The Mad Latinist (jdm314) wrote,
  • Mood:
  • Music:

Valar Pizze Ipradis... yn nyke krespe sindigon jaelan!

Valar Morghulis... Valun JomorghulisSeems like I'm always getting these done at this last minute, even when I have two weeks. This post was actually all but done on Saturday night, when I went to bed. I figured I would go over it and post it in the morning. I wanted to get it posted before David J. Peterson posted—he had, after all, promised a big, all-grammar post, which would at least render my post obsolete, and possibly even "put me out of business," as it were. I figured Sunday morning was fine, because, from past experience, I expected his post to come out Sunday night at the earliest. I was wrong. When I woke up there was a new post, and sure enough this contained a lot of information that needed to be incorporated. I figured I'd just put it off, and post as soon as I could rewrite it.

Well, as you can see, I couldn't get that taken care of all week. At least I'm consistent in always posting right before new episodes of Game of Thrones airs. ETA: OK, OK, right after it airs. Sigh.

But back to our actual episode, "Second Sons," which aired back on May 19th. Once again, there was not much Valyrian this time around. On the other hand, Mr. Peterson has taken advantage of this time to teach us a good deal more than has been usual. We will begin with some of this new information, as well as some of my theories, arranged by part-of-speech, before moving on to the actual episode.

The relevant DJP posts are:
  1. Tȳni Tresi (TT)
  2. Some High Valyrian Inflection (SHVI)
Mr. Peterson also appeared on The Jeff Rubin Jeff Rubin Show, where he gave some more useful information, but given how long it took me to incorporate SHVI, I decided to save this for a later post.


In Second Sons' Valyrian scene, which we will discuss below, we finally got a 1p pronominal form in High Valyrian, something we have not yet gotten even in AV. Curiously the form was Ä«lvyz, which seems very odd. In TT, Peterson explains this as follows:
As I mentioned somewhere at some point in time, adjectives in High Valyrian have a different form depending on whether they come before or after the noun they modify. In this case, the full form would be Ä«lvyzy. The final y drops out if the adjective precedes the noun it modifies, though, and the z devoices unless the next word begins with a voiced sound. Since “guests” is zentyssy, then, the form of the adjective is Ä«lvyz and not Ä«lvys.
In other words:
  • Ä«lvyzy if it comes after the noun (recall that Peterson has said adjectives which precede the noun are also allowed to follow the noun, but it sounds more "official," or something.)
  • Ä«lvyz if it precedes the noun, and the next word begins with a voiced sound (do vowels count as "voiced" for this purpose?)
  • Ä«lvys otherwise.

Given other prepositive adjectives we've seen so far (e.g. all the adjectives we've seen before ēngoso: Quptenkos "common," doros "no," mirros "any"), I strongly suspect it is a general rule that prepositive adjectives loose final short vowels. By the way, this very likely explains kirímvose/kirímvos "thank you"—in origin it is probably an adjective in the instrumental case.</li></ul>

Note that this probably also explains the odd "rūso zȳhosy" in Talisa's letter, though I have no idea what rūso means, and I'm a bit confused about the cases: rūso looks like a genitive, but zȳhosy looks, if anything, like an instrumental (which, if it inflects like the 2nd declension solar, as we learned in SHVI, it shouldn't look like this).

Over at his tumblr (currently the third most popular. Way to go!), Mr. Peterson posted about the idea of translating names between languages—being a Mad Latinist I do not entirely agree with him, but that's beside the point. For our purposes, the most important thing is the cool HV example sentence he gave:
Davído zaldrī́zes aṓhos zaldrī́zose rovýktys íssa.
David’s dragon is bigger than your dragon.
We learn several things from this:
  1. The word for "bigger" is rovýktys, clearly cognate to AV rovája. While comparatives and superlatives of words for "big" are often irregular, I would not be surprised if -ktys, or the like, turns out to be a regular comparative ending.
  2. The instrumental of zaldrī́zes is zaldrī́zose. Generally it seems that instrumental singulars are formed by the infix -os- followed by the stem-vowel of the word in question (e.g. zaldrī́zeszaldrī́zose, ḗngosḗngoso, válavalósa, and so on. But for more on stem vowels, see the section on nouns, below.)
  3. High Valyrian uses an "instrumental of comparison" construction—in other words, one of the meanings of the instrumental case is "than."


Let's start with something small. We already knew that in some cases the collective number can mean "all" of something (e.g. válar "all men"), and in others it means an assemblage of something (e.g. azántyr "army.") It turns out this is at least partially predictable: apparently it regularly means "all" an the case of "ordinary human beings not associated with a profession."

