From PIE to Primal, Sounds

[Note: I have combined multiple past posts under a single section: From PIE to Primal] (AMB, 3/22/18)

Over a series of posts we’ll discuss exactly what we, the creators of Wenja, do each time we create – or rather derive – a new word for the primary language of Oros. How do PIE words like *bʰere- ‘to carry’ and *h₂odyo ‘today’ become Wenja bara and shaja, respectively?

Our first post will begin with the largest class of consonants in PIE — stops (aka occlusives). Stops are found in all of the world’s languages.  When you make a stop, the air is stopped in the mouth in the initial production of the sound, and then the air is released. For this reason you can’t hold a stop out like an s or an m; try holding out a p — you’ll find that it’s impossible!

There are six stops in English:

English stops are organized according to place of articulation and voicing. Stops can either be voiced or voiceless. Voiced sounds require vibration of the vocal folds (in your throat); voiceless sounds do not. The terms bilabial, alveolar, and velar refer to the place of articulation, or where in the mouth a sound is articulated. You can see in the diagram to the left that bilabial sounds are made with the lips, alveolars just behind the top front teeth, and velars towards the back of the mouth.

(Click here for an interactive overview of English phonetics)

While there are only six basic stops in English, it’s very likely that PIE had fifteen. That’s a lot compared to most languages of the world!  The majority of Indo-Europeanists assume five places of articulation (bilabial, dental, palatal, velar, and labiovelar (velar consonant with lip rounding)) and three different ways to make stops (voiceless, voiced, and voiced aspirated). The voiced aspirated stops, which are perhaps better described as breathy, are characterized by a voiced puff of air following the stop in question.

Now let’s see how all of this plays out in Wenja. In the lists below, we first give you an example of a PIE word beginning with the reconstructed stops in question (marked in bold), followed by descendant words in actual Indo-European languages. We then conclude with the Wenja derivative (marked in red).

• Labials:
  • PIE *ped/pod- ‘foot’ > Luvian pāta-, Sanskrit pad-, Greek pod-, Latin ped-, English foot, Armenian otn, Wenja padi ‘foot’
  • PIE *bel- ‘strong’ > Sanskrit bála- ‘strength’, Greek beltíōn ‘better’, Latin dē-bilis ‘lacking strength’, Old Church Slavonic bolĭjĭ ‘bigger’, Wenja bala ‘strong’
  • PIE *bʰer- ‘carry’ > Sanskrit b^hárāmi ‘I carry’, Greek p^hérō, Latin ferō, Armenian berem, English bear, Old Church Slavonic berǫ ‘I take’, Old Irish ·beir, Wenja bara ‘carry’
• Dentals:
  • PIE *ters- ‘be dry, thirst’ > Sanskrit tarṣáyati ‘makes thirsty’, Greek térsetai ‘becomes dry’, Latin terra ‘(dry) land’, English thirst, Albanian ter ‘I dry’, Wenja tarsa ‘become dry’
  • PIE *doru ‘(oak) tree’ > Hittite tāru ‘wood, tree’, Sanskrit dā́ru, Greek dóru, Old Irish daur, Old Church Slavonic drěvo, Albanian dru, English tree, Wenja daru ‘wood’
  • PIE *dʰeh₁- ‘put, do’ > Sanskrit d^hā- ‘put, do’, Greek t^hē- ‘put’, Latin faciō, English do, Old Russian dě, Wenja daha ‘do, make, put’

