Manúskta

Consonants
Consonants written as digraphs and trigraphs are pronounced as a single sound, but are not phonemes in their own right. Underlyingly they are sequences of two phonemes that can be split up by infixes.

The voicing contrast in fricatives (central and lateral) is gradually being lost in the language. The only place where it is consistently maintained is word initially. Furthermore fricatives cannot be found in syllable codas. If one is written, it is not pronounced, but puts breathy voice on the preceding vowel.

Voiceless lateral fricatives can also be pronounced as voiceless lateral approximants at the same point of articulation. All lateral fricatives can also optionally be pronounced as clusters of fricative+lateral.

Consonants in square brackets [] are only found in formal speech. Consonants in round brackets are only found in some dialects.

It is debatably whether the glottal stop is a phoneme in its own right. It only occurs between vowels that would be next to each other because of affixing or compounding. It never occurs morpheme internally. However it is written in the orthography with a hyphen.

The uvular approximant written as g is in free variation with a trill, and a voiced stop. It can also be velar or pharyngeal for some speakers.

Ejectives, implosives, glottalised sonorants, consonants with faucalised and breathy voice as well as voiceless aspirated stops and affricates can all be heard in Manúskta. However these can be predicted from their position relative to accented syllables and breathy voiced vowels. Therefore they are not phonemes and are not listed.

Vowels
All diphthongs listed are centering diphthongs that start at the position labelled on the diagram and center to schwa. Ee and Oo are normally monophthongs, except in syllables with long accent.

Vowels with breathy voice are pronounced with a markedly lower pitch than vowel with modal voice. Also Manúskta has a phonation spreading process that works in a similar way to tone spreading. Breathy voice can spreading from the vowel in one syllable to the vowel in the preceding syllable, as long as none of the intervening consonants are "blocking". Blocking consonants are those that were historically voiceless. They are written in the orthography as p, t, k, f, th, s, h and - (the glottal stop). This is why the word for 1 is aw, IPA: [aw̰], X_Sampa: [aw_k], the word for 100 is awhaanztî, IPA: [awxa̤ə̤tʼɨː], X-Sampa: [awxa@_tt_>1:] (even though the following vowel is breathy voiced, the h blocks the phonation spreading) but the word for 110 is aawmwuastî, IPA: [a̤ə̤wŋʷy̤a̤tʼɨː], X-Sampa: [a@_twN_wya_tt_>1:]. There is nothing to stop the breathy voice on the vowel "ua" spreading to the previous vowel, since neither "m" nor "w" are blocking consonants.

Initial Consonant Mutation
Word initial clusters of fricative + voiceless stop are normally ejectives at the point of articulation of the stop. The only exception to this is if there is another ejective later in the word. In this case the cluster is pronounced as a tenuis stop. Under normal conditions, word initial voiceless stops are aspirated, however, in a similar fashion to the ejectives, they become tenuis if there is another aspirated stop later in the word. A similar thing happens with voiced stops at the beginning of words, which become implosives unless there is another implosive later in the word, in which case they become tenuis stops. Likewise, word initial voiceless fricatives (central and lateral) become voiced if there is another fricative later in the word that is pronounced as voiceless (if it is written, but elided or pronounced as a voiced fricative, then this process doesn't apply). A few speakers glottalise all word initial sonorants, except if there is another glottalised sonorant later in the word.

Suprasegmentals
Each word in Manúskta contains exactly one accented syllable. There are two types of accent, Short, and Long (which lengthens the vowel it is on). Word final syllables with no coda consonant (note that -y and -w count as coda consonants) can only have Long accent.

