Derived from the "Alphabets des langues africaines" (Unesco-SIL 1993), in IPA and with ISO language mapping: http://sumale.vjf.cnrs.fr/phono/PhonologieN.php

This should happen after the code changes for the next update. Latest files are here #57.

We have retracted tongue root for vowels in https://github.com/cldf-clts/clts/blob/master/data/features.tsv, but not for consonants. There are some instances in our data:

• k̙ retracted-tongue root voiceless velar stop consonant
• ɡ̙ retracted-tongue root voiced velar stop consonant
• d̙ retracted-tongue root voiced alveolar stop consonant
• t̪̙ retracted-tongue root voiceless dental stop consonant
• ʕ̙ retracted-tongue root voiced pharyngeal fricative consonant

I have not come across any cases with ATR...

I would suggest that we use modifier letter small r U+02B3 for this feature, which should apply only to stops and affricates.

• tr > tʳ with-trilled-release voiceless alveolar stop consonant
• dr > dʳ with-trilled-release voiced alveolar stop consonant
• ⁿdr > ⁿdʳ with-trilled-release pre-nasalized voiced alveolar stop consonant
• ʈɽ > ʈʳ with-trilled-release voiceless retroflex stop consonant
• ɖɽ > ɖʳ with-trilled-release voiced retroflex stop consonant
• ⁿɖɽ > ⁿɖʳ with-trilled-release pre-nasalized voiced retroflex stop consonant
• tʃɾ > tʃʳ with-trilled-release voiceless post-alveolar affricate consonant
• br > bʳ with-trilled-release voiced bilabial stop consonant
• kpr > kpʳ with-trilled-release voiceless velar-and-bilabial stop consonant
• gbr > gbʳ with-trilled-release voiced velar-and-bilabial stop consonant

Further examples from PHOIBLE are not always consistent in the place of stop and rhotic.
The bilabially post-trilled dental t̪ʙ is dealt with under coarticulated segments #44

• http://wa.amu.edu.pl/nadcalc/

Very interesting, as the data could be linked (some 10 languages), and they offer some estimates on clusters, and how likely they are in the phonotactics of a given language.

Paper here:

• wa.amu.edu.pl/wa/files/The_NAD_Phonotactic_Calculator.pdf

This is a new secondary localisation feature, relatively rare. I propose the modifier letter small capital inverted R, U+02B6 for this. The only data available has this on coronal stops and affricates.

• tx > tʶ uvularized voiceless alveolar stop consonant
• t̪x > t̪ʶ uvularized voiceless dental stop consonant
• dx > dʶ uvularized voiced alveolar stop consonant
• d̪x > d̪ʶ uvularized voiced dental stop consonant
• tʃx > tʃʶ uvularized voiceless post-alveolar affricate consonant
• dʒx > dʒʶ uvularized voiced post-alveolar affricate consonant

This is applicable to vowels, consonants, and clicks.
On vowels:

• i͓ with-friction unrounded close front vowel
• u͓ with-friction rounded close back vowel
• ɯ͓ with-friction unrounded close back vowel

On consonants:

• r͓ with-friction voiced alveolar trill consonant
• ɭ͓ with-friction voiced retroflex lateral approximant consonant
• ʟ͓ with-friction voiced velar lateral approximant consonant

On clicks:

• ǀ͓ with-friction voiceless dental click consonant
• ǁ͓ with-friction voiceless lateral click consonant
• ǂ͓ with-friction voiceless palatal click consonant
• ǃ͓ with-friction voiceless alveolar click consonant

cf. http://phoible.github.io/conventions/

https://github.com/MKilani/FAAL/

Paper is here: https://brill.com/view/journals/ldc/aop/article-10.1163-22105832-01001300/article-10.1163-22105832-01001300.xml

@tresoldi will make sure to refine these, as he's anyway working on that.

