Did Proto-Indo-European Really Only Have 2 Vowels?
ฝัง
- เผยแพร่เมื่อ 14 ส.ค. 2023
- In this video, I explore the reasons why a lot of modern reconstructions of Proto-Indo-European only have two vowels: *e and *o.
________
This channel's Patreon (thank you very much to anybody who donates): / simonroper
Simon didn’t come in from the garden due to adverse weather conditions. I want my money back
he found a glitch in the british isles, we’re meant to have way too much rain for such a thing not to happen
I have to watch these videos alone. My family looks at me when I sit here repeating goat, goal, goat, goal.
You and me both!
My parents both studied linguistics, so the whole family's used to it.
SAME. I'll watch one of these in school or be reading about language and I'll notice I'm saying stuff and I'll get really self-consious.
Always exciting when Simon goes back before Proto-Germanic.
Whoop whoop! 🎉
Gotta love vids on Proto-Proto-Germanic
@@DTux5249 Proto-Proto-Proto English is my favorite reconstructed language for sure.
i see your comments everywhere!
@@skinkrootI watch a lot of these linguistics guys, Simon, Meta, Luke, Dr. Crawford, etc.
It might be similar to some languages in Caucasus, for example, Kabardian, which are analyzed to have only two vowel phonemes, /a/ /ə/, but have various allophones, such as [u]
It’s not *so* rare for languages to have “vertical” vowel systems like that, which contrast vowels primarily by height. For example, in Mandarin, there’s an old-fashioned system of phonemic transcription called Bopomofo, which nicely shows a two-vowel analysis-glides like /j/ and /w/ can be syllabic or influence how adjacent vowels are pronounced, but the nucleus of a syllable is otherwise just /a/ or /ə/.
Whereas, offhand I don’t know of a natural language where the vowel system is primarily “horizontal” like the reconstructed PIE /e, o/ would be, if we took it literally. A vowel system like that could show up in the middle of a vowel shift, but there’s a lot of selective pressure toward something more stable.
@@EvincarOfAutumn I've read speculation online that early IE had only one phonemic vowel /ə/, which subsequently developed an unaccented allophone which later phonemicised /ɐ/, under the assumption of Caucasian influence. When syllabic consonants emerged /j, w/ developed syllabic allophones [i, u] and laryngeals neutralised in syllabic position to [ə]. The new laryngeal allophone forced the old /ə/ to front, and pressure from /i/ caused it to gain a fairly low quality, likely /ɛ ~ æ/ shuffled about. /ɐ/ backed to /ɑ/ to remain balanced. Laryngeal colouring by *h₂ produced a new [ä] phone, causing /ɑ/ to raise to /ɔ/. The laryngeal-coloured allophone further backed to [ɑ]. By Late PIE, *a was probably [ɑ] and *o was probably [ɔ].
I am pretty sure Proto European did not have only 2 vowels - it had all 5 main vowels A / E / I / O / U and then the other ones like Norse and other Germanic languages also got the fancy ones Æ / Ö / Ü etc! I’ve even seen a list of reconstructed Proto European words before and it had words that had all 5 main vowels! There are even some words that haven’t even been changed a lot and some that may still be the same, which is why some words have exactly the same spelling in most European languages, so one could assume those words looked the same in Proto European or were only modified a bit!
@@evefreyasyrenathegoddessev4016If you watched the video, PIE had 2 vowel PHONEMES, which appear as the full 5 vowel ALLOPHONES. Meaning there are only 2 distinct differences in the distribution of vowels. Based on the phoneme, you can tell what the allophone is by the surrounding sounds. It's like how st palatalizes before r in English (E.g. street) and p voices between s and l (e.g. splendid)
commenting to save this
You're an awful lot more well informed than most linguists I know, and everyone makes mistakes, you're brave in putting yourself forward for all our benefit!
I'm sure you know a ton of them too... I'm an interpreter and have been for 25 years and I know ONE professional linguist.
@@Jaggerbushmaybe you’d meet more if you weren’t so rude
@@user-ti8sc6up7t being obnoxiously polite is how you meet 'linguists' that aren't actually linguists.
Most of my university teachers were fraudsters. At least if you study humanities, it's quite possible to become a doctor or professor even if you're just a super mediocre person with a brown tongue.
21:00 as a Punjabi speaker I will say a LOT of very commonly used words such as ਯਾਰ/یار [jäːɾ] meaning friend are loan words, and [j] is actually a phoneme that only shows up in loan words from Persian and Sanskrit (and probably English now), so this sounds believable to me (also Punjabi having Sanskrit loans may sound odd but it's pretty much the same as all the Latin loans in modern European languages, including the Romance languages who are also the descendant of Latin. In regular sound changes from Sanskrit to Punjabi [j] became [d͡ʒ] which is pretty regular).
Edit: coming back to this a couple months later, just want to clarify the /j/ to /d͡ʒ/ sound change from Sanskrit to Punjabi is more complicated than that, I think I can be fairly to confident in saying that in initial positions it seems to have become /d͡ʒ/ but in clusters it seems to behave differently. Though my original point still seems to remain that I can't find any native Punjabi words that have [j] (except maybe as a phonological process of vowels in hiatus but I'm not a phonologist), with all examples I could find of [j] being loan words.
Also I've been talking to an Urdu speaker recently and he was saying that the voiceless velar fricative [x] (from my understanding the sound that makes in German like in ) is regularly pronounced by Urdu speakers but also only found in loan words (from Persian and Arabic mostly). Punjabi has many of these same loan words, but in Punjabi (at least spoken by non Muslim Punjabis) /x/ is instead realized as [kʰ].
Also just an interesting socio linguistics thing going on here, where the ability to pronounce [x] becomes a marker of ones religion, where it would seem like, Muslim speakers of Punjabi and Urdu want to differentiate themselves from non Muslims by using [x] to show a relation to these other Muslim peoples that do have this phone. I haven't studied this exact phenomenon so I can't be sure if that's what's happening here, but I've seen similar things happen in other cases. One example I remember from a lecture is that on the island of Martha's vinyard in Massachusetts, people who don't feel attached to the island will not have a specific feature in their speech that's found on the island, but not Boston where they'd rather be living. So by not using this feature (it's called Canadian raising and is the "aboot" that Americans hear, though it's not usually that extreme and is also found in a lot of the north east United States) these people are showing loyalty not to the island but to Boston.