On the large scale, we now have, for the first time ever, some complete noun paradigms. We have the first declension lunar (TT), example word vála "man" (no surprise there), and the second declension solar (SHVI), example word lóktys "sailor." For both of these paradigms we now have every case and every number, and it is glorious. I am not going to reproduce the full paradigms here, so do be sure to check them out yourselves, and look for patterns.

For vála the instumental and comitative are particularly interesting. As is the vocative: we already know the vocative singular válus from múñus (which seems strange enough already for those used to Latin), but now we have the plural vális, the paucal valússa, and the collective valárza. Stranger yet, in the paucal and the collective the vocative forms are identical to the instrumental. I've heard of a lot of case mergers in my study of linguistics, but instrumental and vocative just blows my mind.

In lóktys, on the other hand, the instrumental (which pesumably would have been *lóktosy) merges into the comitative (lóktomy). This is very likely because most -ys words are animate, and it's harder (albeit not impossible) to come up with a sentence where an animate entity needs to be in the instrumental than an inanimate one. Also worth noting that I previously thought the ins.s. form of this declension was -o, on the basis of Perzo VÅ«jita "kissed by fire," but now we know this can only be exactly what it looks like, the genitive. Apparently in some cases, the agent is marked by the genitive. In the plural, there is an interesting anomaly: the locative plural is not identical with the genitive and dative plural, which we had previously thought was the case for every noun in the language. We now know that the second declension is the only exception to this rule, among all noun and adjective classes. Note as well that the nominative plural ends in -yssy—an ending I have been stumbling over for a long time, but the solution turns out to be very simple.

Now, let's review what we already knew about noun morphology:
  • There are four genders: lunar, solar, terrestrial, aquatic. These seem to correspond so far to lexical forms ending in a vowel, an -s, a -n, and an ­-r respectively.
  • There are six declension classes, each one called after a number, just as in Greek and Latin. We already knew that vála was first declension, trḗsy second, ā́eksio third, but nothing beyond that.
  • The total number of paradigms is 21 (at present, at any rate: Mr. Peterson reserves the right to invent further paradigms later should it become necessary.)
    I hadn't gotten very far into figuring out how the declension classes worked. All three examples we had happened to be "lunar" nouns (ending in a vowel), so I asked about this, and Peterson's reply gave me enough of a hint to cause a sudden insight:
  • The 21 paradigms represent 6 declensions times 4 genders, with three or so combinations not occurring ("... so in a declension class, they’ll come in that order, if present.")
What else is there six of in the High Valyrian language? Vowel qualities, namely a, e, i, o, u, y. The declension classes correspond to stem-vowels!
This chart may make the theory clearer:
Lunar Solar Terrestrial Aquatic
1st decl.
-a *-as *-an -ar
2nd decl.
-y -ys X
3rd decl.
-o -os -on -or
?th decl.
-e -es -en *-er
?th decl.
*-i *-is *-in *-ir
?th decl.
*-u *-us *-un *-ur

Now keep in mind that these endings represent the nominative singular. We cannot count paucals (which seem to end in -n) and collectives (which seem to end in -r­). Endings listed in black are firmly attested. Those in dark grey have uncertain attestation, and those in light grey no attestation at all that I have been able to find. Note in particular the following:
  • We know that the 2nd declension only occurs in the solar and lunar paradigms, so any -yn or -yr nouns we might encounter (e.g. témbyr "book") must be relexicalized paucals or collectives, respectively.
  • E-stem lunar may be attested in the names Rhaelle, Alysanne, and possibly Rhae.
  • The buzdari declension used for foreign words may have originated as the i-stem lunar. For all we know, some foreign nouns wind up as i-stem solars, terrestrials, or aquatics. We don't really have any evidence here.
  • I-stem solar might be attested in Volantis. The problem is that we don't know for certain this is a High Valyrian form: names such as Tyrosh and Lorath suggest that the Free Cities' names are given in the local variety of Low Valyrian.
  • No u-stem forms are attested at all.

Depending on how we count, we have thus far seen between 10 and 15 of the 21 paradigms. Of course, languages are never as pristine as the theories we construct to describe them, so some complications are to be expected, and indeed Mr. Peterson has already stated that knowing what letter a noun ends in in the nominative singular will only get you the gender most of the time: if a noun ends in -y it is probably lunar, if in -ys it is probably solar, but you can't be 100% certain (One might compare Latin where -a (gen. ) is overwhelmingly feminine, and -us (gen. -Ä«) overwhelmingly masculine, but there are plenty of exceptions, like agricola m. and fraxinus f.)