Check the PanPhon data (as in the legacy issue cldf-clts/clts-legacy#121 ) and verify if a better integration with PanPhon is doable/desirable

Many affricates are not in BIPA when they have diacritics. Here an unordered list from PHOIBLE: d̥ʒ̥ t̻s̪̻ t̪s d̟ːʑ̟ d̪z d̠z̠ʲ d̟ʑ̟ d̻ʑ̻ d̻ʑ̻ʱ d̪ʒ d̺ʒ̺ d̻ʒ̻ d̟ʒ̟ d̥ʒ̊ d̺ʒ̺ʷ t̟ːɕ̟ t̺ɕ̺ t͉ɕ͉ t̟ɕ̟ t̻ɕ̻ t̻ɕ̻ʰ t̺ɕ̺ʼ t̠s̠ʲʰ t̺ʃ̺ t̻ʃ̻ t̟ʃ̟ t̺ʃ̺ʷ ˀt̪ɬ ˀt̪s d̥ʑ̥

@tresoldi, @cormacanderson, I need a list of these affricates, ideally exhaustive, with their name, their bipa character, and their features, Ideally in our standard form, so we can just add them. We also need a feature that we would add here. What would the feature be?

http://www.coli.uni-saarland.de/elaut/

They list IPA and SAMPA for several languages. Worth a look, how easily it could be added.

In line with the transcription in Lexibank ASJP, particularly the Transcription.md and orthographic profile as discussed in lexibank/asjp#4

• http://starling.rinet.ru/cgi-bin/response.cgi?root=config&morpho=0&basename=\data\typol\typology&first=1

If you look at this example, and compare it with the NAPA description. It seems, that by adding the NAPA, we could link a lot of the data in Ruhlen's database, similar to the way we use IPA to link to phoible...

I just ran pytest --doctest-modules on a piece of code that somewhere imports pyclts. I'm not entirely sure why that leads down the rabbit hole of checking the sanity of some clts modules, but it does:

>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>> traceback >>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>
clts/sources/apics/scraper.py:3: in <module>
    from pyclts.util import pkg_path
E   ImportError: cannot import name 'pkg_path' from 'pyclts.util' (/home/gereon/.local/etc/lexedata/lib/python3.8/site-packages/pyclts/util.py)

This is true, there is that import there and also in some of the other scrapers:

and the pyclts package does not seem to contain the string pkg_path anywhere at all, neither in the current version on GitHub nor in the version on pypi.

Maddieson uses the combining double arch below U+032B for whistled sibilants. This feature should only apply to sibilants.

• s̫ whistled voiceless alveolar sibilant fricative consonant
• ʃ̫ whistled voiceless post-alveolar sibilant fricative consonant
• z̫ whistled voiced alveolar sibilant fricative consonant
• ʒ̫ whistled voiced post-alveolar sibilant fricative consonant
• ts̫ whistled voiceless alveolar sibilant affricate consonant
• t̠ʃ̫ whistled voiceless post-alveolar sibilant affricate consonant
• dz̫ whistled voiced alveolar sibilant affricate consonant
• d̠ʒ̫ > dʒ̫ whistled voiced post-alveolar sibilant affricate consonant

We should be careful not to confuse with combining seagull below, U+033C, in use for linguolabials.

I have to admit I don't understand exactly what this is, but it seems they provide some laboratory phonetic data on speech sounds in different languages:

The google drive with all data:

• https://drive.google.com/drive/folders/10jsQgT5anb-E97DZMbGJjwPZ-Ux91xMM

The github, without data:

• https://voxclamantisproject.github.io/

The ACL paper:

• https://voxclamantisproject.github.io/assets/pdfs/VoxClamantis_Phonetic_Typology_Corpus-ACL_2020-CR.pdf

Help appreciated, if somebody understands if we can link this in some way to clts.

I seem to remember us discussing the use of these before for the labiodental b̪ and p̪ respectively. Graphemically, they are preferable and Maddieson uses them, although they are not IPA. Minimally, they should be entered as aliases. I thought they were but cannot find them at https://clts.clld.org/values

A number of coarticulated sounds are not being parsed.

• ŋm voiced velar-and-labial nasal consonant
• ŋmkp > ⁿkp pre-nasalized voiceless velar-and-bilabial stop consonant
• ŋmgb > ⁿgb pre-nasalized voiced velar-and-bilabial stop consonant

From LAPSyD:

• nm voiced alveolar-and-bilabial nasal consonant
• n̠m retracted voiced alveolar-and-bilabial nasal consonant
• nmtp > ⁿtp pre-nasalized voiceless alveolar-and-bilabial stop consonant
• n̠mt̠p > ⁿtp, pre-nasalized retracted voiceless alveolar-and-bilabial stop consonant

Three rare sounds but legitimate sounds should also be added.

