The Sound System of aćharêtâ'

Articulatory Location
Lip Tongue Nasal Glottis Larynx
  tip up tip down      
front back front back retracted
Stricture Type dental alveolar palatal velar pharyngeal
Con-
son-
ants
Ob-
stru-
ents
Affricates ć č     [vls]
  ǰ
Stops p t   k q [vls]
b d g
Spirants f s š x [vls]
  z   ǧ h
Sono-
rants
Flaps upward     r    
forward
Lateral l  
Approx-
imants
w   y
Nasal
Stops
m n   [nas]
Vowels Short Close u   i   ˜
Opened o   e a
Long Close û   î  
Opened ô   ê â
Pitch Accent primary    ʹ 
secondary    ʹ 
Terminal
Contours
full drop   .
steady   ,
half drop   ?

Table 1. The Phonemes of Aćharêtâʹ.

([vls] [voiceless] and [nas] [nasalized] represent coproduced components of the sounds listed to their left in the row.)

Phonemic Inventory. The symbols in Table 1 stand for the distinctive speech processes that produce the sounds of Aćharêtâʹ. These symbols are used in lieu of customary phonetic symbols, which are not yet handled by most web browsers.


(The nature of the phonatory and laryngeal processes of Aćharêtâʹ have been determined from acoustic, visual, and proprioceptive observations of speech. The absence of more sophisticated observational techniques in the difficult field conditions of the region leaves the precise anatomical nature of the following tensing processes undetermined.)


Phonation and Laryngeal Accent.  The normal unaccented phonation of vowels in Aćharêtâʹ is produced with a back tensing of the larynx, possibly with a degree of concomitant posterior voicing.*  Throughout the vowel this process produces a fundamental frequency (F0) that may be more or less progressively lowered, the degree of lowering corresponding to the intensity of the speech. There is also a slight megaphonic effect. Back Tensing strengthens slightly over utterance segments, producing a characteristic step-down lowering of F0 over the length of an utterance (see Fig. 1).

Accented phonation is produced with front tensing, an upward pulling of the larynx that occurs along with normal back-tensed phonation. This process produces a raised F0. The range of pitch between accented and unaccented vowels is somewhat narrow, because of the restriction of larynx raising caused by the concomitant pull of back tensing.

Accented short vowels are indicated by the accent mark (ʹ) preceding the vowel's symbol. The point during the vowel of highest F0 varies under the influence of speech intensity and terminal contours.

The position of accent on long vowels is distinctive. On long vowels with beginning accent, indicated by the accent mark placed before the vowel, front tensing occurs at the beginning of the vowel and then drops, producing a falling pitch throughout the vowel (see Fig. 2). On long vowels with ending accent, indicated by the accent mark placed after the vowel, front tensing occurs at the end of the vowel, while the entire vowel sustains normal back tensing. The result is a level, falling-rising, or rising pitch across the vowel, depending on the intensity of speech (see Fig. 3). Thus there is a contrast between, e.g., ending-accented râʹt 'night and day [24-hour period]' vs. beginning-accented rʹât 'blood'.

Almost all long vowels with ending accent occur in the final syllable of a word. When such words take an inflectional suffix, the vowel of the suffix normally receives the accent, in keeping with the position of the accent at the end of the word; e.g., kâʹl 'year', plural kâlʹa. The two exceptions noted are hâʹt 'hand', plural hʹâta, and jhâʹṭ 'hair', plural jhʹâṭa.

In a few forms ending-accented long vowels appear penultimately. Most of these, like the two exceptions just noted, contain aspirated voiced consonants which may have influenced their development: dhrîʹṣṭo 'seen', lhôʹko 'little', mhôʹro 'sweet', ghâʹnu 'big'. The interrogative pronouns kôʹṛo, kôʹso, etc. 'which one' are compounds of kôʹ 'who' plus a deictic pronoun. Only ǰâʹbli 'runny sap' is indeed anomalous; it appears to be based on a Pashto loanword.

Accented phonation may occur on the penultimate or the final syllable in a word. Antepenultimate accent apparently occurs only when suffixes are added to a word; e.g., šʹâlak 'locust', plural šʹâlaka, mhʹâruc̣i mʹuṭ 'mulberry tree', cf. mhârʹôc̣o 'mulberry'.