• Palatals:
  • PIE *ḱerd- ‘heart’ > Sanskrit śrad-, Old Church Slavonic srĭdĭce, Lithuanian širdìs, Hittite kard-, Greek kardíā, Latin cord- [kord], English heart, Wenja charda ‘heart’
  • PIE *ǵónu ‘knee’ > Sanskrit jā́nu, Avestan žnum, Hittite gēnu, Greek gónu, Latin genū, Eng. knee, Wenja janwa ‘knee’
  • PIE *ǵʰeu- ‘pour’ > Sanskrit hūyáte ‘is poured’, Avestan zaotar- ‘priest’, Greek k^he(w)ō ‘I pour’, Tocharian B kewu ‘I will pour’, German giessen, Wenja jawa ‘pour’
• Velars:
  •  PIE *kes- ‘hair’ > Old Church Slavonic kosa ‘hair’, Lithuanian kasà ‘braid’, Hittite kiss-, Greek késkeon, Old English heord ‘hair’, Wenja kasa ‘braid, weave’
  • PIE *gras- ‘grass’ > Sanskrit grásate ‘eats’, Greek grástis ‘grass’, Latin grāmen ‘grass’, Wenja grasti ‘straw’
  • PIE *gʰrebʰ(hᵪ)- ‘grab’ > Sanskrit /g^hrab^h-/, Av. grab-, OCS grabiti, Eng. grab, Wenja grabasha ‘grab, catch’

• Labiovelars:
  • PIE *kʷis ‘who’ > Sanskrit kás ‘who’, Old Church Slavonic kŭ-to ‘who, Lithuanian kàs, Hittite kuis ([k^wis]), Latin quis ([k^wis]), Old English hwæt, Wenja kway ‘who, what’
  • PIE *gʷen- ‘woman’ > Sanskrit jáni-, Old Church Slavonic žena, Old Prussian genna, Hittite kuinnas ([k^winnas]), Greek gunḗ, English queen, Wenja gwani ‘woman’
  • PIE *gʷʰen- ‘kill’ > Sanskrit hánti ‘slays’, Avestan jaiṇti, Old Church Slavonic ženǫ ‘I hunt’, Hittite kuenzi ([k^wentsi]) ‘slays’, Greek -p^honos ‘slayer’, Latin dē-fen-dit, English bane, Wenja gwana ‘kill’

If you’ve made it this far and have looked closely at the PIE & Wenja examples, you’ll note two basic changes in the Wenja stop system.

1. Voiced aspirates are pronounced as normal voiced stops in Wenja: *bʰ, dʰ, ǵʰ, gʰ, gʷʰ > b, d, j, g, gw, respectively
2. Palatal stops are pronounced as (alveopalatal) affricates in Wenja: *ḱ, ǵ, ǵ > ch, j, j, respectively. Recall that ch  = < ch > (cheese) & j = < j > (juice).

Though there are some modifications here and there (we’ll get to those), it’s pretty much that simple! This leaves us with the following stop inventory for Wenja:

Next time we’ll talk about the fricatives – the hissy sounds – of Wenja: s, z, h, sh, and f.

Welcome back!

Having examined the stops of PIE & Wenja, we can now turn to the fricatives. Fricatives are noisy sounds, characterized by significant (but partial) obstruction within the vocal tract. In English, we have nine fricatives (in case you’re wondering, that’s a lot!) — they can be made with the lips & teeth together (labiodental), with the tongue in between the teeth ([inter]dental), just behind the top front teeth (alveolar), retracted slightly behind that (alveopalatal), or in the throat (glottal). Note the voicing distinction found in the stops is also present for all the fricatives except for < h >.

The sounds < f >, < v >, < s >, < z >, and < h> are pretty self-explanatory — they’re the sounds at the beginning of the words fish, van, sit, zoo, and hi, respectively. But what about those remaining, funny-looking characters? The symbol < θ > =  < th >, as in thick, < ð > = < th > as in then, < ʃ > = < sh > as in shoot, and < ʒ > = < j > as in judge.

And how about PIE?  While the proto-language had fifteen different types of stops, it probably only utilized four different types of fricatives.  They were:

We’ll discuss each of these in turn.

The fricative *s was likely a dental sound, more like the Spanish s than the English one. We find this sound all over the place in PIE, for instance in the widespread root ‘to sit’. PIE *sed- ‘sit’ > Ved. sáda ‘sit!’, Lat. sedēre, Eng. sit, OCS sěděti, Gk. hézomai, Arm. hecanim, Wenja sada ‘sit’.