Also, sounds existing due to loanwords is not that rare - vast majority of [d͡ʒ] in English, both in "soft G" and J, are of Norman French origin, or Latin being viewed through French lens. The presence of [d͡ʒ] in Old English is a contentious topic.
What does ਯਾਰ descend from btw?
Is d͡ʒ even commonly used now?
What about the word Punjabi, it does contain a j. Is it a loanword? Is it just the English word and the native word is completely different? If so, how do you call your language?
@@UniDocs_Mahapushpa_Cyavana Sorry I don't think I made this clear, [j] is the loan sound, [d͡ʒ] is pretty common in modern punjabi and is the result of Sanskrit [j] (in certain positions) as well as Sanskrit [d͡ʑ] and maybe palatalization of Sanskrit [g] though I'm not sure about that one. So a lot of sounds in Sanskrit merged into modern Punjabi [d͡ʒ].
The example of "simple" reminded me of listening to a Brazilian friend pronounce the names Caio and Kyle almost identically - the trailing vowel/consonant thing really stood out.
That’s funny, I know people from Bristol who would pronounce them both the same too
Yeah, in Brazilian Portuguese /l/ becomes a [w] glide at the end of a syllable. It’s a neat little feature.
@@modalmixture Yes, and Polish has ł for that sound.
Another example from Brazilian Portuguese: the same Maicon is pronounced like Michael. The "n" at the end also works like a vowel not completely dissimilar to "l" in English.
@@koyrehme4361I would analyze the "n" there as a nasalization rather than as a vowel or glide, so ['mai.kõ], kinda lika the French do with their coda "n". "Michael" will in fact be most likely pronounced as ['maj.kow] if the person is aware of how the name is "supposed" to be pronounced. If not, the Brazillian Portuguese orthographic intuition would lead the person to say [mi.ʃa.'ɛw]
I can’t believe you didn’t mention Wymysiöeryś! /s
I'm not a linguist myself, but when thinking about PIE, I'd suggest we should always keep in mind that the reconstructed language describes what might've been a set of closely related 'dialects', with local/regional variations, as well as class/social differences, within a linguistic continuum, just as any other modern language. A 'standard', unified variaty like the reconstructed PIE tries to describe may never actually have been spoken, as far as we know.
Yeah like, imagine trying to reconstruct modern English from a set of descendants, people would look at it and think "no that can't be, the vowels are way too asymmetrical... And that R-colored vowel being so prominent just seems too unlikely."
Plus, there were probably things in the language that couldn't be reconstructed. A parallel can be found if you compare proto-Romance (a comparative reconstruction of the Romance languages most common ancestor) and Latin. There are plenty of things in Latin that didn't make it into the Romance languages, such as the passive voice (romance languages form it differently, eg, amatur vs. se ama, both meaning "he is loved") or the Locative and Vocative cases. Similarly, who knows what could have been in Proto-Indo-European. There could have been 10 or 12 cases instead of 8. Maybe even moods past the Subjunctive and Optative. Glottalic theory could have been right. There could also have been roots that were lost that could tie IE with other families such as Uralic. But we'll never know unless time travel or at least time viewing are invented.
Also, keep in mind that the features of the reconstructed PIE might not even have all existed at the same time even if the language didn't vary too much dialectically at any one time. Features attested by only a few related modern languages may only be reconstructable back to late PIE, and may not have developed until that period, while features attested across the family may give insight all the way back to early PIE.
Yeah, like how far back exactly are the laryngeal consonants affecting vowel quality? Could have been very recent. Or it could have been very old, and also happened 1000 years apart from the Centum-Satam split. @@JonBrase
@@akl2k7 isn't PIE to IE languages as PR is to Romance languages?
The Shuswap language of western Canada provides an interesting parallel to the PIE vowel/resonant system, with a series of phonemes that can either act as consonants or vowels. There's [m]~[əm], [n]~[ən], [l]~[əl], [j]~[iː], [ɰ]~[əː], [w]~[uː], [ʕ]~[aː], [ʕʷ]~[ɔː] (plus glottalized versions of all those phonemes).
Funny how i've actually never noticed this even though i've been looking at PIE reconstructed forms for years😅
it _is_ a little obscured by the fact that /*y *w *h/ are often written as *i, *u, and *a when they're syllabic.
An interesting video, Simon. I studied Latin, French and Sanskrit when I was young, and my first wife who passed away in 2015 was from a Latvian family, so you might say I've had experience with a number of Indo-European languages. Back in 2017 I married a Filipina. Her mother tongues are Higaonon Binukid and Cebuano. She speaks Tagalog (Filipino) and a number of other Filipino languages. They are all in the Austronesian language family. I only mention them because prior to the Spanish colonial times in the Philippines, there were only three vowels, "a", "i" and "u". Interestingly, there are also the consonants "y" and "w" which function as vowels in certain cases. In Cebuano, for example, the word for "house" is "balay", where the "y" sounds like "i". The word for "water buffalo" is "kabaw", where the "w" sounds like "u". Loan words from Spanish include the vowels "e" and "o" as required. Sometimes people spell words with "e" in place of "i" and "o" in place of "u", while retaining the pronunciation of the original vowels. The saying goes that Filipinos spell with their ears! Aren't languages wonderful?
You'll be surprised to learn that there exists a fourth vowel phoneme, the schwa, which has largely been merged with the other vowels but mainly with /u/ in Cebuano. It still exists as a separate phoneme in certain dialects of Bohol Cebuano and in other Philippine languages.
I don’t think you’ve ever done the facial expression at 2:28 before on this channel, I’m so used to your usual array of expressions so it caught me so joyfully off-guard
Glad I’m not the only one😂
Every time i watch and listen to your monologues my head fills with thoughts quite incompatible with activity. Prompting drinking tea and having a lie down. Certainly you have a remarkable ability to get the neurons buzzing. Your videos are so very compelling and i commend your scholarship. Thank you Simon. Thank you.
really appreciated your reading there at the end - lovely
Your work is amazing!! Thanks for posting!
I don’t even remember how I first found your channel but I’m so glad I did. I absolutely love your stuff! It’s all so interesting and well done, plus you have a voice that is quite pleasant to the ear :)
19:40, I may be wrong, but I’m most reconstructions of PIE aren’t the laryngeals different even when syllabic? Based of how in Ancient Greek, when syllabic, h1>e, h2>a, h3>o. I might be wrong, but this is what I remember. Amazing video overall btw!!!