If a vowel with short accent is immediately followed by a stop, then that stop becomes implosive (with the exception of k and g which become glottalised uvular approximants IPA: [ʁ̰], X-Sampa: [R_k], owing to the difficulty of making an implosive in the back of the mouth). e.g. cot - kót, IPA: [kʰɔɗ] X-Sampa: [k_hOd_<]. to stay overnight - pák, IPA: [pʰaʁ̰], X-Sampa: [p_haR_k]. If the stop is written b, d or g, then it itself is pronounced the same way, but the vowel is changed to breathy e.g. bottle - kóed, IPA: [kʰo̤ɛ̤ɗ], X-Sampa: [k_hoE_td_<], vegetable - páag, IPA: [pʰa̤ə̤ʁ̰], X-Sampa: [p_ha@_tR_k]. If the stop is written p, t or k, then the vowel can never be breathy.

Also breathy voice can never spread onto (or through) syllables with short accent e.g. cots - kótuaz, IPA: [kʰɔɗy̤a̤], X-Sampa: [k_hOd_If a vowel with a short accent is immediately followed by s and then a voiceless stop, then the s is not pronounced and the voiceless stop becomes ejective e.g. Manúskta is pronounced IPA: [mʲanɵkʼta], X-Sampa: [m_jan8k_>ta].

If a vowel with a short accent is immediately followed by a sonorant, then that sonorant acquires creaky voice and also blocks the spreading of breathy voice e.g. human - mánuas, IPA: [mʲan̰y̤a̤], X-Sampa: [m_jan_kya_t]. Here we can see that the vowel in the second syllable has breathy voice, which would normally spread through n (as it is a sonorant) were it not for the fact that n has acquired creaky voice.

If the vowel with a short accent is immediately followed by a fricative and then a sonorant, the fricative is not pronounced and the sonoarant acquires creaky voice. However the vowel acquires breathy voice e.g. Muslim - móozlim IPA: [mʲṳl̰ɨm], X-Sampa: [m_ju_tl_k1m]. If such a sonorant is a vowel, then a glottal stop is inserted before it e.g. Pacific - paséefik IPA: [pʰazi̤ʔɨ̰k], X-Sampa: [p_hazi_t?1_tk].

In contrast to vowels with short accent, vowels with long accent often modify the pronunciation of the preceding consonant, in addition to lengthening the vowel e.g. if they are preceded by a sequence of fricative + voiceless stop, then the fricative will be deleted and the stop will become an ejective e.g. twenty seven - twenskông, IPA: [ʈʰe̤nkʼoːŋ], X-Sampa: [t`_he_tnk_>o:N]. Otherwise a voiceless stop before a vowel with long accent will become aspirated e.g. eggplant - makûax, IPA: [m ʲ akʰy̤ːa̤], X-Sampa: [m_jak_hya_t:]. As mentioned previously, fricatives are normally voiced between vowels, but a long accent on the following vowel blocks this for f, s and h e.g. to collect tax - pasêeh, IPA: [pʰasi̤e̤], X-Sampa: [p_hasie_t]

The pronunciation of the long accent in other cases depends on whether or not the vowel is breathy voiced. Voiced fricatives, voiced stops, nasals and approximants (i.e. all the non blocking consonants) all become breathy voiced before a vowel with long accent and breathy voice. The vowel in this case acquires a "peaking" tone that starts low, rises to mid and then falls to low again e.g. zoo - zôoh, IPA: [z̤ṳ̂ː], X-Sampa: [z_tû_t:]. Furthermore the breathy voice can spread to preceding syllables in the same word e.g. to visit - tám, IPA: [tʰam̰], X-Sampa: [t_ham_k]. This word by itself clearly has no breathy voice, but when we look at the verb meaning to visit a zoo - taamzôoh, IPA: [tʰa̤ə̤m̤z̤ṳ̂ː], X-Sampa: [t_ha@_tmz_tû_t:] we can see that the breathy voice has spread to the first syllable.

If a syllable with long accent begins with a voiced fricative, a voiced stop, a nasal or an approximant, but there is no breathy voice on the vowel, then the vowel acquires a sharply rising tone and the consonant is pronounced with faucalised voice e.g. twelve - lô, IPA: [l̬ǒː], X-Sampa: [l_vǒː].