A few words contain an accented vowel immediately following an unaccented vowel: dhiʹûṛo 'daughter's son', lôbiʹa 'string bean', etc. In cases where a long vowel precedes another vowel, the accent mark indicates beginning accent on the following vowel, rather than an ending accent on the preceding vowel: kôṭaghâʹûčo 'male from kôṭaghâʹ', etc.

Secondary Accent.  The presence of a secondary accent appears confirmed by full waveform plots of speech, which show a greater amplitude or duration for vowels that are subjectively of secondary prominence. Such vowels do not show the raised F0 of laryngeal front-tensing; they are probably produced by increased glottal opening.

Vocalism.  The articulatory processes of mandibular opening, lingual fronting, and labial rounding produce the five contrasting vocalic qualities of Aćharêtâʹ (see Table 1). To these may be added the process of lengthening^ over a vowel's symbol), which produces a contrast between long and short (unmarked) vowels; e.g.,

Laryngeal Effect on Vocalism.  There is a degree of linkage between the tensing of the larynx and the position of the base of the tongue: normal laryngeal back-tensing tends to pull the tongue back and down, while laryngeal front-tensing pushes the base of the tongue up. Back-tensing restricts the fronting of the short vowels e and i and the opening of a, while accentual front-tensing has engendered vocalic changes at times during the evolution of Aćharêtâʹ.

Vocalic Processes within the Word.  Both laryngeal accent and the anticipation of the fronting of a following i ("umlaut") have at different times determined the vocalic makeup of modern words in Aćharêtâʹ. The vowels in such words contrast with those in morphologically related forms that have not been affected by these processes, giving rise to vocalic alternations such as:

The determining processes of accent and anticipation apparently spawned processes of vocalic change in the following stages:1
There are exceptions to each process; none is automatic in the current language. Chronological details of these processes of change are still being sorted out to best account for the exceptional cases. The processes are discussed below.

Closing.  Open, beginning-accented long vowels became closed under the tongue-raising effect of laryngeal front-tensing, while corresponding ending-accented or unaccented long vowels remained open; e.g.:

Fronting.  The fronting process of the close front vowel (i) was anticipated on an â in the preceding syllable, producing ê:

Such anticipation apparently did not occur if the preceding vowel was originally short: Other exceptions include ǰʹâmi 'mandible' and ac̣hʹâṛi 'walnut tree'.

Rounding.  In addition to the direct effect of laryngeal tensing on vowels, there was a sympathetic process of lip-rounding on beginning-accented long ʹâ, which became ʹô. Ending-accented or unaccented â remained; e.g.:

Most instances of ʹô in the current language arose from this process in earlier times.

Vocalic Tensing ("Lengthening").  During vocalic tensing the articulatory processes are strengthened, producing increased opening or fronting and lengthening the vowel. Accented short vowels doubled in length as the tongue was pushed farther up and out, while unaccented or sequential short vowels remained short, as seen in the alternations of vowels in inflected forms:

Within such forms the same total vocalic strength, as measured in moras, is maintained in both inflected and uninflected words, the length of a tense vowel being equivalent to that of two short vowels.

However, there are many morphologically related forms that show no such alternations, having derived from earlier forms with tense vowels:

Sequences of Vowels. 

Nasalization. 

Consonantism. 

Voicing. Consonants may be voiced (Figures 5 and 7) or voiceless (Figures 8 and 9). Voiced consonants have the default back-tensed voicing, slightly strengthened to compensate for the back pressure generated by the oral stricture, as indicated by a characteristic drop in F0 across the consonant. Unaspirated y appears to lack this drop, indicating that it may be produced with front tensing; see Fig. 5. Voiceless consonants are produced with an interruption of the airflow through the glottis, by an unclear process. The F0 component of vowels following voiceless consonants is unperturbed, indicating that laryngeal back tensing is relaxed (or possibly the larynx is slightly front-tensed) during voiceless consonants. Distinctive voiced–voiceless contrasts appear among the obstruents, with voicelessness as the accentual ("marked") component; sonorants are by nature voiced. Voiced and voiceless consonants may appear in initial, medial, or final position.