God bless you Simon! Love your videos.
One of the many things I love about your videos is that you talk about specific linguistic questions, but you always give insight into the deeper way that we think about and engage with language. I never watch one of your videos without leaving with a deeper appreciation of language generally
you're my hero. nothing but admiration for your knowledge!! I was looking for this information for years!! It was also so shocking to believe, that there was just a reduced set of vowels in the reconstructed version of proto-IE
Very interesting observation about "L" at the end of words. I always used to notice that Michael Howard (former Tory cabinet minister) would say "local" or any other word ending in L with that L emphasised and it sounded really weird. In fact, it wasn't emphasised, he was just pronouncing the final L like the opening L.
I lived in Bristol for 30 years, which is famous for the "Bristol L", apparently unique to Bristol. Words ending in A or O have an L added. "It's a nice areal round 'ere" "Oim droivin moi Ford Fiestal to go shoppin' in ASDAL" and so on. But it's not a hard L - it's as though they're about to pronounce the L but stop, just like Simon's example. In East Bristol, the final L does become a W: "Ford Fiestaw".
The most famous example is the name Bristol itself. In Elizabethan England, the city was called Bristow (originally Brigstowe, place of the bridge), but locals added an L to become "Bristowl". And in East Bristol today, you'll still hear it as "Bristawww".
Interesting. Thank you for sharing. Do you have any insight into why sometimes English commentators add an R at the end of words ending in a vowel, but keep the R silent when it is actually written? Examples below:
Written: Put some water in the radiator of the car.
Spoken: Put some wataa in the radiataa of the caa.
Written: India and Australia are playing tomorrow.
Spoken: IndiaR and AustraliaR aa playing tomorrow.
@@pmaitrasm "R" at the end of a word is called a "rhotic" R. In most of the US and in parts of the UK (especially the West of England), this R is always pronounced. So most US speakers will pronounce "butter" as "butteR", with the R clearly pronounced; most UK speakers will pronounce it as "buttah", unless it's followed by a vowel - so "bread and buttah", but "butteR and jam", where you will hear the R.
The reason people pronounce that R when it's followed by a vowel is it's just easier and makes it flow - "buttah and jam" involves a "glottal stop" between the "buttah" and the "and".
So in your examples, it's easier (but technically incorrect) to say "IndiaR and Australia", so you'll probably hear British BBC commentators saying "India and Australia", but most Brits will say "IndiaR and Australia".
So the "rule" (if one can call it that!) is that if you speak "BBC" English, then you don't pronounce R at the end of a word unless it's followed by a vowel, and you shouldn't put in an unnecessary R with words ending in vowels just because the next word starts with a vowel. But most people people break that last bit of the rule, and do add in an unnecessary R in words ending in a vowel if there's a vowel following.
@@southvillechris, Excellent explanation. Thank you. 👍
One thing I've realized comparing Simon with other TH-camrs is that the quality of the information presented seems to correlate strongly with the humility of the person presenting.
I hadn't found such an interesting channel in years! As somebody with a college degree in English literature and language, I find your channel really useful and a means to learn further about Modern English and its origins, what we didn't study much at uni.
You've got yourself a new subscriber 😊
Eála þā Spania!
Sorry that you had to suffer through college. :(
Thanks for the "simple" example, that was very helpful!
"Sonorant consonants" are not simply consonants that can be said "continuously." That definition better describes the distinctive feature known as [+/− continuant]. Stops, or plosives, are by definition [-continuant], while fricatives, glides (approximants), and non-consonantal vowels are [+continuant].
Sonority is a vaguer concept, and has to do with sounds that have a sort of resonating quality. One way to describe it is a sense that you could "sing" them, in a way you can't sing a [z] or a [v]. Vowels are [+sonorant], as obviously you can sing those, as well as glides, trills and most interestingly, nasals. Nasals are important, because nasal stops are not considered [+continuant]. This is because the [+/− continuant] feature is defined by the airflow in the mouth, which in the case of nasals is blocked. You can, however, say nasal "stops" such as [m] and [n] continuously. Because of this sonorous quality they have, you can "sing" them, or rather hum them. Of course this "sing" explanation isn't perfect though, as there are also unvoiced sonorants, which obviously are not "sung" (no vocal chord vibration), but none the less have this resonating quality to their airflow.
One definition I have heard for sonorance is the ratio of air pressure behind vs. in front of the point of constriction. In a plosive, a lot of air pressure is built up behind the point of occlusion, whereas in an approximant, hardly any is.
I thought it was just volume
Like i have a hunch humans universally scream (actually you could apply this to animals too) something along the lines of ahhh because /a/ is a sonorous (=loud) vowel; the low tongue position and relatively free airway maximises volume for minimal energy (try being loud with a consonant like /t/)
and languages like japanese that have simple phonotactics rely on ending loanwords with 'u' or 'i' because they're quiet (=low sonority) vowels and aren't nearly as jarringly obvious as something like 'a' (eg 'dog' would be borrowed as 'doggu', not *dogga)
@@haharmageddontv6581 You have the idea of +/- sonorant tangled up with +/-voice.
You can't say [t] loud because it is not voiced. It is an unvoiced sound. Your vocal chords can make sounds much louder than anything articulated in the mouth. You can say a [d] much louder than you can say a [t] because of +voice, but both are -sonorant.
Furthermore, I speak Japanese, and know Japanese linguistics very well. The vowels you are talking about, /ɯ/ and /i/, are sonorants. They do not have more or less sonority than any other vowel. Distinctive features are binary; they are not a continuum. The relevant feature that these two phonemes, /ɯ/ and /i/, have in Japanese is that they are often realized as voiceless vowels. This is what you are calling "low sonority," which is not a thing.
You are right that sonorant sounds can be said "louder" than obstruent (non-sonorant) sounds. This is because of the resonating quality I described. Resonance makes sounds louder. But sonority is not simply volume. It is a characteristic of the articulation of the phone, not the sound of the phone.