While < z > is a full-fledged sound in English (note the pair sit ~ zit), it was not in PIE. In linguistics we would call *z an allophone of the phoneme *s. By this we mean that PIE speakers didn’t hear *z as a different sound from *s, despite their difference in pronunciation. In fact, the only time we can reconstruct *z is when *s was situated in front of a voiced stop (*d, *gʰ, etc.).  To give you an example, the *e vowel in *sed- ‘sit’ was sometimes deleted to produce *sd-, which was automatically pronounced as *-zd-. This famously explains the source of PIE *ni-zd-ó- ‘nest’, literally the ‘place (for a bird) to sit down’, continued by Sanskrit nīḍás, Latin nīdus, Old Church Slavonic gnězdo, English nest, Wenja nizda ‘nest, lair’. As for the element of *nizdó-, ni ‘down’, if you watch Brenna’s Winja Warshta: Brina Winja dachaya, you’ll hear her give the command U ni sada! “Sit down!” The basic word for ‘down’ in both Wenja & PIE is ni.

Just like PIE, < z > really isn’t used in Wenja, except when it’s before voiced stops in words such as mazga ‘to descend; marrow, semen’ or in borrowed words (Izila < Iz. His-hílax).

The other three fricatives of PIE, *h₁,* h₂, *h₃, are known as the laryngeals. These were all fricatives produced in the back of the throat, which were largely eliminated / altered beyond recognition in the daughter languages.

Here’s the funny thing about the laryngeals. While they were technically consonants in PIE, they are primarily continued as vowels in the Indo-European languages.  So if they’re usually vowels, why do we think they were consonants? This is largely due to the Anatolian languages, such as Hittite. Let’s look at some examples:

1. *h₁es- ‘be’ > Hittite ēszi ‘is’, Sanskrit asti, Greek esti, Latin est, English is, Wenja hasa ‘be (formal), exist’
2. *h₂ent- ‘face’ > Hittite ḫanti ‘in the face of’, Sanskrit ánti ‘before’, Latin ante, Greek antí, English end, Wenja shantiyi ‘near’
3. *h₃er- ‘large bird’ > Hittite ḫāraš ‘eagle’, Greek órnis ‘bird’; Gothic ara, Old Irish irar, Old Church Slavonic orĭlĭ ‘eagle’, Wenja faran ‘eagle’

You’ll note that in the words above both *h₂ and *h₃ are continued as < ḫ > in Hittite, a sound which was either a velar or pharyngeal fricative.  For more on the laryngeals (and PIE phonology) in general, I recommend that you read through a recent paper of mine, posted here.

So why do the laryngeals become vowels in the non-Anatolian languages? Well, often laryngeals were situated in difficult-to-pronounce consonant sequences, such as *ph₂ter- ‘father’. They ‘vocalized‘, which really means that they inserted a short vowel (schwa, the *uh* sound in among) next to it in order to make the sequence pronounceable.

1. *d^hh₁s- ‘sacred, religious’ → *d^həh₁s- > Gk. t^hés-p^hatos ‘decreed by god’, Lat. fānum ‘temple’ (< *fasno-), Skt. dhíṣṇya- ‘devout’, HLuv. tasan-za ‘votive stele’, Wenja dahisna ‘temple’
2. *sth₂-to- ‘standing, made to stand’ → *stəh₂-to- > Gk. statós, Ved. sthitá-, Lat. status, ON staðr ‘obstinate, restive (of horses)’, Wenja tashta ‘stand, pedestal’
4. *dh₃-ti- ‘gift’ → *dəh₃-ti- > Gk. dósis, Ved. díti-, Lat. datiō, Wenja dafti ‘(mutual) exchange’

You can see that Wenja does pretty much the same thing as PIE — it inserts a vowel next to the laryngeal to make the sequence easier to say.  While PIE used schwa, Wenja uses either < i > or < a >, which depends on other factors we can’t get into here.