• xʀ̥ voiceless velar-and-uvular fricative consonant
• t̪ʙ with-trill-release voiceless dental-and-bilabial stop consonant
• βɾ voiced bilabial-and-alveolar fricative consonant

ipapy offers some feature representations of some ipa characters / letters. The problem is that they need to be extracted somehow, it is not entirely clear. But I think it would be nice to list for each CLTS BIPA character if it would be accepted by IPAPY and also how it would be encoded in terms of features there.

• https://www.biodiversitylibrary.org/item/121160#page/5/mode/1up

This phonetic alphabet may be of historical interest (only).

• http://phonology.us/featurize

This should be easy to add to CLTS and would provide yet one more feature set.

• asjp
• bipa
• upa (description available and also nice to quote in papers)
• teutonia (or similar, a German dialect IPA for which a mapping is available)
• apa (important one, the American linguists' alphabet, used in SA and NA languages)
• gld/uts (important for conversion between tower of babel and other datasets)

• https://eric.ed.gov/?id=ED526778

This corpus is difficult to find, but many scholars seem to use it (?).

see for example here: https://www.biorxiv.org/content/biorxiv/early/2018/06/14/346965.full.pdf

and here: https://github.com/soskuthy/u-fronting

If the data is in any way available in IPA, we could/should add it.

• ä̠ not recognised in PHOIBLE, although both retracted and centralised features are available in https://github.com/cldf-clts/clts/blob/master/data/features.tsv

https://link.springer.com/article/10.1007/s10579-019-09483-3

It's a pity the authors didn't do this themselves as part of their database. But well, in this case, we definitely have to do it.

Assign write-order https://github.com/cldf-clts/pyclts/blob/dfddc2836bf66c83408e5c8a55b1f8f19d56551f/src/pyclts/models.py#L268-L276

Assign name-order https://github.com/cldf-clts/pyclts/blob/dfddc2836bf66c83408e5c8a55b1f8f19d56551f/src/pyclts/models.py#L277-L283

Cf. https://github.com/cldf-clts/pyclts/blob/dfddc2836bf66c83408e5c8a55b1f8f19d56551f/src/pyclts/models.py#L356-L371

The well-known collection of the phonological inventories of South American indigenous languages, hosted as UC Berkeley. Data is in IPA.

https://linguistics.berkeley.edu/~saphon/en/phonemes.php

I keep finding them, but consider these two gems: a fake schwa ә (04D9). Should be listed as alias for bipa right away.

ɾ̪ ɾ̪
ʐ͇ ʐ͇
ɫ̪ ɫ̪
ɾ̪ ᴅ
ð̪̺ ð̪̺
ɹ̪̩ ɹ̪̩
z͇ z͇
ʂ͇ ʂ͇
ts͇ ts͇
z͇ z̪͇|z͇
ɹ̪̹̩ ɹ̪̹̩
ɾ̪ˠ ɾ̪ˠ
ɯ͓ ɯ͓
k̙ k̙
ɗʒ ɗʒ
ɡǀ͓ ɡǀ͓
i̯au̯ i̯au̯
iou iou
ǁ͓ʰ kǁ͓ʰ
l̠˞ l̠˞
ʀʁ ʀʁ
s͇θ s̻θ
d̪ʒ d̪ʒ
ˀt̪ɬ ˀt̪ɬ
kɬ kɬ
ɖr̠ ɖr̠
ɖr̠͓ ɖr̠͓
ð͇ˠ ð͇ˠ
dz͇ d͇z͇
ⁿt͇s͇ʰ ⁿt͇s͇ʰ
s͇ s͇
ʃ͇ ʃ͇
t͇s͇ t͇s͇
ts͇ʰ ts͇ʰ
t͇s͇ʰ t͇s͇ʰ
ʈʂ͇ ʈʂ͇
ʒ͇ ʒ͇
t̪ʙ t̪ʙ
xʀ̥ xʀ̥
ŋ̥m̥ ŋ̥m̥
ð̙ˤ ð̪̙ˤ
ɹ̪̰̩ ɹ̪̰̩
ɹ̪ˠ ɹ̪̩ˠ
ɾ̪̊ ɾ̪̊
ɾ̪̊ʰ ɾ̪̊ʰ
ɾ̪ʲ ɾ̪ʲ
i͓ i͓
ʟ͓̥ ʟ͓̥
ɭ͓ ɭ͓
r͓ r̪͓|r͓
ɾ͓ ɾ͓
u͓ u͓
d̪l̪ d̪l̪
d̙ˤ d̙ˤ
t̪̙ˤ t̪̙ˤ
ʕ̙ ʕ̙
ɾ̪ ᴅ̪
ɾ̪̰ ᴅ̪̰
fʃ fʃ
ŋǂx ɡ̰ǂx
ŋǃx ɡ̰ǃx
ld ld
ʟ̝̊ɬ ɬʟ͓̥
ⁿbz mbz
ⁿd͇z͇ ⁿd͇z͇
r̠̙ r̠̙
st st
s̙ˤ s̙ˤ
ʃt ʃt
tθ̪ t̪θ̪
ˀy ˀy