Regarding "sonorant" - I assume by "sing" you mean "sing resonantly, so that it carries"? Because I assure you, I can sing a [v] :P
@@klop4228 Yes, I am using "sing" in such a sense, like a holding a smooth, clear note. No death metal growls, lol. Again, it's not a perfect description, as you can have voiceless sonorants, but I thought "sing" was a decent way to make the "resonating" quality that defines sonorants more concrete.
a great topic !
edit: the explanation using h2 at the end was particularly good
So, interesting thing here... when you set the wasp and the orange as the background, it became incredibly difficult to comprehend what you were saying. My brain couldn't comprehend both at the same time. But when you momentarily had the camera out of focus and on the "grey" area, it could focus on your words again! Interesting how that happens autonomically.
There's an opposite development in the "Chippewa" indigenous American language. My spell checker isn't familiar with the proper name of that people, Ojibwa, yet it knows the misunderstood transliteration into European languages (first French, then English). The Europeans heard "Chippewa" because the Ojibwa speakers had a habit of dropping the initial vowel. I don't have the symbols, so I'll just use common letters here. The |j| sounds like |ch| to native French and English speakers; the |b| sounds like |p|, and there seems to have been an added vowel sound inserted following the |b|, or preceding the |w| (probably the latter was the rule). There were quite a few other words that were mis-transliterated by the French and English because of the rules that native speakers internalized and didn't think about. I'm not an Ojibwa native speaker, but live in their region on the Great Lakes. I only found out about this linguistic phenomenon because I read a local newspaper column about boats and ships that sailed the Great Lakes in the past. One of the packet and passenger transport vessels had a name that sounded like an indigenous word, but the author couldn't trace it. An Ojibwe woman who lives in this area and reads the column contacted the author to solve the mystery of the name of the vessel and it's meaning in English.
One thing I want to add on, there are similar arguments (primarily from Ringe, 2006) that PIE also had phonetic /i/ and /u/ distinct from the syllabic counterparts /j/ and /w/. He follows a similar process to present some words that can't easily be tied to any equivalent with a syllabic counterpart, according to him anyway.
Interested but undereducated viewer here. A lot of your content goes winging off over my head so quickly I couldn't write a reminder to ask about it later. But when you said IF proto-Indo-European speakers had been made to develop a writing system...I stopped the video to ask if they did not have writing. Then I realized of course they didn't: By the time they wanted to keep records they were already speaking descendent languages of Indo-European. What a fascinating field of study. Thanks for communicating clearly enough that I can get excited and curious about linguistics. Subscribed!
I have a degree in Slavic languages and I have studied historical linguistics since I was a kid. I believe an open phoneme *a existed in PIE. I also believe there was only one laryngeal, or two if you count voiceless and voiced allophones as separate phonemes. They did not have a colouring effect though. A long *a was present in the word for Mother, which in my view came from baby language, extended with a suffix. Labials and open unrounded vowels are the first speech sounds a baby produces. There's more evidence for *a. The laryngeals are also present in the so called aspirated consonants. I believe aspiration could not have existed without a separate h sound. I also reject the Semitic view on root structure, I mean words could have a vowel for the initial phoneme. I don't know all the answers but I do believe every piece of the puzzle can still be found in the daughter languages. The linguist I can relate to the most, is Jouna Pyysalo, although I don't agree with everything he proposes, because, as I said, the daughter languages must show the evidence. Let me conclude with the condition that reconstructions should be pronounceable rather than being mathematical formulas. Maybe I will find the time to corroborate my views. Until then, these remain opinions without scientific backup, I do admit. Then again, many theories start with opinions. Thanks for the video. You know enough about the subject to call yourself a linguist. We need more fresh views as linguists are divided and don't seem to have an open mind on new developments in the field.
"Every expert is closed-minded" Is the go to excuse for crackpots to explain why their ideas don't see any acceptance.
Great video Simon! An example of the phenomenon of semivowels disappearing when appearing next to vowels that you mention would be the word ‘sword’ - after all it is of course said as ‘sord’ rather than phonetically.
Yes. Synchronic evidence for that would be the fact that the "w" in German "Schwert" is still pronounced, even if as a "v". For more substantial proof, Wiktionary tells me the word was pronounced in Old English as /swe͜ord/, [swe͜orˠd], so the "w" in current orthography is an etymological relic/artifact.
It would be very cool if there was an obscure English dialect which pronounced "sword" as something like [swo:d] or [swoɹd]
That's a very good point about how native speakers think of their own language differently to how linguists analyse the mechanics of any given language.
Oh perfect timing for this video, I've been thinking about PIE vowels for the past few weeks.
Can it be that slavic languages have kept some of this? There is a famous czech sentence without a vowel: "Strč prst skrz krk", meaning "Stick a finger through/into the throat"
Czech is a really nice example of a modern European language with syllabic sonorants! I wish I'd mentioned it in the video, now!
I believe the syllabic consonants in Czech come from another source than the PIE syllabic sonorants
The Slovenian word for bumblebee? Čmrlj.
I also thought of the Czech language, could there be a part two? Really interesting thank you.
English too! @@simonroper9218
That was so cool. Never thought of the end of words like “simple”. Had to say it 10 times before I could hear difference. 👏😊
Fascinating as always :)
A couple of other commenters have already mentioned Standard Chinese (Mandarin). This obviously has lots of surface vowel sounds, and there have been several competing phonological analyses to account for these. Edwin G. Pulleyblank even did away with vowel phonemes completely in his system.
An example of a two-vowel analysis is that of Mantaro Hashimoto, who posits just /ə/ and /a/ as the vowel phonemes of Standard Chinese. Each syllable can have /ə/, /a/, or a zero nucleus. Then combined with the glides /j, w, ɥ/, they can generate the various surface vowels. For example, -i [i], -u [u], and -ü [y] are just /j, w, ɥ/ combining with a zero nucleus; -ie [je], -uo [wo], -üe [ɥe] are /j, w, ɥ/ combining with /ə/; -ei [eɪ̯], -ai [aɪ̯] are /ə, a/ combining with coda /j/; and -ou [oʊ̯], -ao [aʊ̯] are /ə, a/ combining with coda /w/. Additional codas produce some other vowel qualities, as in -ian [jɛn] and -üan [ɥɛn] which are /j, ɥ/ combining with /a/ and coda /n/.
If Proto-Indo-European was spoken today, we might see lots of competing theories about how best to analyze its vowel phonology as well. Of course, it is likely that there never was a moment in time where PIE was spoken with the exact phonological system as we have reconstructed it, as its various components might date from different points in the past.
