^ abcdefghijklmnopqrstWord-finally, the voiceless-voiced distinction in the obstruent pairs [p–b, t–d, k–ɡ, ts–dz, tʃ–dʒ, f–v, s–z, ɕ–ʑ, ʃ–ʒ, χ–ʁ] is neutralized, mostly in favour of the voiceless obstruents (but see the table titled Suprasegmentals) (Gilles & Trouvain (2013:68)).
^The /ʀ/ phoneme is realized as a trill [ʀ] when pre-vocalic within the same word, as well as often when non-prevocalic in French loanwords (Gilles & Trouvain (2013:68, 71)).
^ abcdefBoth [ɕ, ʑ] and [χ, ʁ] are allophones of /χ, ʁ/. [χ, ʁ] occur after back vowels, whereas [ɕ, ʑ] occur in all other environments, although the voiced [ʑ] occurs only in a few words. Speakers increasingly do not distinguish between [ɕ, ʑ] and [ʃ, ʒ] (Gilles & Trouvain (2013:68–69)).
^ abcdeWhen non-prevocalic within the same word, the /ʀ/ phoneme has quite a long list of allophones:
After short vowels, the non-prevocalic /ʀ/ is realized as a fricative, either voiced [ʁ] or voiceless [χ], depending whether the following consonant is voiced or voiceless;
The unstressed, non-prevocalic sequence /eʀ/ is realized as an open vowel [ɐ];
/ʀ/ is fully absorbed into the preceding /aː/ in the non-prevocalic sequence /aːʀ/, so that e.g. Paart, Taart and waarm are pronounced [paːt], [taːt] and [vaːm], i.e. as if they were spelled Paat, Taat and waam;
^ abcAlthough these sounds do not occur in native Luxembourgish vocabulary, almost all speakers pronounce them correctly.
^ abApart from being the main realization of the phonemes /b, d, dz, ɡ, v, z, ʒ, dʒ/, [b, d, dz, ɡ, v, z, ʒ, dʒ] occur as word-final allophones of both /p, t, ts, k, f, s, ʃ, tʃ/ and /b, d, dz, ɡ, v, z, ʒ, dʒ/ (in this position, some scholars may analyse both of these sets as /p, t, ts, k, f, s, ʃ, tʃ/) if the next word begins with a vowel and is pronounced without a pause. [ʁ, ʑ, bv] also occur as allophones of /χ, χ, pf/ in the same environment, but note that [bv] does not occur in other circumstances, and that /h/ is always voiceless [h], as it cannot even appear in the syllable coda, and thus is not subjected to the resyllabification and voicing. In this context, the final voiceless obstruents are not only voiced, but also resyllabified, i.e. moved to the onset of the first syllable of the following word. Therefore, a more phonetically accurate transcription of sech eens would be [zə‿ˈʑeːns] or simply [zəˈʑeːns] (Gilles & Trouvain (2013:68, 72)), although we transcribe it [zəʑ‿ˈeːns] for simplicity.
^Word-initial /pf/ tends to merge with /f/, although that is not how we transcribe it here. In other positions, /pf/ is retained as a true affricate.
^ ab[ə] and [e] are allophones of a single phoneme /e/. [e] appears before velar consonants, and [ə] elsewhere. Note that unlike Standard German, [ə] appears in both stressed and unstressed syllables, and that unstressed sequences of [ə] and a sonorant (excluding the unstressed, non-prevocalic sequence /eʀ/, which is realized as [ɐ]) do not form syllabic sonorants (Gilles & Trouvain (2013:68 and 70)).