And like Hittite, those consonant sounds haven’t gone anywhere, though two of them have shifted in pronunciation. You’ve probably picked up on how the three laryngeals change into Wenja:

• *h₁ > h   (no change!)
• *h₂ > sh
• *h₃ > f

While the second & third laryngeals become < sh > and < f > consistently throughout Wenja, you’ll often see the first laryngeal changing to < sh > in certain contexts, namely before a consonant or at the end of a word.

• PIE *wih₁ró- ‘hero’ > Sanskrit vīra- ‘hero’, Latin vir ‘man’, English were(wolf), Wenja wishra ‘hero; the one’
• PIE *d(e)h₁só- ‘god’ > Greek theós, Armenian dik’, Wenja dashka ‘god’

You’ll note that in the second example, *deh₁só- ‘god’, there’s an additional change of < s > to < k >, which is something we’ll discuss in a later post on consonant sequences.

So to wrap things up, here is the fricative inventory of Wenja:

Having taken a wonderful vacation after the end of the semester, we can now return to the blog. It’s been a while since we discussed the derivation of Wenja from PIE and today I’d like to focus on a topic that is near and dear to my heart — consonant clusters. A consonant cluster is a sequence of two or more consonants in a row.  Some languages, such as Hawaiian, do not allow any such sequences.  Think of the Hawaiian words that you know (Honolulu, Maui, Hawai’i, etc.). Each syllable is either a simple vowel (V) or a consonant plus vowel (CV).

In our final post about the creation of the Wenja sound system, we’ll look at perhaps the simplest component : the vowels.

It turns out that the vowels of Proto-Indo-European (PIE) were pretty straightforward, too. Unlike English & French which have close twenty vowels, there were only a handful of them in PIE.

Three of the vowels of PIE, *e, *a, and *o, merge to Wenja a, as you can see in the following words:

• *e : *h₁es ‘is’ > Hitt. ēšzi, Ved. ásti, Gk. estí, Lat. est, OCS jestŭ, Wenja hasa ‘is’
• *a : *sal- ‘salt’ > Ved. sal-ilá- ‘salty’, Gk. hál-, Lat. sal-, Arm. ał, OIr. salann, Wenja sal(i) ‘salt’
• *o : *g^wow– ‘cow’ > Lyc. wawa-, Skt. gáv-, Gk. bó(w)es ‘cows’, Lat. bouēs, Arm. kov, OCS gov-ęždĭ ‘bovine’, Wenja gwaw(i) ‘cow’

Why is this? Well, to a large extent the vowels *e and *o would alternate with each other in the same word, with *e used in some situations and *o in others. This is a process called ablaut. We even find remnants of this within English.

1. sit ~ sat         < PIE *sed- ~ *sod-, Wenja sada
2. sing ~ sang   < PIE *sengwh- ~ *songwh-, Wenja sangwa
3. bind ~ band  < PIE *bhendh- ~ *bhondh-, Wenja banda

The left form in each pair is the present tense of the verb, with the right one being either the past tense or a noun form. All of this is easily explainable by reconstructing the vowel *e as the vowel of the present tense, with *o being used in the past and nouns. (Note: this is a huge oversimplification, but is mostly true.)

A number of Indo-Europeanists believe that while the vowels *e and *o participated in ablaut alternations in late PIE, in proto-PIE (what Wenja was designed to be, in part) this alternation did not yet exist. This may sound like a stretch, but there are lots of examples of vowel alternations being created in the history of language from all over the world.