The two-vowel analysis of Mandarin that you mention is what Karlgren called the ‘phonemic craze’. It is just a kind of amusing game that linguists play, trying to see what ‘rules’ they can come up with and how to limit everything to characters available on a keyboard. But in the real world, phonemic analysis has little significance, and is actually misleading. An example of this is how some people have recommended that the word ‘sushi’ in Japanese should be spelt as ‘susi’, because of the complementary distribution of S and Sh in Japanese. But this not only serves no purpose, but also misleads people who don’t know the language, and annoys those who do. Or to spell ‘hachi’ as ‘hati’. I mean, why? It’s just fashionable nonsense.
@@cangjie12 I wouldn't completely dismiss the value of this kind of phonemic analysis, but you do have to be aware of limitations of course. These abstract phonemic analyses can help us understand the distribution of sounds and help us make sense of unexpected gaps and the like. They might offer clues as to earlier stages of the system. But at some point when such analyses get too abstract, they will be less likely to correspond to the intuition of native speakers, regardless of their explicative power.
Applying phonemic analyses to romanization is a separate issue. For Japanese speakers, many of whom really do think of the two consonants in 'sushi' as the same sound, writing it as 'susi' might feel more straightforward. But if you're romanizing for people who are not necessarily familiar with Japanese and just want to approximate the pronunciation, there is absolutely no reason to write it as 'susi' which would just be confusing. Romanization for general purposes has to balance phonemic and phonetic considerations.
@@cangjie12 susi / hati spelling corresponds to the kana system and it's how japanese people type
Hey simon! Great video as always, could you do a video on PIE cases?
wow! this exact topic did always blow my mind a bit :)
It's interesting having been brought up with the paradigm of dividing all sounds into either consonants or vowels that actually on closer inspection there is much more of a spectrum. I can also see the potential for languages to exist where consonants and vowels cannot be defined at all. That they inevitably do probably relates to how our brains function in terms of producing and processing speech.
Your work is brilliant ... you can safely drop the apologetic intro captions ! No need to be overly humble ;)
I love your work...
Liked and subscribed.
11:35 personally, i'd transcribe the south-england "simple" as [sɪmpʰʊw] or [sɪmpʉw]
*the /p/ could go either way for me, and so could the quality of the quality of the final vowel.
*the word to me seems to end in a closing diphthong (as in, ending with a glide). i believe coda /l/ > /w/ is actually a fairly common sound change, and certainly more common that ut spontantiously becoming a vowel.
this is not a critique of the entire video, which i find very informative, i just thought i should point this out. great work!
11:00 I have an accent that's not a million miles from yours Simon. I think I need an extra three vowels in addition to the set that, say, Geoff Lindsey gives as his standard SSB phonemes. I think I have a "GHOUL" vowel /ow/, a "DOLE" vowel /ɔw/ and a "PULL" vowel /ʊw/ all of which definitely feel like closing diphthongs. And it really doesn't feel like just an allophone of /l/: for me the words "dole" and "doll" are a minimal pair in that the vowel sound is exactly the same, but "dole" has an extra "l" sound on the end, /dɔwl/ versus /dɔw/.
The thing that makes me most think they should be analysed this way is comparing to the MOUTH vowel /aw/, another closing diphthong that is uncontroversially accepted as a vowel. Whatever exactly it is that happens with that written "l" at the end of "pull", say, it seems like it's exactly the same thing that happens at the end of a word like "mall", and "mall" rhymes with "cow" and ends in that MOUTH vowel. So if MOUTH is a standalone vowel phoneme, it seems to me so are GHOUL and DOLE and PULL!
As a midlands near-SSB speaker who doesn't feel to employ that much if any l-vocalization, this is really interesting to me.
Would you say 'pow' and 'pal' have become perfect homophones for you?
@@naufalzaid7500 Yep, that's my perception at any rate. I think it would feel forced to pronounce "Pall Mall" in any way other than homophonously with "pow Mao".
This reminds you of how no one's 100% sure about the vowels in YHWH in ancient Hebrew
I like the suggestion it was "Yahu-Wahu" from Larry Gonick's The Cartoon History of the Universe
Yes, but that is a separate issue. We Semites use abjads, and do not write vowels at all. The Jewish people had an unfortunate history of having been separated from their language in daily use.
@@AKnightofIslamicArabiaTruly, it was even before that. The Name ceased to be used even in the few places it was not taboo-- that being the Second Temple, by the High Priest-- even from c. 300 BCE. So even while Hebrew was spoken as the common tongue, the Name's vowels were known theoretically by one or two people who never said it aloud, but whispered it to themselves while others shouted so that none would hear it. An interesting hypothesis for its vowels interprets the Name as meaning, "He Who creates existence", but again, just a hypothesis.
@@AKnightofIslamicArabialong vowels are written and short vowels can also be written, with diacritics.
This was to thwart pronunciation, so the concept of the right vowels is meaningless.
Good grief! You're terrific! The matter of Semivowels/semiconsonants becoming vowels emerges in the language I am grappling to understand: Hebrew. The consonants Yodth, Waw and Heh often operate as "I" "U/O" and "AH". It was dandy to see your describe the phenomenon from a new, crisp perspective. A review of your other video titles suggests gold strikes ahead for me.
This is some complicated s#!+, Simon. It is rather amazing that linguists have done so much with PIE reconstruction. I can see that there is a necessity to be a specialist in this field.
Can you do a deep dive into proto italic? I love your proto Germanic vids but italic went on to become latin which is pretty cool
It's weird to me how similar PIE's reconstructed vowel system is to modern standard Chinese, which also only has two vowel phonemes if you consider /i/ and /u/ as underlying glides. The Chinese vowels are /a/ and /ə/ so it's a bit different but it's weird to think two extremely influential languages millennia apart may have this same very unusual system.
As far as I remember, this analysis is not the most convincing and widely accepted by linguists.
PIE is just a half baked guess of how it might have looked like.
@@user-qh4dr1vy9dGuess, sure. Half baked? No. Reconstructing this language has kind of been the foundation of academic linguistics for a century and a half, so I think a lot of thought’s been put into it.
@@grahamh.4230don't think you need to attempt to give nuance to this guy, but nice efforts
It's kinda like how R-colored vowels are extremely rare... Except for the 3 billion people who speak the two most spoken languages on earth: Mandarin and English.