Furthermore, if we look at the known High Valyrian names used in the books we have at least the following list of unclassifiable names: Rhaegal, Aemond, Valarr, Aelix, Syrax.
  • Aemond and Valarr look like they must be some sort of Westerosianized forms that have lost their endings entirely (in the same way that, for instance, Marcus Antonius is called "Marc Antony" in English.) But perhaps Peterson has come up with some other way to fit these into the system.
  • Perhaps Rhaegal represents a first declension terrestrial (a combination otherwise not yet attested), and *-an regularly becomes -al? In support of this theory, notice that Dany's other two dragons, Drogon and Viserion both have terrestrial names.
  • Aelix and Syrax don't really fit into the system so far. We do not yet have firm attestations of the solar paradigm in either a-stem or i-stem, so perhaps this, too, is a regular change. But I suspect it is more complicated that that. In any case this might conceivably explain the rather strange form oressiks in Talisa's letter, but another possibility will come up when we discuss verbs.
This analysis makes a good deal of sense. In particularly if you chart them out you will see similarities across gender within each declension class. For instance, in the first declension the locative is marked by a long a, so that vala becomes valā, and ānogar becomes ānogār. In the third declension, on the other hand, the locative merges into the dative, so that āeksio becomes āeksiot, and lentor becomes lentrot. Peterson has gone so far as to say that case mergers are "what defines declension classes in High Valyrian" (SHVI).


A major development occurred in the area of verbs as well. We learned that morghū́lis is not, in fact, a present tense at all, but actually an aorist, like ȳdrássis. This explains the translation "all men must die"—the aorist indicates it's a general truth of the universe, rather than something that's simply happening. As a result it often implies something unavoidable. (Cf. Mr. Peterson's explanation of why Daenerys told the Unsullied that free men iderḗbzi "are choosing," present tense, rather than iderḗbis "choose," aorist.)

How is this possible, formally? Well, apparently the -ssi- suffix is only used with thematic verbs, those which I informally call "-as­ verbs." For athematic verbs, those which I informally call "-zi verbs," a different set of endings is used, roughly the same as for an i-stem verb (but not identical.) The following chart lays it out (to the best of my ability at this point.)
"speak (a language)"
"get drunk"
Present Aorist Present Aorist Present Aorist Present Aorist Present Aorist
1s jáelan jáelin ȳ́dran ȳdrā́ssin síndin sindíssin únden undéssin mṓzun mōzússin
2s jáelā(?) jáelia ȳ́drā ȳdrā́ssia síndī sindíssia úndē undéssia mṓzū mōzússia
3s jáelza jáelis ȳ́dras ȳdrā́ssis síndis sindíssis úndes undéssis mṓzus mōzússis
1pl jáeli jáelī(?) ȳ́drī ȳdrássī(?) síndī sindíssī(?) undī undéssī(?) mōzī mōzússī(?)
2pl jáelāt jáeliāt(?) ȳ́drāt ȳdrā́ssiāt(?) síndīt sindíssiāt(?) úndēt undéssiāt(?) mṓzūt(?) mōzússiāt(?)
3pl jáelzi jáelisi(?) ȳ́drisi ȳdrā́ssisi(?) síndisi sindíssis(?) úndesi undéssisi(?) mṓzusi mōzússisi(?)
This -ia form must explain Thoros's rijī́biā "you worship"—though we'll have to ask Mr. Peterson about the quantity of that final a.

I had previously wondered about múñar ... téptas "parents gave" (with the verb in the singular) squared with válar morghū́lis (with the verb apparently in the plural.) Now I realize morghū́lis is, in fact, in the singular here! Nouns (and possibly adjectives) may have four numbers, but verbs seem to have only the singular and the plural as distinct numbers, so collectives, relexicalized or not, always take a singular verb. This is presumably also true of paucals, but we do not yet have any examples.

In fact, once SHVI came out, it finally became clear that in HV the general 3p ending is not, in fact, -is, as we've long thought it was, but rather -si. The difference between thematic and athematic verbs is not as great as I've been thinking. Rather, the form the endings take (generally: does the vowel precede the consonant or follow it?) is determined by what phonetic conversations are or are not lepermissible. Thus, *jáel·na becomes jáelan "I wish," and **mend·si becomes *méndis "They..." well, I can't translate it because Mr. Peterson chose as his example a hypothetical root, which does not (yet) have any meaning.

Note that we may have some trouble here. A verb ending in -is could be any of the following:
  • athematic, 3s aorist (like jáelis)
  • i-stem, 3s present (like síndis)
  • athematic, 3pl present (like *méndis)
Our only clue will be the phonotactic composition of the verb stem, and any other forms of the verb we've already seen. And since we don't yet know the rules for which combinations are acceptable, we're going to be relying on the latter alone a lot for now.

Note as well that there is apparently a related tense called the "past habitual" (I speculate that it looks like *teptássis and means "he used to give; he would always give."), but that Peterson himself does not always consistently distinguish between this and the aorist.

