Phonology

Consonants

Katnae has the following consonant phonemes, with common allophones and marginal phonemes given in parentheses:

The palatal stop/affricate and postalveolar fricative are allophones of syllable-final stops before a /j/ in various dialects, with postalveolar realizations being by far the most common. The glottal stop, similarly, occurs only as an epenthetic sound at the beginning of a vowel-initial word and as an allophonic realization of a word-final stop before a pause.

All stops are unaspirated in all positions.

Consonant Length

Pairs of two of the same consonant are pronounced long, but there is no evidence in favor of treating these as separate phonemes rather than simple sequences — on the contrary, many of these geminated consonants are the product of assimilation of one member of a cluster of two non-identical consonants to the other, and the presence of clusters of non-identical consonants (as in the name Katnae itself) suggests that consonant clusters, while subject to very narrow phonotactic restrictions, are not impossible. There also does not appear to be any difference in pronunciation between a "long" consonant that is demonstrably a product of assimilation and one whose etymology cannot be traced.

The glides /j/ and /w/ do not occur geminated, but sequences of a diphtong ending in /i/ or /u/ followed by the corresponding glide do occur and contrast consistently with the glide on its own.

Allophonic Voicing

Single stops and fricatives may occasionally become voiced between two vowels or sonorants, although this is far from regular and does not occur for all speakers in all situations. Voicing is never phonemic, and does not play a significant role in distinguishing single consonants from geminates (length is still overwhelmingly the primary distinguishing factor).

Sonorants are always voiced, and display little or no devoicing, even when adjacent to a stop, fricative, or word boundary.

Vowels

The actual vowel inventory is quite straightforward, if analyzed in the simplest way possible:

These five phonemes are realized very similarly to the five cardinal vowels that their IPA symbols correspond to, although the mid vowels are somewhat lower than one might expect, closer to [ɛ ɔ] for many speakers.

Vowel Length

Katnae undeniably contrasts short and long vowels at the level of realization. However, the argument for treating short and long vowels as separate phonemes as opposed to sequences of two of the same vowel is fairly weak — there are morphemes that can be analyzed as beginning with a chroneme, but these can just as easily (and often more correctly — more on that later) be considered to begin with an underspecified vowel that adopts the quality of the preceding vowel.

In contrast, there are quite a few strong arguments for phonetically long vowels to be considered sequences. Firstly, the set of diphthongs present in the language covers the entire space of possible two-vowel combinations, removing any need for pairs of single-mora vowels to be treated as underlying phonological units. Furthermore, the language's long consonants are frequently (although not universally) etymologically derived from sequences of non-identical consonants affected by assimilation, and all exceptions can be trivially analyzed as sequences of two identical consonants — hence there is no need to posit chronemes anywhere else in the language. And lastly, there is no difference in quality between short and long vowels. Thus, we generally assume that long vowels are simply sequences of two identical single-mora vowels.

If we were to accept long vowels as separate phonemes, the table would expand to the following:

Diphthongs

Any pair of vowels may occur adjacent to one another to form a diphthong. Diphthongs are realized as a relatively smooth transition from the first vowel quality to the second, without the reduction of the offglide found in English and many other SAE languages — think something more like Finnish or Japanese. As such, there is no set of accepted diphthongs outside of which vowel sequences will be treated as two separate syllable separated by hiatus.

Vowel sequences in hiatus do not occur, either within words or across word boundaries. A word ending in a vowel and a word beginning with a vowel will be separated by a non-phonemic glottal stop.

Phonotactics

Basic Katnae syllable structure is (C)V(V)(C) — that is, an optional consonant, followed by a vowel or diphthong (including long vowels), and optionally followed by a consonants. However, there are a number of restrictions to be aware of:

• Fricatives and the glides /j/ and /w/ do not occur syllable-finally, ever.
• Syllable-final nasals assimilate in place of articulation to a following stop or fricative, and entirely to a following sonorant.
• Syllable-final stops become identical to a following stop, or are realized as a glottal stop word-finally before a pause.
• Syllable-final stops assimilate to a following fricative, sonorant, or glide in a highly dialect-dependent manner.

Assimilation of Syllable-Final Stops to Fricatives and Sonorants

Typically, a syllable-final stop assimilates to the place of articulation of a following fricative or sonorant, but remains a stop. However, in certain dialects, a syllable-final stop assimilates all the way to some subset of fricatives, nasals, and liquids, resulting in a simple geminate consonant. See the section on dialectal variation for isoglosses and a more in-depth treatment.

Assimilation of Syllable-Final Stops to Glides

A sequence of a syllable-final consonant and /j/ is usually realized as [c:], [tʃ], or [ʃ:] depending on dialect. This is the source of the original (false) assumption that /ʃ/ was a phoneme in this language.

Similarly (and more straightforwardly), a sequence of a final consonant and /w/ near-universally yields [p:].

Word Boundaries and Assimilation

Word-final stops usually assimilate in place of articulation to the first sound of the following word, if that word begins with a consonant. However, outside of actual compounds, the initial consonant of the second word will not be affected by this assimilation.

If a word-final stop occurs immediately before a word beginning with a vowel, it can be realized either as a glottal stop or as its underlying form, depending on dialect (isoglosses and/or elaboration of dialectal variation to come).

There is likely a word-initial, non-phonemic glottal stop present in at least some dialects in words that begin with a vowel, although again, the dialectal variation hasn't been fully worked out yet.

Stress and Prosody

Lexical stress is universally word-initial, without regard for syllable weight.

Orthographic Notes

The phonemes /ŋ/ will be spelled <ng>, and a word-final glottal stop will be written as <t>. There may be some variation as to whether a word-final consonant that has assimilated to the first consonant of the next word will be written as it is pronounced or as <t>, but the <t> grapheme is more common.

All other phonemes will be spelled as their IPA symbols.

There will be a native orthography, although it is currently un(der)developed. When ready, it will have its own section.