Loved the video.
Of the many things I could comment on, I'll stick to the vocalisation of the [ł] into [ʊ] that happens in your dialect of English.
It also happens in Portuguese (at least the Brazilian varieties I'm familiar with): "Brasil" /bra'zil/ becoming [bɾɐ'ziʊ] or even [bɾɐ'ziw].
That's one of my tricks for teaching my Uruguayan middleschoolers why the name "Daniel" sounds completely different for them when pronounced in English (vs Spanish). They immediately recognise the phenomenon of vocalisation, because they can all imitate the Brazilian pronunciation for the country, given that we're exposed to the phrase "Brasil, o mais grande do mundo" since infancy 😅
It's not uncommon, even here on the other side of the pond. My father would pronounce "owl" as /ˈæʊ̯w/. He'd also pronounce "iron" as /ˈɑɹn/; that means that there were more homophones than in most varieties of English: "fire" and "far", "tire" and "tar", etc. It also means that in my childhood, I overcompensated with words like "jaguar" and "reservoir", pronouncing them with /aɪ̯ɚ/ instead of /ɑɹ/.
Maybe another good example in English is the addition of an initial glottal stop (or sometimes an r) to enunciate words starting with vowels. This occurs only when there is no immediately preceding consonant from a previous word. The difference between the presence and absence of the glottal is not phonemic. And native speakers mostly wouldn't even know the difference unless it was brought to their attention.
This is my favourite corner of TH-cam.
This is going to sound weird as hell, friend, but there is like this *one* hair at the corner of your right lower lip that curls in towards your mouth and was making a line over that corner as you talked, and it was super distracting, and I was wondering if it was some artifact of the video recording or something. I eventually got so bothered by it I started pausing and unpausing the video until I caught it. I don't know how it escaped while you were shaving, but that's mad. Anyway, your videos are good and I like to watch them, but that was driving me nuts.
the vowel-glide alternation totally reminded me of the Salish languages, many of which have that exact kind of allophonic alternation. Some languages in that family also have syllabic sonorants making the nucleus inventory of PIE make way more sense
12:39 Wonder how much the way native speakers think about the sounds they're making is affected by knowing how a word is meant to be spelled, given that most adults tend to be literate. It would be interesting to see if a child would describe these differently
Most adults would probably still be able to intuit phonemes in their language through morphology. For example, “star” may be realised as /stā/ in non-rhotic dialects, but the r would be clearly pronounced upon adding the suffix “-y” to yield “starry”.
Not quite so clear cut. A lot of people would pronounce it "stahwy" or "Stowey". Etimology could be lost within a single generation.
I just learned about this and was confused, then this video pops up.
Hey I love your videos. I speak 6 languages including Welsh. I always wanted to learn olde English. I'm a pyglot? But not a linguist
But I love your channel I like listening to you and I think you are very handsome. Xxx no offence at all. Keep up your videos. I watch tham. Xxxx
This reminds me very much of one of the main problems with reconstructing Sumerian phonology. There was no universal system for syllabic spelling in cuneiform inscriptions, only general conventions, which show mild variation across time and both between and within individual regions. Trying to pinpoint underlying phonological structures thus becomes a matter of assessing individual spelling variations and comparison of the behavior of certain loanwords to ascertain what was most likely the most “common” form of a particular word or morpheme. That’s only the tip of the iceberg and I’d be happy to go into further detail, but for brevity I will just say that reconstructing a dead language’s phonology is one of the most difficult feats in linguistics, and one that ultimately remains forever tentative and somewhat speculative.
Really very good: illuminating clarity of explanation (seriously) ! Just one thing: when you summarised IE languages as ' most European and some Indian', what about Persian ? What have you got against Persians ? This is obviously some deep and sinister prejudice ..........
I find your sentence construction and diction significantly easier to comprehend than someone like Jackson Crawford. It is surprising that a fair few people in this field have awkward, halting verbal delivery, and swallow key words so you can't quite catch what they say.
This is really interesting when you think about the extremely rich phonemic vowel structure of Germanic languages!
"Rich" is one way of putting it.
I have wondered that. Thank you.
Does the story continue?
Overall a great video, but I think its an overstatement to say that phonological systems devised by linguists necessarily closely resemble our own mental understanding of our sound systems. We have no way of knowing whether PIE speakers thought of [i] as a type of /j/, or if they thought of [j] as a type of /i/. It's clear that theres a phoneme encompassing both [j] and [i], but whether it's fundamentally a /j/ or a /i/ is somewhat arbitrary and maybe not even meaningful. Fitting this phoneme in with the other sonorants which can be syllabic is convenient and creates a neat symmetrical system, but it's not necessarily a description of an underlying systematic reality.
A great example of where people's intuitions about their phonological systems differ from linguistic models based on systematic sets is the Danish "soft d" approximant sound which may broadly be written /ð/ or /ɤ/, depending on what linguists you listen to. Danish people very much perceive it as a consonant, a kind of d or eth, since that is what it has been historically and it is how it is represented in the written language. Some linguists like Ruben Schachtenhausen argue instead that it is a vowel, since it is both acoustically a vowel (though the acoustic boundary between an approximant and a vowel is arbitrary) and it belongs to a set of halfvowels in terms of its systematic behavior. So linguistic phonological models can absolutely differ from native speaker mental models.
A great blog post (in Danish) by Schachtenhausen about the decision to treat soft d as a vowel is called "Utilpasset IPA (bonus): Det bløde d".
At what point are you going to admit you are a bloody linguist 😂
A rose by any other name --
When he gets his phd
Been a minute since I checked in. Looking healthy Simon 💪
"You too could have discovered laryngeals." This is the first time I heard or read and explanation of laryngeals that is memorable and motivated. Thanks for the clear explanation.
Proto-Proto-PIE had no vowels at all.
Your remarks about the closing "l" (as in "simple") in English are interesting. It's also something I've noticed before, and although I'm very much not a linguist, I wonder if it's connected to something similar observed in other languages. For example, in Polish the letter "ł" is nowadays pronounced straightforwardly as /w/, but descends directly from standard /l/ (or л) as its orthography, and cognates in other Slavic languages, betray. However I don't know of any further examples of this /l/→/w/ shift, so I can't say how widespread it is.