As hosted at https://github.com/segbo-db/segbo (Zenodo: https://zenodo.org/record/3633918).

The dental diacritic in segments such as the dental tap ɾ̪ is currently not being recognised. Examples:

• r̪ voiced dental trill consonant
• ɾ̪ voiced dental tap consonant
• ɾ̪̊ voiceless dental tap consonant
• ɾ̪ˠ velarized voiced dental tap consonant
• ɾ̪̊ʰ aspirated voiceless dental tap consonant
• ɾ̪ʲ palatalised voiced dental tap consonant
• ɹ̪̹̩ rounded voiced dental approximant consonant
• ɹ̪̰̩ creaky voiced dental approximant consonant
• ɹ̪ˠ velarized voiced dental approximant consonant
• ɫ̪ voiceless dental lateral fricative consonant
• ɮ̪ voiced dental lateral fricative consonant
• ɮ̪ʲ palatalized dental lateral fricative consonant
• t̪ɬ voiceless dental lateral affricate consonant

Didn't see it before, it is Eden (2018): https://discovery.ucl.ac.uk/id/eprint/10058348/

We should update the reference upon new release.

We have more or less clarified this in code already:

• normalize is a one to many conversion procedure, only single characters are allowed, it is transcriptionsystem specific, as it is possible that different systems normalize in different ways
• confusables going beyond this are excluded and placed into the alias section

But we also started to collect things in cldf/multicode. Many of the examples there belong to what we would use to normalize a dataset. But not all.

I think we can drop multicode, as it was never really followed up, and we'd have to think how to integrate it into any of our tools (maybe one could use it for normalization in linse, where we also have a small normalization procedure for bipa only, to be able to use linse without depending on pyclts). But we should thoroughly check to have harvested all major characters from the unicode confusables list.

PHOIBLE uses the combining equals sign below U+0347 for non-sibilant fricatives. This is the alveolar diacritic in the extended IPA so is liable to cause confusion. This is a bit of a problem in the IPA, with no good way to deal with these.

• s͇
• z͇
• ts͇
• dz͇
• ʃ͇
• ʒ͇
• ʂ͇
• ʐ͇
• ʈʂ͇

In contrast, this one from PHOIBLE looks more like the alveolar diacritic, as it's not on a sibilant.

• ð͇

This should be metadata, I suppose, as Grimes (1959) lists features for only some 40 sounds relevant for Romance languages. He also describes a metric to define distances between sounds, and this is interesting. Ideally we should also add this data, but the question is: in which form? Distances call for a matrix, but csv is not really apt for matrices, due to the large number of column names. JSON could handle this. In any way, this will call for a custom script to create the data, based on features of Grimes, linkings, some description, a csv file with the sounds and our feature names, and the matrix in JSON. Question is again, how to do this in the most consistent way. We may decide for some json file with metadata that contains the additional information AND the table information for CSV, @xrotwang, is this possible in the current cldf spec?

• http://people.ds.cam.ac.uk/ssb22/gradint/lexconvert.html

Not sure yet, but they seem to have options to synthesize sounds, which would be cool, but the Q is, to which degree this might be feasible for us to do...