Aspiration. Aspiration is a turbulent flow of air through the glottis. Aspiration may stand alone as a consonantal sound (represented by h) or as an accentual component of a consonant, forming an aspirated consonant (represented by h following the symbol for the consonant). The turbulence peaks after the release of the oral stricture, more so for continuants than for stops (see the waveforms in Figures 4, 6, and 8). Aspirated consonants contrast with plain (unaspirated) consonants (Figures 5, 7, and 9), which are released directly into a following sound without intervening turbulence. Aspirated consonants do not occur in final position.

The location of aspirational turbulence in the glottis is determined by the voicing component of the consonant: in voiced consonants the location is posterior glottal; in voiceless consonants the location is (apparently) full-glottal.

In aspirated voiced consonants the posterior glottal stricture is produced in concert with the normal laryngeal back tensing of phonation. The result is a posterior whispery voiced* sound. Whispery voice occurs alone as non-initial h, as a component of an aspirated voiced stop, e.g.,

ghôʹm 'wheat' (cf. gôʹ 'where'), J̣haṇJ̣ʹîr 'chain',
ǰhʹip 'tongue' (cf. ǰêʹb 'pocket'), dhîʹ 'daughter' (cf. dʹî 'from'),
ḍhôʹr 'hopper' (cf. ḍʹôk 'back'), bhʹîro 'male goat' (cf. bʹîḍo 'much'),
or as a component of an aspirated voiced continuant, e.g.,
yhʹûm 'I will come' (cf. yʹû 'barley'), lhôʹṇ 'salt' (cf. lôʹ 'cedar'),
rhôʹ 'song' (cf. rʹôt 'night'), zhâʹy 'place' (cf. zʹaṛ 'poison'),
mhârʹânu 'killing' (cf. mârʹânu 'dying'), whʹî 'he will come down' (cf. wʹî 'water').
No aspiration has been recorded with the continuants n and , nor with or the rare, borrowed sounds and ǧ.

After a voiceless consonant the accentual turbulence appears as full-glottal voiceless breathy phonation.*  The heightened air flow of the full-glottal opening is barely contained by the labial closure of the aspirated stop ph, which releases with a spirantal off-glide or merges with bilabial f. Examples of aspirated voiceless consonants include:

khâʹl 'threshing floor' (cf. kâʹl 'year') c̣hʹôṇ 'hollyoak' (cf. c̣o˜ṭ- 'write')
čhʹay 'shade' (cf. čʹay 'tea') ṭhʹongi 'ax', (cf. ṭʹombo 'trunk')
thʹî 'thy' (cf. tʹî 'he [ergative]') phalʹûṛo 'grain' (cf. palʹûlo 'male Aćharêtâ' speaker')
baćhâʹr 'young ox' (cf. kućʹa 'trap door')

The turbulence of an initial h starts out voiceless and becomes whispery voiced by its end. The delayed appearance of whispery voice indicates a latency in that process.

Within a simple, noncompounded word only one point of aspiration occurs, almost always at the beginning of the word. Non-initial aspiration may occur alone (lohʹoylo 'red', rohʹiṇ 'urial') and with affricates (ac̣hʹî 'eye', gućhʹî 'morel'), and it occurs in small group of forms that show etymological pecularities; e.g., forms with ancient prefixes (âghâʹ 'sky', oḍhʹôl 'flood'), loanwords (sithâʹr 'sitar'), and a few other words (buṭhʹe 'all', baḍhʹîr 'sledgehammer', gôkhʹî 'Chitrali', etc.).

Transition.  The system of transcription used here indicates word boundaries for clarity. Sounds within a word are produced in close transition, which may imply the overlapping (coarticulation) of certain articulatory components of adjacent sounds. Across a word boundary (indicated by a space) sounds are produced with lexical transition, in which fewer types of components overlap. However, the sounds across some word boundaries may be produced in close transition, in which case the words are linked together by an underline (_) rather than separated by a space. The normal overlap of lexical transition may be broken, as indicated by a comma (,).

