, rhotacised, rtr, short, syllabic, tap, trill, uvular, velar, velarised, voiced, voiceless