In Belarusian and Ukrainian. The Belarusian “л” turned into “ў” and in the Ukrainian “л” turned into “в”. They are read the same way as “w” (ў) although Ukrainians do not know how to read their language. In Russian the “л” remains as it should.
Ukrainian and Belarusian were the same language in the 16th century
It's been a thing in Southern British English for quite a while. For instance, the city of Bristol is named what it is due to a hypercorrection: the historically accurate name would be "Bristow", but due to the vocalisation of /l/ to /w/, people perceived it as ending in /l/.
You are too humble Simon, you are a very qualified linguist!
In this case qualified probably means going through formal education on linguistics, which as far as I'm aware Simon has not.
@@poudink5791 I think that part was obvious.
Re the w/u, y/i thing - you can see this in many English words that are borrowings from Latin - e.g. "solVe" vs "solUtion" - when you understand that in Latin a "V" was actually pronounced as a "W". The Latin verb had the stem "solW-" and the W was realised as a "v" or a "u" depending on where it appeared in the syllable. One thing that you said about syllabic consonants made me think a little: "... and destroyed this kind of symmetry that existed within the system...". The problem is that when reconstructing you are going to choose the most parsimonious (i.e. simple and powerful) solution to your problem i.e. it is almost guaranteed that you will travel from something that is more complex to something that is more simple... So, is the "destruction of symmetry" when you start travelling forwards again merely an artefact of your reconstructive process or did it actually happen :-) ?
Thinking a little more about this - I would think it unlikely (although I haven't investigated this) that any child language would inject unnecessary irregularity or complexity into its parent. I would expect that as each generation reinterprets and regenerates their languages grammar they must make a similar number of decisions to their parents about the way that the language is patterned and sometimes these decisions will be different i.e. things will be grouped differently - but perhaps the total simplicity/complexity ratio of the language as a whole does not change [much]. In which case, the reason above for travelling from complexity to simplicity is that you are travelling from a number of child languages to a single parent language and not that any given child is necessarily more complex/less regular than its parent.
Classical Latin had no 'y' making do only with an 'i' - for exactly this reason thus the Romans had no problem spelling their God's name "Iupiter" (notice the two 'i's one at the beginning of a syllable one in the middle) whereas we prefer "Jupiter" (OK - you also have to understand that 'j' was pronounced 'y' at some point... :-) - nothing is easy...)
i think of the le at the end of simple as a syllable - ul - thats how i say it too, symppul. maybe the ending L becomes a vocal fry on the u instead sometimes - like how an r after a vowel lengthens the vowel
Whispering "simpulll... simple... simpouu" to myself here at work, getting strange looks
Can you do some spoken PG like you do with OE?
I think the two vowel system makes a lot of sense considering that PIE had a lot of consonants. This means that there is many distinct syllables you can make with even 2 vowels and you don’t need many to differentiate between words.
It’s like the modern languages of the caucuses(idk how to spell it)
I've seen theories that the two consonants may have been an areal feature of the region, since PIE was just north of the caucases which is legendary for their very extreme consonant-to-vowel ratios.
@swagmund_freud6669 There's also a fringe theory that Indo-European was related to some of the languages of the Caucasus called Proto-Pontic. Who knows how valid it actually is, but the languages still could have interacted thousands of years ago, even if it's not.
I am pretty sure Proto European did not have only 2 vowels - it had all 5 main vowels A / E / I / O / U and then the other ones like Norse and other Germanic languages also got the fancy ones Æ / Ö / Ü etc! I’ve even seen a list of reconstructed Proto European words before and it had words that had all 5 main vowels! There are even some words that haven’t even been changed a lot and some that may still be the same, which is why some words have exactly the same spelling in most European languages, so one could assume those words looked the same in Proto European or were only modified a bit!
It actually makes no sense not to have at least the 5 main vowels that all kids naturally produce - obviously the dude that made Proto European knew that one can produce at least the 5 main vowels, in fact, it may have even had a schwa sound, at least in pronunciation, not necessarily in spelling!
From 2 vowels in PIE to 26 vowels in Danish
im 90% sure danish is a hoax
Had the early PIE speakers simply decided to talk like they had five potatoes in their mouths, they could have shortened the period for that vocalic evolution by several millenia!
Many New Zealanders (not all, probably not most) make a vowel of postvocalic l (L). I think this is because we have a couple of vowel segments that are pronounced differently when followed by an l. The vowel in 'fool' is a high back rounded vowel, whereas that in 'food' is a high central rounded vowel. Similarly, the diphthong in 'coal' glides from an upper mid back vowel to a high back one, while that in 'code' goes from a lowish central position to a high central one. So when someone pronounces one of back vowel versions without the following l, the Kiwi brain supplies it. I only notice the lack if I'm listening for it.
And these L-droppers have extended this to other sequences of vowel+l, turning it into a diphthong of high back rounded vowel+l. And as in your dialect, a syllabic l, as in 'simple', becomes a rounded high back vowel.
The /w/ and /j/ being more of consonants makes sense since the Semitic written languages had no vowels but did have letters for /j/ and /w/.
Thanks a lot
Thank you,
Proto-Indo-European sounds like it went through a similar process that Proto-Slavic did, where reduced vowels (in Proto-Slavic, front ь and back ъ) began to be assimilated into or reanalyzed as consonants.
In what Slavic language(s) did the yers get reanalyzed as consonants? As far as I know, they were either lost, remained a shewa (in Bulgarian) or merged with another vowel (a, e, o)
The vowel system of Mandarin Chinese comes to mind here, at least some interpretation of it. More structuralist analyses of Mandarin pose two phonemes, a ~ ɛ and ə ~ i ~ ɤ ~ o, with the vowels agreeing on neighbouring consonants' backness and labialisation. Similarly to PIE, this analysis also proposes vocalisation of approximants /j/ /w/ /ɥ/ /ɹ/ when they form a syllable
Viceversa, PIE speakers might have thought of *y and *w as vowels which had consonant allophones in syllable onset.
The Romans did indeed spell /i/-/j/ and /u/-/w/ the same, respectively I (i) and V (v).
I was kind of surprised that despite being English, you had not mentioned words like bottom, bottle and button as examples of syllabic consonants, but the explanation of your own dialect made it clear why you didn't think of that.