The types of transition are summarized in Table 2.


Transition
Lexical Close Broken
[space] _ ,

Table 2. Transition Types in Aćharêtâʹ.





Notes

* I follow Catford's (1977: 99ff.) nomenclature of phonation types.


1. Stage 1 must precede Stage 2, otherwise the ʹê from Stage 2 would be closed during Stage 1, giving *nʹîṛi 'creek'. Likewise, Stage 1 must precede Stage 3, otherwise the ʹô from Stage 3 would be closed during Stage 1, giving *dʹûdo 'father's father'. Stage 2 probably preceded Stage 3, otherwise both â and ʹô (from Stage 3) would have to be fronted during Stage 2. Stage 4 must follow the other stages, otherwise accented short vowels would have been lengthened and subjected to the processes of Stages 1–3. It appears that if the position of the accent changed after determining this cascade of processes, the cascade occurred again, so that *uḍhâwʹi 'having fled' > uḍhêwʹi 'having fled' > *uḍhʹêwum 'I'll flee' > uḍhʹîum 'I'll flee'.





Figures

The following graphs were produced using SIL International's Speech Analyzer 1.5 for Windows, which is available as a freeware download from the SIL website.


Aćharêtâʹ F0 Plot Showing Downdrift
Figure 1.
F0 plot of the first first few seconds of the audio sample of the Aćharêʹt History text, showing characteristic downdrift over the segments of the following utterence:
bʹe, čilâsʹî, gʹupusî pânda de, lʹaspura the yhʹôla. lʹaspura čʹûr brhâwʹu de;


Aćharêtâʹ F0 Plot of Initially Accented Long Vowels
Figure 2.
F0 plot of Initially Accented long vowels, showing pitch drop over the vowels. Sampled words are:
thʹî 'thy', tʹî 'he [erg.]', pʹûtr 'son', dʹôda 'grandfathers', bhanǰayʹânu 'hitting'.


Aćharêtâʹ F0 Plot of Terminally Accented Long Vowels
Figure 3.
F0 plot of Terminally Accented long vowels, showing level, falling-rising, or rising pitch over the vowels. Sampled words are:
kâʹl 'year', dhrôʹk 'side valley', asâʹm 'us', zhâʹy 'place', hukumâʹt 'government'.


Aćharêtâʹ Aspirated Voiced Continuant Consonants
Figure 4.
Waveform and F0 plot of aspirated voiced continuant consonants. Sampled words are:
pahurta 'after', whaʹi 'having come down', yhʹôla 'came [plural]', zhâʹy 'place'.


Aćharêtâʹ Plain Voiced Continuant Consonants
Figure 5.
Waveform and F0 plot of plain voiced continuant consonants. Sampled words are:
wʹaxt 'time', yʹêy 'mother', zabâʹn 'language'.


Aćharêtâʹ Aspirated Voiced Stop Consonants
Figure 6.
Waveform and F0 plot of aspirated voiced stop consonants. Sampled words are:
bheṇ[pʹûtr] 'sister's son', dharʹîto 'remained', uḍhêwʹi 'having fled', ghinʹi 'having gotten'.


Aćharêtâʹ Plain Voiced Stop Consonants
Figure 7.
Waveform and F0 plot of plain voiced stop consonants. Sampled words are:
be 'we', dʹôda 'grandfathers', bʹîḍi 'much', ǰangʹî 'of war', gʹiḍ 'Dameḷ [place name]'.


Aćharêtâʹ Aspirated Voiceless Stop Consonants
Figure 8.
Waveform and F0 plot of aspirated voiceless stop consonants. Sampled words are:
phʹatah 'victory', thʹî 'thy', c̣hatrʹôla 'in Chitral', khoǰʹôlo 'asked'.


Aćharêtâʹ Plain Voiceless Stop Consonants
Figure 9.
Waveform and F0 plot of plain voiceless stop consonants. Sampled words are:
pʹûtra 'sons', tʹâ 'there', čʹûr 'four', mac̣ôʹk 'Mac̣ôʹk', kâʹl 'year'.

[Under Construction]


(Phonemic transcription updated 3 June 2011)