Mr. Peterson also let a passive form slip: válar iprádiks "all men must be eaten." This suggests another possible interpretation for oress oressiks: oressiks appears to be an aorist passive. Zhalio suggests it might mean "they [VERB] and are [VERB]ed" (for oress(is) oressiks?)—his favored guess is "kill," as a description of Westerosi politics, but in any case we've had other words for "kill."

Misc. Vocabulary & Phonology:

The HV etymon of AV tórgo "worm" is now known to be túrgon, which shows that I can't have been right about Dany repeating the name Tórgo Núdho back in its HV form (which, I'm guessing, would be Túrgon *NÅ«don, or the like).

I already mentioned valar ipradiks, but in fact this was a joke on the active form valar ipradis, which he had already given. On Facebook (here is the link, but it won't be useful for most people) Zhalio pointed out that this raises an interesting issue: stress is determined in High Valyrian by syllable weight, but this could theoretically be íp-ra-dis or i-prá-dis (in Latin such words can scan either way, according to the rule called muta cum liquida, though for purposes of accent, at any rate, they seem to have always divided the first way.) Mr. Peterson tells us that in HV the syllable will devide the latter way, and you'll get iprádis. I'll have to keep this in mind should I ever get around to writing hexameters in HV!

 A similar datum is that the "on-glide diphthongs," that is to say vowel clusters such as io, iō, ue, uē scan as if they were one vowel, short if the second vowel is short, long if it's long. Critically, the on-glide does not count as a consonant for these purposes: in other words, a word like obÅ«ljarion does not scan Ë£o·bÅ«l·ja·rí·on, nor as Ë£o·bÅ«l·jár·i̯on, but rather o·bū́l·ja·riÍ¡on. I could not quite picture this in my mind's ear until I heard how Mr. Peterson pronounced these diphthongs himself on The Jeff Rubin Jeff Rubin Show.

There are some cases where a combination of vowels that could be a diphthong is treated as two separate vowes, but they are not very common. The most notable example is daor "not," which may be pronounced either dáor (something like *[ˈdao̯r]) or daö́r (something like [daˈor]). The latter pronunciation seems to be used throughout the show.

One last note:

Zhalio's font is out! Go download it!

There are really only two audible lines of Valyrian in this episode. Apparently in the Second Sons' camp scene there was Low Valyrian in the background, but I have been unable to catch anything I could trasncribe. Also, I got confused at one point and thought Stannis was speaking Valyrian, but I was in fact comically wrong. (In my defense, Peterson himself was also confused when he first saw that scene.)

Our only actual dialog occurs in the scene where Dany meets with the leaders of the Second Sons:
Grey Worm: Nýa dáre, béza unéhtelas jáa éngo ózy?
Subtitle: “My queen, shall this one slice out his tongue for you?”
Nýa dáre "my queen," HV dā́ria (I'm guessing the complete phrase would be something like Ñúhys dā́rius?)

béza unéhtelas: "Shall this one cut out?" Other than the 3s ending -as this verb is a little difficult to analyze.  The u- could be the same as in uderḗbagon, which apparently means "choose," but without the applicative i-, the -t- could conceivably be the perfect ending, but it is followed by -el- which might possibly be a variant on the -il- imperfect ending; this would leave neh- as the stem... but it is very unlikely this analysis is entirely correct anyway.

jáa éngo: "His tongue." The main thing of interest here is jáa instead of zýa. I'm not entirely certain what the difference is,  but Peterson did hint that there might be more than one third person pronoun. Perhaps jáa is necessary because Grey Worm refers to himself in the third person, and if he said béza unéhtelas zýa éngo, it would mean "Shall this one cut out his own tongue?"

ózy: "For you," actually o "you" and zy "for," written as one word. Interestingly, in HV the pronoun would have to be in the genitive. The AV o- here could either be the attested 2s dative (from HV aōt), or conceivably it could be a fossilized reflex of the HV genitive (unfortunately we don't know what that is! *Aō syt, maybe?)

Daenerys: Bísi (váli) ī́lvyz zentýssy íssi.
Subtitle: “These men are our guests.”
Bísi (váli): "These (men)." Daenerys omits the word vali, but no big deal. Maybe. We will need to discuss HV demonstratives in more detail next week.

íssi: "are," 3pl. By private correspondence, Mr. Peterson has confirmed that issi is the 3pl, and the correct 2pl form is iksāt.

ī́lvyz zentýssy: "our guests." We have already discussed the fascinating ī́lvyz above. Zentýssy, being substantive, keeps its -y, and we now know this is the caracteristic of the second declension solar, so the singular must be zéntys.
Tags: adjectives, aorist, astapori valyrian, conlangs, david j. peterson, declension,, game of thrones, high valyrian, linguistics, nouns, valyrian, verbs
  • Post a new comment


    default userpic

    Your IP address will be recorded