I have doubts that they would be considered syllabic consonants in his dialect of English
It’s way more of a US thing imo
My previous comment seems to have disappeared, YT can’t like people adding links to Reddit but there’s a discussion there of exactly this phenomenon of PIE having only 2 vowels that’s also worth a read. It’s not covered as well there as here in Simon’s channel though of course!
Yachoob doesn't like external links.
0:50 whoo Tocharian, I am a Tocharian fanatic.
I'm not convinced that PIE even only has two vowel phonemes. Or even two vowel morphophonemes.
Morphophonologically, *i appears in reduplicated verb stems in locations where it doesn't correspond to a consonant, as if it was actually an otherwise-lost ablaut grade
Phonemically, a coda glide in a root is always vocalised in the nasal-infixed stem, even when the nasal infix takes the Ø-grade. For example, in the verb *kʷrinéh₂ti we have the 3rd person plural present indicative *kʷrinh₂énti, whereas if vocalisation of resonants was purely allophonic we'd expect kʷṛyṇh₂énti
Phonetically of course, we have at least five. Vowel colouring by laryngeals is attested in all branches and so must be assumed to have already taken place in PIE, even if it was still purely allophonic at that stage. Additionally, if Brugmann's law truly doesn't affect o resulting from vowel colouring we either need to reconstruct two separate o-like vowels (e.g. *o & *ɔ), one as the original *o, and the other from a coloured *e
So, from my perspective we have to consider three levels of analysis:
Morphophonologically: 3 vowel morphophonemes, of which one (*i) is marginal
Phonemically: 4 vowel phonemes, of which two (*i & *u) are marginal, as well as 2 marginal syllabic resonants and 3 marginal syllabic laryngeals
Phonetically: 5 or 6 vowel phonemes, as well as 2 syllabic resonants and 3 syllabic laryngeals
I can't believe you didn't mention Esperanto. :)
Thanks from sunny Vienna, Scott
How did Greek get the PT sound, like in Ptolemy.
Did there used to be a vowle between them? Or is it something derived from some other, non Indoeuropean, preexisting language that it speculated that Greek merged with ?
Syncope. There indeed used to be a vowel between the /p/ and /t/, but it was lost in Ancient Greek. This might have been due to patterns of stress in Proto-Greek triggering vowel reduction and loss. I would guess πτερόν would be a useful example. You can see this lost vowel in a cognate word in English with the same meaning: feather. I expect something similar happened with πτόλεμος, but my knowledge of Ancient Greek is thin at best.
Seems like latin i/j might be an instance of a vowel mascarading as a consonant. "iugera" = "jugera" etc.
Same with u/v, Latin didn’t differentiate these letters.
12:27 "In certain situations what's underlyingly a consonant can surface as a vowû"
"Most modern European languages and some Indian subcontinent languages," I know you acknowledged that you can't list them all, but for the record, that summation excludes the Iranian branch entirely, so like Farsi, Dari, Tajik, Kurdish, etc. You could say that description excludes Armenian too, although Armenian could be sort of be lumped in with "European languages."
Goodness when ARE you gonna go get that linguistics degree?
Fascinating
0:45 can't forget Persian and all its siblings! And other cousins like Kurdish and Pashto
At 19:40 you say that all the syllabic laryngeals are neutralized, but then how do you explain Greek (and I believe Armenian?) having separate reflexes for all of them?
Armenian may have neutralized the laryngeals outside of word-initial syllables, but of course due to the complex development of the phonology of Armenian, there is some dispute about the outcomes even in word-initial position. It seems there is some agreement that a similar 'triple reflex' to the one found in Greek is also found in the Phrygian language, but it would sure be nice if we had more material in Phrygian than we do.
Another thing I'd mention is that when laryngeals were pronounced as syllabic nuclei in Sanskrit, they appear to have been reflected as /i/ rather than /a/, so there are certainly complexities in all of this. This development is not found in the Old Iranian languages.
I'm not too familiar with Russian, but it might be a good example of natives analyzing sounds based on phonemes/archiphonemes differently than nonnatives. That's my guess based on its excessive palatalization (why else would they have a specific letter for "don't palatalize")
You seem to conflate writing and phonetics/phonology.
The letter you mean (the hard sign, Ъ) is not used to say "don't palatalize", but more often as a divisor between morphemes (typically prefix from root). Also, it's typically mean a /j/ sound.
The hard sign is here in Russian writing for mostly historical reasons. Historically it was used to represent a vowel sound.
It doesn't have a specific letter for "don't palatalize". It has 2 letters (ь and ъ) that mean the "vowel letter" и, е, я, ё or ю should be "read" (pronounced as red) as ji, je, ja, jo or ju respectively (I didn't take vowel reduction into account, if I would, then it will be ji, ji, ji, ji or ju in unstressed syllable). ь just means that the consonant before jV is soft, and ъ means that it's hard (that explanation is little bit simplified)
Examples (I tried to include only examples where jV is stressed):
ь:
воробьи́ [vərɐˈbʲji]
релье́ф [rʲɪˈlʲjef]
судья́ [sʊˈdʲja]
жильё [ʐɨˈlʲjɵ]
бью ['bʲju]
ъ:
(there seem to be no words with ъи)
подъе́зд [pɐˈdjest]
объя́ть [ɐˈbjætʲ]
объём [ɐˈbjɵm]
отъюли́ть [ɐtjʉˈlʲitʲ] (couldn't find stressed example)
P.s.
And without ь or ъ, и, е, я, ё and ю are pronounced ʲi, ʲe, ʲa, ʲo and ʲu after consonants and i, je, ja, jo and ju after vowel (or, taking vowel reduction into account and talking about unstressed syllabels, ʲi, ʲi, ʲi, ʲi and ʲu after consonants and i, ji, ji, ji and ju after vowels (tho ji after vowel can turn into i sometimes)
@@F_A_F123actually there are words with ъи, but they spelled with ы (ъi)
@@michailreznik5670 not really. They would be pronounced with C(hard) + /ji/, but words with ы are pronounced as C(hard) + /i/.
@@F_A_F123 why they would be pronounced with C(hard)j if /i/ only denotes ʲi but not ʲji? At the same time ы is actually a ligature of ъ and и, съ + играть > сыграть
Much of the problem is an artifact of the vowel-consonant definition. If you recategorize vowels as syllable nuclei, and consonants as syllable onsets and codas, many difficulties issues evaporate. PIE had many different sounds forming nuclei, so the "2 vowel" claim becomes nonsensical.