We don’t have to go far — check out some more recent examples of vowel alternations in English:

1. keep ~ kept < *keept — the “eh” vowel was historically the same vowel as the < ee > one (before the Great Vowel Shift)
2. man ~ men < *maniz — the “eh” vowel in men used to be the same as the < a > one (before Germanic Umlaut occurred)

In PIE there were other vowels as well, such as the high vowels *i & *u :

• *i : *mizd^ho- ‘reward’ > Av. mižda-, Gk. mist^hós, Goth. mizdo, OCS mĭzda, Wenja mizda
• *u : *nu ‘now’ > Sanskrit nu, Latin nunc, English now, Wenja nu ‘now, and’

These vowels remained as such in Wenja, resulting in a three-vowel inventory within Wenja. Such inventories are quite common among the languages of the world, found in languages like Arabic.

⇓

There were also long vowels in PIE, vowels that were twice as long as short ones. In English we don’t really make a distinction between the two, at least not in making contrasts between different words. But you’ll note the difference in vowel length in a pair of words such as bet ~ bed, where the < e > vowel is roughly twice as long in bed. Because the actors were primarily monolingual English speakers, the team decided to get rid of vowel length all together.

• *ē : *h₃rḗĝ-s ‘king’ >  Latin rēx, Old Irish rí, Wenja fraji (not fraaji)
• *ā : *nās- ‘nose’ > Latin nasus, English nose, Wenja nas (not naas)
• *ō : *swesōr ‘sister’ > Latin sorōr, Sanskrit svasAr, English sister, Wenja swasar (not swaasar)

And finally, there were the diphthongs in PIE. Diphthongs are complex vowels that consist of two vowel-like elements squished together. In English we have sounds like < i > (as in “I” and “sky”), < o > (as in “no” and “toe”), and < ow > (as in “cow” and “bout”) which begin with a vowel that transitions to another vowel. In PIE there were six diphthongs altogether (excluding long diphthongs). And since *e, *a, and *o all became a in Wenja, these six diphthongs collapsed together to two : ay & aw.

Diphthongs: *e, *a, *o + *u & *i

• *ei : *ḱei- ‘lie down’ > Skt. śaye ‘lies’, Gk. keĩmai ‘I lie’, Wenja chaya ‘lie down’
• *ai : *kaikos ‘blind’ > Lat. caecus, Goth. haihs ‘one-eyed’, Wenja kayka ‘one-eyed’
• *oi : *mei- ‘exchange’ > OLat. moenus ‘duty, tribute, payment’, Wenja maya ‘trade, exchange, replace’
• *eu : *sreumn̥   ‘river’ > Gk. rheũma, English stream, Wenja shrawman ‘river’
• *au : *h₂sauso- ‘dry’ > Gk. haũos, Lith. saũsas, English sear, Wenja shisawsa ‘dry’
• *ou : *mouro– ‘idiot’ > Gk. mōró-, Sanskrit mūrá-, Wenja mawra ‘stupid, foolish; idiot’

Lastly, let’s talk about the rhythm of the language. Many people have noticed Wenja’s sharp, staccato rhythm, which directly contrasts with the sing-songy nature of Izila. We believe that PIE was exactly like Izila; as a pitch-accented language, its speakers differentiated stressed syllables through pitch, not loudness. The stress moves back and forth in both PIE & Izila, as any syllable could be stressed.

Not so in Wenja. Using linguistics terminology, Wenja’s rhythm is trochaic, with its trochees assigned at the left side of the word. This means that in a word with two syllables, the stress will fall on the first syllable : bada ‘dig’ = [BA da].  In a word of four syllables, the stress will fall on the first and third syllables, with the stress being strongest on the first syllable: bada-bada ‘keep on digging’ = [BA da BA da].

This results in a language with a very “caveman” rhythm. But this wasn’t created out of whole cloth. In fact, there’s a theory by Paul Kiparsky whereby the “default” accent is assigned to the leftmost syllable of the word, all things being equal. It’s called the “Basic Accentation Principle” (the BAP for short), and is becoming increasingly more popular within the field. I personally am a fan of it. It’s very heady stuff, but if you’re interested in reading about it more, check out this paper of Kiparsky’s.

Well, that’s it for sounds.  Next up: verbs!