• ɺ̣ voiced retroflex lateral tap consonant
• ṛ > ɽr voiced retroflex trill consonant
• ʈθ̣ voiceless retroflex affricate

• https://doi.org/10.1044/2018_AJSLP-17-0100

The data contains some detailed consonants, so nice to add to clts as an example usage of linking things up. Thanks to @SimonGreenhill for pointing this out.

Creaky is a recognised feature on consonants here https://github.com/cldf-clts/clts/blob/master/data/features.tsv and I find d̰ here https://clts.clld.org/values. However, there was failure to parse:

• s̰̪ > s̪̰ creaky voiced voiceless dental sibilant fricative consonant

Note that s̪ is included at https://clts.clld.org/values

In parallel to the prenasalised feature, we can use a postposed nasal here.

From LAPSyD:

• kŋ > kⁿ with-nasal-release voiceless velar stop consonant
• kpŋm > kpⁿ with-nasal-release voiceless velar-and-labial stop consonant
• t̠n̠ > t̠ⁿ with-nasal-release retracted voiceless alveolar stop consonant
• t̠pn̠m > t̠pⁿ with-nasal-release retracted voiceless alveolar-and-bilabial stop consonant

From PHOIBLE:

• gm > gbⁿ with-nasal-release voiced velar-and-bilabial stop consonant
• km > kpⁿ with-nasal-release voiceless velar-and-bilabial stop consonant

I would propose to deal also with prestopped nasals here, (https://en.wikipedia.org/wiki/Pre-stopped_consonant).
From PHOIBLE:

• kŋ > kⁿ with-nasal-release voiceless velar stop consonant
• pm > pⁿ with-nasal-release voiceless bilabial stop consonant
• tn > tⁿ with-nasal-release voiceless alveolar stop consonant
• t̪n̪ > t̪ⁿ with-nasal-release voiceless dental stop consonant
• cn̠ > cⁿ with-nasal-release voiceless palatal stop consonant
• cɲ > cⁿ with-nasal-release voiceless palatal stop consonant
• dn > dⁿ with-nasal-release voiced alveolar stop consonant
• d̪n̪ > d̪ⁿ with-nasal-release voiced dental stop consonant
• ɖɳ > ɖⁿ with-nasal-release voiced retroflex stop consonant
• t̠n̠ > t̠ⁿ with-nasal-release retracted voiceless alveolar stop consonant
• ȶȵ > ȶⁿ with-nasal-release voiceless alveolo-palatal stop consonant

The attached book has an index of all sounds used in the Pikean tradition of transcriptions, phonemic analyses and practical orthographies. The sounds are accompanied by articulatory descriptions. Nice to have it in CLTS for the future.

Smalley_Pike_Tradition.pdf

This is from an older paper by Mielke, data not supplemented, as far as I know, but he had it online on his web-site. Article was a talk, in draft form:

• https://pdfs.semanticscholar.org/59c8/a6ff82f80d9519eb56c92da5d1ba86fda40d.pdf

This dataset is available on osf. It accompanies the article by Johansson et al. 2020.

The data offers transcriptions in different flavors, converting large lists into a unified IPA, that is not yet linked to CLTS, but should be. They also distinguish "cardinal sounds", to break down complexity. It is an excellent use-case for CLTS, especially also since we might use the breaking down to cardinal sounds as an alternative sound class system (!).

https://github.com/cldf-datasets/easterdaysyllablestructure/

After discussion with Cormac, these are the issues with Phoible graphemes that should be addressed:

