Lardil phonology

Consonants
The consonant inventory is as follows, with the practical orthography in parentheses.

Lardil's consonant inventory is fairly typical with respect to Australian phonology; it does not distinguish between voiced and unvoiced stops (such as b/p and g/k), and features a full set of stops and nasals at six places of articulation The distinction between ‘apical’ and ‘laminal’ consonants lies in whether the tip (apex) of the tongue or its flattened blade makes contact with the place of articulation. Hale's 1997 practical orthography has ‘k’ for  in order to disambiguate nasal+velar clusters (as in wanka ‘arm’ ) from instances of the velar nasal phoneme  (as in wangal ‘boomerang’ ) and to avoid suggesting -gemination in  clusters (as in ngangkirr ‘together’ ). The sounds represented by the digraphs ‘nh’ and ‘ly’ are not common in Lardil, but speakers perceive them as distinct, respectively, from and, and they do occur in some words (e.g. minhal ‘burnt ground’, balyarriny [title of a social subsection]).

Vowels
Lardil has eight phonemically distinct vowels, differentiated by short and long variants at each of four places of articulation. Phonemic vowel length is an important feature of many Australian languages; minimal pairs in Lardil with a vowel length distinction include waaka/waka ‘crow’/’armpit’ and thaldi/thaldii ‘come here!’/’to stand up’. Long vowels are roughly twice as long as their short counterparts.

Stress
Primary word stress in Lardil falls on the initial syllable, and primary phrase stress on the final word in the phrase. These stress rules have some exceptions, notably compounds containing tangka ‘man’ as a head noun modified by a demonstrative or another nominal; these expressions, and other compound phrases, have phrase-initial stress.

Common alternations (consonants)

 * /rr ~ d/, _#
 * The distinction between /rr/ and /d/ is lost word-finally, as in yarburr ‘bird/snake’, which may be realized as [yarburr] or [yarbud], depending on the instance.


 * /d ~ n, j ~ ny/, _N
 * /d/ and /j/ may assimilate to a following nasal, as in bidngen > binngen ‘woman’, or yuujmen > yuunymen ‘oldtime’.


 * /r ~ l/, #_
 * Word-initial /r/ is often expressed as /l/; as with /rr ~ d/, either (e.g.) [leman] or [reman] may be heard for ‘mouth’.

Word-final phonology
In addition to the common phonological alterations noted above, Lardil features some complex word-final phonology which is affected by both morphological and lexical factors.

Augmentation acts on many monomoraic forms, producing, for example, /ʈera/ ‘thigh’ from underlying *ter.

High vowels tend to undergo Lowering at the end of bimoraic forms, as in *penki > penke ‘lagoon’. In several historical locative/ergatives, Lowering does not occur. It does occur in at least one long, u-final stem, and it coexists with the Raising of certain stem-final /a/s.

In some trimoraic (or longer) forms, final, underlying short vowels undergo Apocope (deletion), as in *jalulu > jalul ‘fire’. Front-vowel Apocope fails to occur in locatives, verbal negatives, many historical locative/ergatives, and a number of i-final stems such as wan̪t̪alŋi ‘fish sp.’ Back-vowel Apocope also has lexically-governed exceptions.

Cluster Reduction simplifies underlying word-final consonant clusters, as in *makark > makar ‘anthill’. This process is ‘fed’ in a sense by Apocope, since some forms that would otherwise end in a short vowel arise as cluster-final after Apocope (e.g. *jukarpa > *jukarp > jukar ‘husband’).

Non-apical Truncation results in forms like ŋalu from underlying *ŋaluk, in which the underlying form would end in a non-apical consonant (i.e. one not produced with the tip of the tongue). This process is also fed by Apocope, and seems to be lexically governed to an extent, since Lardil words can end in a laminal; c.f. kakawuɲ ‘bird sp.’, kulkic ‘shark sp.’

In addition to the dropping of non-apicals, a process of Apicalization is at work, giving forms such as ŋawit from underlying laminal-final *ŋawic. It has been proposed that the process responsible for some of these forms is better described as Laminalization (i.e. nawit is underlying and nawic occurs in inflected forms), but Apicalization explains the variation between alveolar /t/ and dental /t̺/ (contrastive but both Apical) in surface forms with an underlying non-apical, and does not predict/generate as many invalid forms as does the Laminalization model.