Note that aspirated stops, ejectives, implosives and voiceless fricatives are each considered special categories of consonant in Manúskta. A word can have a maximum of one phoneme from each category. If it would have more than one, then all except the last are changed to tenuis stops (or voiced fricatives in the case of f, s and h) e.g. to visit - tám, IPA: [tʰam̰ ʲ ], X-Sampa: [t_ham_k_j], to make a surprise attack- kyâp, IPA: [cʰa:p], X-Sampa: [c_ha:p], to travel a distance to make a surprise hit and run raid - tamkyâp , IPA: [tam ʲ cʰa:p], X-Sampa: [tam_jc_ha:p]. Here the first t would normally be aspirated, but because there is an aspirated palatal stop later in the word, it is deaspirated to a tenuis stop.

Manúskta has an extremely productive process of producing nouns from verbs and vice versa by changing the location of the pitch accent. This process originated with English stress derived nouns (e.g. to inVITE vs. an INvite), but was borrowed into Manúskta and became much more productive e.g. to invite- inváoyt, IPA: [ɨnvʲɔj̤], X-Sampa: [1nv_jOj_k] vs. an invite (e.g. an invitation) - ínvaoyt IPA: [ɨn̰vʲɔj], X-Sampa: [1n_kv_jOj], a convict - káonvikt, IPA: [kʰɔn̰vʲɨk], X-Sampa: [k_hOn_kv_j1k] vs to convict - kaonvíkt IPA: [kʰɔnvʲɨʁ̰], X-Sampa: [k_hOn_kv_j1R_k].

It is very common for the type of accent to shift from short to long or long to short when it is moved from one syllable to another. The reason for this is that, once upon a time, every syllable in a word in Manúskta could be long or short. However a sound change occurred that shortened all the unaccented vowels. An example of this is the following pair of words: contrast (noun) - kóntraast, IPA: [kʰɔn̰ʈa̤ə̤t], X-Sampa: [k_hOn_kt`a@_tt] vs. to contrast - kaontrâast, IPA: [kɔnʈʰa̤ə̤ːt], X-Sampa: [kOnt_ha@_t:t] Note that this is another example of the rule that Manúskta does not like words with more than one aspirated consonant. In the verbal form, the initial k would normally be aspirated, but is tenuis because there is another aspirated stop later in the word.

Phonotactics
Manúskta does not like syllables that are more complex than CVC, where V is a vowel and C is a consonant sound. For the purposes of phonotactics, clusters like -tr- behave as single sounds even though, underlyingly, they are clusters of two consonants. It is also worth noting here that, in most dialects, "r" and "l" are deleted in syllable codas.

Manúskta orthography contains many instances of consonant clusters that would be forbidden by the phonotactics specified above (sometimes formed by affixation and compounding). These are simplified to permissible clusters by applying a series of rules. These are applied in layers, and keep going until only permissible syllable shapes are obtained (maximum CVC):

1) Delete all fricatives unless they come immediately before a vowel (this may lead to breathy voice being put on the preceding vowel, as long as there is no intervening voiceless consonant to block it). 2) Excluding digraphs listed in the consonant chart above, begin deleting all approximants other than approximants immediately after a vowel. This rule starts working from right to left. <p style="border-style: initial; border-color: initial; font-style: normal; border-style: initial; border-color: initial;">3) Begin deleting all stops in the cluster. Start working from right to left. <p style="border-style: initial; border-color: initial; font-style: normal; border-style: initial; border-color: initial;">4) Begin deleting all nasals in the cluster. Start working from right to left.

<p style="border-style: initial; border-color: initial; font-style: normal; border-style: initial; border-color: initial;">Take for instance the transitive verb skrúym - to scream at someone or something. Both the initial and final consonant clusters as written are not allowed in Manúskta. They are simplified as follows:

<p style="border-style: initial; border-color: initial; font-style: normal; border-style: initial; border-color: initial;">a) The initial s does not come before a vowel. It is therefore deleted (though it makes the k ejective). We now have [kʼɻɵj̰m ʲ ], which is not yet permissible, so we go to the next step. <p style="border-style: initial; border-color: initial; font-style: normal; border-style: initial; border-color: initial;">b) The [ɻ] is not immediately after a vowel, so it gets deleted. However the [j] is, so it is retained. We now have [kʼɵj̰m ʲ ], in which the syllable onset is now phonotactically permissible, but the coda is not. <p style="border-style: initial; border-color: initial; font-style: normal; border-style: initial; border-color: initial;">c) The final [m ʲ ] is deleted. We now have [kʼɵj̰] which is permissible.

