Nahtak family Scratchpad (NP: PN phonology)
Anthologica Universe Atlas / Forums / Department of Creativity / Nahtak family Scratchpad (NP: PN phonology)

? dhok posts: 235
, Alkali Metal, Greenfield, United States
This is the umpteenth iteration, give or take, of the project affectionately referred to as 'Shit Algonquian'. We'll start, as usual, with phonology.

*p	*t	*ts	*tʃ	*k	*kʷ	*ʔ
*m	*n			*ŋ
	*θ	*s	*ʃ			*h
	*l		*j		*w

*ts *tʃ *ʃ *j are written c č š y, respectively, in accordance with usual Americanist practice.

*i *i:	*ɨ *ɨ:	
*e *e:		*o *o:
	*a *a:

Long vowels are transcribed with a macron (but only if unaccented; see next section): ā ē ī ō ɨ̄.


Proto-Nahtak is reconstructed as having a system of pitch-accent. The accent may fall on any of the last three morae; long vowels are considered to have two morae.

The accent in a word may be either lexically or morphologically determined. If lexically determined (mainly nouns and particles), it will attach to a particular vowel (or mora of a particular long vowel), and moved only when forced to, usually by a suffix. If morphologically determined, it will attach to a certain mora- the final mora, penultimate mora, or antepenultimate- and then stay in that position regardless of what vowel is assigned to it.

A penultimate syllable with a short vowel but a coda consonant (that is, one followed by a true cluster- see Phonotactics below) is considered to be two morae long. This has no effect on the type of accent that can fall on the vowel, but it prevents the accent from landing on the antepenultimate syllable.

An accented short vowel is written with a grave accent: <è>. A long vowel with an accented first mora has a falling tone, and is written with a circumflex: <ê>. A long vowel with an accented second mora has rising tone, written with an acute accent: <é>. An unaccented long vowel is written with a macron (<ē>), and an unaccented short vowel of course will not be marked with any sort of diacritic.

For an example of lexical accent, we can consider the noun *θāmè 'hawk'. Its accent is attached to the final short /e/, and this remains even in the presence of suffixes: *θāmèwa 'hawks'. On the other hand, the independent indicative mood of the verb has morphologically-determined penultimate accent, and this does change when suffixes are added: thus, from the root *mač- 'cook', we derive the 1sg form *nɨmàčen 'I cook it', but the 1pl exclusive form *nɨmačèhken.

Some suffixes cause a preceding short vowel to lengthen; in this situation, a lexically determined accent is automatically assigned to the newly long vowel's first mora. Thus, the singular obviative of *θāmè is *θāmêta, with suffix *-:ta.


The phonotactic structure of PN is reasonably simple. All nuclei must be a vowel, which may be short or long; any singular consonant except , or none may begin a word (but no clusters); words may only end in a vowel, *n, *k, *t, or *s.

Medially, clusters of up to two consonants may appear, as follows:

py ty cy    ky     my ny    θy sy    hy ly
   tw cw čw           nw ŋw θw sw šw hw lw

ʔp ʔt ʔc ʔč ʔk ʔkʷ ʔm ʔn ʔŋ ʔθ ʔs ʔš ʔl 
mp nt nc nč ŋk ŋkʷ          nθ ns nš nl                        
θp          θk θkʷ                   
sp st sc    sk skʷ                  
šp št    šč šk škʷ                  
hp ht hc hč hk hkʷ hm hn hŋ hθ hs hš hl
lp lt lc lč lk lkʷ lm ln lŋ     

The clusters ending in a semivowel- *w or *y- pattern differently from the others, and are called 'semi-clusters', as opposed to 'true clusters'. In particular, semi-clusters do not cause a short penult to act long for the purposes of accent placement, and they differentiate cluster-initial nasals, which otherwise all act identically.

Right, we'll leave that be and get to the grammar tomorrow...or later tonight, I haven't decided...