I created this account because I want to add phonemic versions of Luxembourgish IPA. I saw one or two users do this and at first I thought it unnecessary, but then I thought: Why not, but let's do it right then!
These are the points that I've thought half-baked in the phonemic IPA around here.
Long vowels and diphthongs
- [aː] vs. [ɑ] is either /aː/ vs. /a/ or alternatively /a/ vs. /ɑ/. I think the former is preferable for several reasons.
- Land – /lant/, [lɑnt]
- [ɛː] (in French and German words) should be /æː/ (long equivalent of the native vowel /æ/). The realisation seems to tend towards [æː] anyway. This loan phoneme /æː/ does not exist before /ʀ/, where [ɛː] is an allophone of the native vowel /eː/.
- It doesn't seem justified to analyse [e] and [ə] as allophones of /e/ but at the same time to distinguish not only /ə/ but even /ɜ/ in diphthongs. Diphthongs should be noted as combinations of the existing monophthongal phonemes:
- /ai̯/, [ɑɪ̯] — /au̯/, [ɑʊ̯]
- /æi̯/, [æˑɪ̯] — /æu̯/, [æˑʊ̯]
- /ei̯/, [ɜɪ̯] — /eu̯/, [əʊ̯]
- /ie̯/, [iə̯] — /ue̯/, [uə̯]
- Vowel length is neutralised (1.) in stressed syllables word-finally and before unstressed vowels, (2.) in all unstressed syllables (i.e. with neither primary nor secondary stress). Under 1, such neutralised vowels are phonetically long; under 2, they are short.
- The quality oppositions within the phoneme pairs /a/, /aː/ and /e/, /eː/ are also neutralised word-finally and before unstressed vowels. They are phonetically [a(ː)] and [e(ː)]. (Length depending on the stress as above.)
- Eifeler-Regel variants like da [dɑ] (from dann [dɑn]) are no exception to this because phonemically both variants can be analysed as /da(n)/.
- The only problem arises with the three reduced monosyllables de (“you”), se (“she, they”), ze (“to”). I propose to analyse these as phonemically /d/, /z/, /ts/; and in fact, they usually lose their final [ə] before vowels and also sometimes before liquids.
- Word-internally, however, there's an—albeit rare—opposition between unstressed [a]/[ɑ] and [e]/[ə], so [a], [e] must be distinguished as /aː/, /eː/.
- Lycéesschüler – /ˈli.seːsˌʃyː.leʀ/, [ˈli.sesˌʃyː.lɐ]
Vowels with coda /ʀ/
- /ʀ/ remains consonantal after short vowels (except unstressed /e/). This is chiefly restricted to non-native words. /ʀ/ becomes [χ] before unvoiced consonants. This is still distinct from /χ/ because it doesn't alternate with [ɕ].
- Sport – /ʃpoʀt/, [ʃpoχt]
- The same is also possible after /aː/, but /ʀ/ may also be deleted in this case, so it should be given in brackets.
- /eːʀ/ becomes [ɛə̯].
- Häerz – /heːʀts/, [hɛə̯ts]
- [ɛə̯] is also a non-obligatory allophone of [ɐ] (= unstressed /eʀ/) in word-initial position.
- After /ie̯/ and /ue̯/, stem-internal /ʀ/ is now always deleted and should thus not be noted at all. The spelling has become purely etymological.
- kuerz – /kue̯ts/, [kuə̯ts]
- Stem-finally, the disyllabic pronunciations [iː.ɐ], [uː.ɐ] are possible. Since /ie̯/, /ue̯/ cannot be stem-final without an underlying /ʀ/, these disyllabic forms may be analysed as (non-obligatory) allophones of the same diphthongs, used in stem-final position (or: before morpheme boundaries). However, since this is not a purely phonological but partly grammatical analysis, I've thought it preferable to use /iːʀ/, /uːʀ/ in these cases.
- It is true that some speakers may still distinguish /iːʀ/, /uːʀ/ also stem-internally, but this is now a dialectal feature (i.e. not part of "general Luxembourgish" or "Koiné"). These speakers might also have a phonemic /æː/ in native words or other features that we do not indicate.
- Other diphthongs and long vowels + /ʀ/ exist stem-finally only. Such combinations are phonetically disyllabic. Monosyllabic realisations are possible but are merely due to elision in fast speech (to which all unstressed syllables are liable). Unfortunately, the straightforward notation "vowel + /ʀ/" would put the respective words in the wrong categories concerning number of syllables. So to mend this I've added (e), where the brackets may indicate that the vowel is actually not phonemic.
- When a vowel is added, these disyllabic pronunciations usually become monosyllabic again, though they may also remain disyllabic. The spelling has no bearing on this and there's no phonemic distinction.
- In compounds, component-final obstruents are resyllabified with following vowels and undergo allophonic voicing. Phonemically they are unvoiced (or rather: neutralised) as in the coda.
- Word-final -n must be given in brackets when deletable (Eifeler Regel), e.g. dann /da(n)/, [dɑn] and da /da(n)/, [dɑ], as mentioned above. Although the spelling distinguishes these two pronunciations, they are merely allophonic variants of the same word and therefore have the same phonemic notation. However, deletable -n and non-deletable -n are phonemically distinct because not every -n is deletable; hence the brackets. Only in unstressed word-final /en/ is the deletion universal, so the brackets can be left out. — Strictly speaking, an /n/ would also have to be added in compounds like Bouneschlupp /ˈbeu̯nenˌʃlup/, [ˈbəʊ̯nəˌʃlup], but I suppose that would be a bit pedantic.
Calques from German
I've also begun to add a lot of etymologies for Luxembourgish words derived from German. With simplexes you can usually see whether it is inherited or not, but with derivatives and compounds it's more difficult.
Generally, any Luxembourgish compound or derivative to which there is an etymological equivalent with the same meaning in German must be considered a loan translation unless the same word is attested with the same sense already in Middle High German (before 1400). However, I've excluded words which are attested in Early Modern German (before 1650) and which at the same time are so everyday and semantically so obvious that they could be formed independently (e.g. Iessläffel). Another exception are, of course, such hyper-productive affixes like mat-, which can be added randomly to any verb.
Even compounds that are in fact attested in Middle High German are nevertheless likely to be loan translations, but it's hard to prove. The only two hints are if the "Luxemburger Wörterbuch" or the "Rheinisches Wörterbuch" say "nach dem Nhd." (after modern German) or "Neol." (neologism, which usually means the same), or if another dictionary mentions that the word was regional in Middle High German and spread only later on.
All learned terms (like Insekt) are also borrowed through German, unless they are borrowed through French (if there's a phonetic or semantic hint at that).
By applying these standards it will be shown that the bulk of Luxembourgish vocabulary is derived from German (one way or another), but that's just the way it is with a language only recently standardised.