Consonant clusters

[FredericBlum]

There are a couple of very long consonant clusters which should be checked. @SimonGreenhill Maybe we can provide some easy fixes for those through the orthography profiles?

Language_ID	Length	Cluster	Words
Nalik_368	3	n g t	r a n g t a n g a n // rangtangan // m a n g t u l // mangtul
HaliaSelau_647	3	r n g	n g o r n g o r o // ngorngoro
Patpatar_178	3	r n g	n g a r n g a r // ngarngar
Avava_456	3	r n g	n g a r n g a r // ngarngar
AngavaeDixonReef_1238	4	t n t n	t n t n // -tntn
Avava_456	4	t n t n	t n t n // tntn
Neveei_484	4	t n t n	t n t n // tntn
NindeLawa_1217	4	t n t n	t n t n // -tntn
NombotkoteUsusMelkin_1230	4	t n t n	t n t n // -tntn // t n t n // -tntn
Kokota_365	4	ɣ ŋ ɣ ŋ	ɣ ŋ ɣ ŋ a u // ɣŋɣŋau
Mangseng_1335	4	k n g r	ŋ r e k n g r e k // ŋrekngrek
Nahavaq_453	4	n v y/j n	t i n v y/j n // tinvyn
Roria_1006	4	n ts n t	n a n ts n t a r a ⁿp // nantˢntaraᵐp
SakaoPortOlry_230	4	β y/j l j/dʒ	a β y/j l j/dʒ ɛ + ɣ // aβyljɛ-ɣ // a β y/j l j/dʒ a + n // aβylja-n
SakaoPortOlry_230	4	r s y/j m	k ɔ r s y/j m // kɔrsym
Iaai_471	5	y/j ɲ k y/j m	y/j ɲ k y/j m e // yñkyme
Nahavaq_453	5	m b y/j s y/j	m b y/j s y/j // mbysy-
Nahavaq_453	5	m b y/j n ʔ	l e m b y/j n ʔ a j/dʒ // lembynʔaj
NatiWindua_1229	5	β y/j l y/j n	n o + s i β y/j l y/j n + v a r a n // no-siβylyn_varan
SakaoPortOlry_230	6	β c/ts ɣ β c/ts ɣ	β c/ts ɣ β c/ts ɣ // βcɣβcɣ
SakaoPortOlry_230	6	β y/j t β y/j t	β y/j t β y/j t // βytβyt
SakaoPortOlry_230	6	β y/j l β y/j l	β y/j l β y/j l // βylβyl
Nahavaq_453	8	m b y/j ʔ m b y/j ʔ	m b y/j ʔ m b y/j ʔ // mbyʔmbyʔ

[SimonGreenhill]

@tihomirrangelov -- can you please take a look at these?

[tihomirrangelov]

• The first four, involving 'n g', my guess is these are not two segments, but one, namely ŋ, in some cases the consonant clusters span across morpheme boundaries, as some of these examples clearly involve reduplication, but then there is nothing unusual.
• AngavaeDixonReef_1238. I was able to verify this against the sound file in VV. It seems like the transcription is wrong. I would transcribe this as [tu-titn̩] (with the final n syllabic).
• Avava_456. This language allows up to five-consonant clusters word finally, but the sonorants in such clusters are usually syllabic (Crowley 2006). So the proper transcription should be [tn̩tn̩] where these are two syllables. This is also my impression from the audio in VV. The same seems to be the case with the other three varieties with this cluster.
• Kokota_365. Probably syllabic ŋ
• Mangseng_1335. Clearly the 'n g' should be ŋ
• Nahavaq_453. I couldn't find this example in VV to hear it, but if we choose 'y', rather than 'j', then this is phonotactically very simple, i.e no consonant clusters to write home aboutː CVC.CVC
• Roria_1006. There are no descriptions or sound files for this languages, but looking at the wordlist, my best guess is that 'n ts' is one prenasalized affricate segment [ⁿt͡s]. 'nt' could be either a [ⁿd] or a n+t cluster in a syllable onset. Possibly, but less likely, a prenasalized t.
• SakaoPortOlry_230. /y/ is a vowel here.
• Iaai_471. /y/ is a vowel
• Nahavaq_453. /y/ is a vowel. I would also bet that /mb/ is actually a prenasalized b than a cluster.
• NatiWindua_1229: /y/ is a vowel.
• SakaoPortOlry_230. βcɣβcɣ - this is probably a mistake in transcription. I have collected data on this language and for 'good' I have βɔɣβɔɣ, where ɔ is a vowel. In the other two examples, /y/ is a vowel.
• Nahavaq_453. /y/ is a vowel. I also bet that mb is a prenasalized b, rather than a cluster.

[FredericBlum]

Most of those cases (ŋ, y) can be adopted through the orthography profiles. Do you have updated versions on some database, or are the ones in this repository the most up-to-date ones?

Can I change the /mb/ occurrences to prenasaliyzed /b/ in the languages with such clusters? I would then tag you in order to revise the mappings and double-check on the changes.

How can we fix the transcription errors? Manual resolving of transcription errors here and in the database on your side at the same time?

[tihomirrangelov]

Regarding mb, as long as a language has a prenasalisation contrast, it is a safe bet to go for a prenasalised b.