• add dlʷ as an alias (voiced lateral affricate)
• add d̠ʓ as an alias ( ʒʲ )
• add d̪l̪ as an alias (voiced lateral affricate)
• add d̪ð̪ as an alias (voiced dental affricate)
• add d͇z͇ (new sound, alveolar diacritic)
• add i͓, u͓, and ɯ͓ as new sounds (frictionalised diacritic U+0353, might require new feature)
• decide on click notation/alias for kǀ̪, kǀ͓x, kǀ͓ˀ, kǀ͓ˠʰ, kǁ͓xʰ, kǁ͓ˀ, kǂ͓ˡ, kǂ͓ˡx, kǂ͓ˡʰ, kǃ̪, kǃ͓, k‼, k‼x, k‼xʼ, k‼ʰ, k‼ʰʼ, k‼ʼ, qǀ, qǀʼ, qǁ, qǁʼ, qǂ, qǂʼ, qǃ, qǃʼ, qʘ, qʘʼ, ŋǂ͓ˡ, ŋ̤ǀ, ŋ̤ǀ͓, ŋ̤ǁ, ŋ̤ǂ, ŋ̤ǂ͓ˡ, ŋ̤ɡǃ, ŋ̥ǀ͓xˀ, ŋ̥ǁ͓ʰ, ŋ̥ǂxˀ, ŋ̥ǂʰ, ŋ̥ǂˀ, ŋ̥ǂ͓ˡxˀ, ŋ̥ǂ͓ˡʰ, ŋ̥ǂ͓ˡˀ, ŋ̥ǃˠˀ, ŋ̥ǃ̠ʰ, ŋ‼, ŋ‼ʱ, ɡǀ͓x, ɡǁ͓, ɡǂ͓ˡ, ɡǂ͓ˡx, ɡ̤ǀ, ɡ̤ǀ͓, ɡ̤ǁ, ɡ̤ǂ, ɡ̤ǂ͓ˡ, ɡ̰ǀ͓x, ɡ̰ǂx, ɡ̰ǂ͓ˡx, ɡ̰ǃx, ɡ‼, ɡ‼x, ɡ‼xʼ, ɡ‼ʱ, ɢǀ, ɢǀqʰ, ɢǁ, ɢǁqʰ, ɢǂ, ɢǃ, ɢǃqʰ, ɢʘ, ʔŋǀ, ʔŋǁ, ʔŋǂ, ʔŋǃ, ʔŋʘ,
• correct entries for voiced alveolar lateral affricates kʟ̥ʼ and kʟ͓̥ʼ (we currently use pre-Unicode SIL extensions, k and kʼ, see here and here)
• add mw as an alias (labialized bilabial)
• add ndl as an alias (voiced lateral affricate)
• add aʲː as an alias (long diphthong)
• add ãʲ as an alias (long diphthong)
• add aʲ as an alias (diphthong)
• add eʲ as an alias (diphthong)
• add oʲ as an alias (diphthong)
• add õʲ as an alias (diphthong)
• add i̩ː as an alias (even though arguably tautological)
• add dl as an alias (voiced lateral affricate)
• add tl as an alias (voiceless lateral affricate)
• decide on "less rounded" diacritic for consonants, particularly for w̜ and w̜ʲ
• add co-articulated stops ŋmʲ, ŋmʷ, ŋmˤ, ŋɡmb, ŋ̥m̥
• decide what to do with missing tone that include other segmental or supra-segmental features: ˥˧̰, ˥˩˩˥, ˦ˀ, ˦˥̰, ˦̰, ˧˨ˤ, ˧˨̤, ˧˩̤, ˧˩̰, ˨˩̤, ↓˦˨, ↓˦↓˦
• add ŋm and ŋmɡb (co-articulated nasal stops)

In the same round, the following issues could be addressed as well:

• #25
• #13

nm nm
n̠m n̠m
ɗ̪ʲ ɗ̪ʲ
ɽr ṛ
tx tx
βɾ βɾ
ŋ̊ǀh̃ ŋ̥ǀh̃
ŋ̊ǀx ŋ̥ǀx
ŋ̊ǀʔ ŋ̥ǀʔ
ŋ̊ǁh̃ ŋ̥ǁh̃
ŋ̊ǁx ŋ̥ǁx
ŋ̊ǁʔ ŋ̥ǁʔ
ŋ̊ǂh̃ ŋ̥ǂh̃
ŋ̊ǂx ŋ̥ǂx
ŋ̊ǂʔ ŋ̥ǂʔ
ŋ̊ǃh̃ ŋ̥ǃh̃
ŋ̊ǃx ŋ̥ǃx
ŋ̊ǃʔ ŋ̥ǃʔ
ɗ̪ ɗ̪
nmʲ nmʲ
n̠mʲ n̠mʲ
ɾ̪ ɾ̪
ᵗl̪ d̪l̪

see #59 for the same explanation.

following up discussion from here.

It seems we don't have this sound yet
https://en.wikipedia.org/wiki/Retroflex_trill