<p style="border-style: initial; border-color: initial; font-style: normal; border-style: initial; border-color: initial;">Now let's say we want to make the verb antipassive. To do this, the infix "it" is inserted after the first underlying consonant in the onset of the final syllable, this gives sitkrúym which simplifies as follows:

<p style="border-style: initial; border-color: initial; font-style: normal; border-style: initial; border-color: initial;">a) The initial s comes before a vowel this time. It is therefore not deleted. To start with we have [sɨtkɻɵj̰m ʲ ], where both the medial and final clusters are not yet permissible, so we go to the next step. <p style="border-style: initial; border-color: initial; font-style: normal; border-style: initial; border-color: initial;">b) The [ɻ] is not immediately after a vowel, so it gets deleted. However the [j] is, so it is retained. We now have [sɨtkɵj̰m ʲ ], with a coda that is still not phonotactically permissible. <p style="border-style: initial; border-color: initial; font-style: normal; border-style: initial; border-color: initial;">c) The final [m] is deleted. We now have [sɨtkɵj̰] which is permissible.

<p style="border-style: initial; border-color: initial; font-style: normal; border-style: initial; border-color: initial;">Now say we prefix it with the deontic mood prefix waona-, to make woenaeskrúym (remember that the s is deleted and puts breathy voice on the preceding vowel, with spreads through "n" to the first vowel in the prefix). Also not that the k is no longer word initial, so it is not ejective.

<p style="border-style: initial; border-color: initial; font-style: normal; border-style: initial; border-color: initial;">a) The s is deleted, so we initially have [wo̤ɛ̤na̤ə̤kɻɵj̰m ʲ ]. <p style="border-style: initial; border-color: initial; font-style: normal; border-style: initial; border-color: initial;">b) The [ɻ] is not deleted, because the consonant cluster that it is in conforms to Manúskta phonotactics. Nothing happens to the final [j] for similar reasons as before. <p style="border-style: initial; border-color: initial; font-style: normal; border-style: initial; border-color: initial;">c) The final [m ʲ ] is deleted.

<p style="border-style: initial; border-color: initial; font-style: normal; border-style: initial; border-color: initial;">Now let's say that we want to put both the antipassive infix and the deontic prefix on the verb. First waona- is prefixed, since the mood inflexion takes place before voice inflexion. So first we have woenaeskrúym like before. However the antipassive infix -it- attaches to the first underlying consonant in the onset of the initial syllable. For the purposes of infixing, Manúskta grammar would divide a word like this up as woenaes.krúym. Therefore the written form is woenaeskitrúym.

<p style="border-style: initial; border-color: initial; font-style: normal; border-style: initial; border-color: initial;">a) tr coalesces into [ʈ], and the s is deleted with the same breathy voice effect as before. So now we have [wo̤ɛ̤na̤ə̤kɨʈɵj̰m ʲ ]. <p style="border-style: initial; border-color: initial; font-style: normal; border-style: initial; border-color: initial;">b) The [m] is deleted for the same reasons as in the other words. So the word is pronounced [wo̤ɛ̤na̤ə̤kɨʈɵj̰m ʲ ].

<p style="border-style: initial; border-color: initial; font-style: normal; border-style: initial; border-color: initial;">So when is the m actually pronounced? Well Manúskta has an extremely productive process that suffixes -ar (pronounced [a] except in the few dialects that allow [ɻ] in codas) to create a noun from a verb meaning "the one who does ...". So woenaeskitrúymar means "the one who I hope screams", and is pronounced [wo̤ɛ̤na̤ə̤kɨʈɵj̰m ʲ a].