Valyrian Adjectives

Okay, as I write that title, I’m now thinking I can’t promise I’ll say everything about adjectives, but I’ll say some things. Is that cool?

High Valyrian was primarily head-final, meaning that adjectives usually preceded the nouns they modified. It actually makes more sense to start a discussion of adjectival inflection by looking at adjectives that appear after the nouns they modify, though, as prepositive adjectival inflection can be seen as a reduction of postpositive adjectival inflection. (I’ll explain this in a second.)

Though nouns have a number of different declensions, adjectives comes in three major varieties which I’ll exemplify using these three adjectives below (for expository purposes, the endings are lunar nominative singular):

  • Class I: kasta “blue, green”
  • Class II: adere “sleek, smooth, slippery, fast, quick”
  • Class III: ēlie “first”

Couple semantic notes on the above. Kasta is a word that can refer to anything that’s in the blue-green spectrum. Such words are common in older languages which tend not to have as many lexical color terms as a modern language eventually does. For a more in-depth treatment of this phenomenon, check out this post on Dothraki color terms from a while back. Second note is that adere probably first meant “slippery”, and then went on to develop the other senses.

The adjectives above are grouped the way they are because they inflect differently. Class I adjectives are the most informative, as they will decline differently for every case, gender and number combination—or almost. As with subject-verb agreement, adjectives only display partial number agreement (all adjectives, not just Class I adjectives). While a noun can appear in the singular, plural, paucal or collective numbers, adjectives only have singular and plural forms. In agreeing with a noun, an adjective will show singular agreement with singular and collective nouns, and plural agreement with plural and paucal nouns. The same is true of subject-verb agreement.

With that out of the way, this is what the inflection of kasta looks like in the singular:

Class I
Lunar Solar Terrestrial Aquatic
Nominative kasta kastys kaston kastor
Accusative kaste kasti kaston kastor
Genitive kasto kasto kasto kastro
Dative kastot kastot kastot kastrot
Locative kastā kastȳ kastot kastrot
Instrumental kastosa kastosy kastoso kastroso
Comitative kastoma kastomy kastomo kastromo
Vocative kastus kastys kastos kastos

And here it is in the plural:

Class I
Lunar Solar Terrestrial Aquatic
Nominative kasti kastyzy kasta kastra
Accusative kastī kastī kasta kastra
Genitive kastoti kastoti kastoti kastroti
Dative kastoti kastoti kastoti kastroti
Locative kastoti kastī kastoti kastroti
Instrumental kastossi kastossi kastossi kastrossi
Comitative kastommi kastommi kastommi kastrommi
Vocative kastis kastyzys kastas kastas

Adjectives of Class II and Class III are distinguished by not having declensions that correspond to each gender. Instead, both classes group the solar and lunar genders together and then the terrestrial and aquatic genders together. Thus (and what is, by far, the most exciting part for me) each class can be represented with a single table. Behold!

Class II Solar/Lunar Terrestrial/Aquatic
Singular Plural Singular Plural
Nominative adere aderi aderior aderiar
Accusative adere aderi aderior aderiar
Genitive adero aderoti aderȳro aderȳti
Dative aderot aderoti aderȳro aderȳti
Locative aderē aderoti aderȳro aderȳti
Instrumental aderose aderossi aderȳso aderȳssi
Comitative aderome aderommi aderȳmo aderȳmmi
Vocative aderes aderis aderios aderīs

And now for Class III:

Class III Solar/Lunar Terrestrial/Aquatic
Singular Plural Singular Plural
Nominative ēlie ēlī ēlior ēliar
Accusative ēlie ēlī ēlior ēliar
Genitive ēlio ēlȳti ēlȳro ēlȳti
Dative ēliot ēlȳti ēlȳrot ēlȳti
Locative ēliē ēlȳti ēlȳrot ēlȳti
Instrumental ēlȳse ēlȳssi ēlȳso ēlȳssi
Comitative ēlȳme ēlȳmmi ēlȳmo ēlȳmmi
Vocative ēlies ēlīs ēlios ēlīs

Now Class II has a couple of subclasses which I won’t get into here, but these are the main three declension patterns you’ll need to know to correctly inflect postpositive adjectives.

Now for prepositive adjectives.

Rather than redoing the tables, I’ll just make some comments. For the most part, a prepositive adjectival form will lose its final syllable when the inflection is disyllabic. This means that you’ll lose the -ti in all forms that have it, as well as the -si in instrumentals and -mi in comitatives. Word-final -t is also lost unless the adjective modifies a vowel-initial word. Here’s an illustrative example using the dative:

  • aderot ābrot “to the quick woman”
  • adero Dovaogēdot “to the quick Unsullied”

This does mean that in the nominative and vocative plural you get, for example, kastyz rather than kastyzy (nominative) or kastyzys (vocative). That word-final -z usually devoices to -s unless the following word begins with a vowel or a voiced consonant. Another example:

  • kastys hobresse “blue goats”
  • kastyz dāryssy “blue kings”

Where a disyllabic inflectional form is simply VCV, only the final vowel is lost, not the final syllable. For example:

  • ānogro ēlȳro “of the first blood”
  • ēlȳr ānogro “of the first blood”

You’ll see this most often in singular instrumentals and comitatives, in addition to terrestrial/aquatic genitives of Classes II and III.

Finally, Class III needs some special attention. For forms that modify a solar or lunar word, where a shortening would leave the final syllable with ȳ, that vowel changes to io. The same is not true of the terrestrial/aquatic. Here are some illustrative examples:

  • valosa ēlȳse “with the first man”
  • ēlios valosa “with the first man”
  • daomȳssi ēlȳssi “with the first rains”
  • ēlȳs daomȳssi “with the first rains”

And a couple of final notes. First, as those who’ve been studying High Valyrian nominal declension will know, many paradigms often level the distinction between the instrumental and comitative (some using a comitative m form for both and some using an instrumental s form for both). When an adjective modifies a noun, it will agree with the split. All adjectives, as a result, have distinctive m and s forms, but for a particular paradigm, it may only inflect with one of the two.

Second, High Valyrian is in the process of eliminating word-final m (or, to put that more accurately, High Valyrian’s never liked word-final m), so contracted forms that end in m often only keep that m if the following word begins with a vowel or a labial consonant. Otherwise, that m becomes an n.

That should be enough to get things going with adjectives! To conclude, here are a couple notes on some things that came out in recent interviews. First, while I have provided translations to George R. R. Martin when he requested them (whether he used them or how can only be determined when the books the translations were requested for are published. I still haven’t gotten a chance to look at the maps book to see how those translations worked out), I never said I provided Valyrian translations. That was an assumption on the reporter’s part. Second, I recently did an interview for Entertainment Weekly’s radio program. Somehow my middle name came up, and at the end of the spot, one of the hosts guessed my middle name—or so I thought! When they repeated it at the interview’s close, I could have sworn they said “David Jasper Peterson”. If that is the case, then I’m afraid I misheard them the first time—i.e. they said “Jasper”, but I thought I heard my actual middle name. I hereby go on record to say that my middle name on my birth certificate is not Jasper, though I’d certainly like that name better than my actual middle name, which is terrible. My apologies to EW!

That concludes this initial look at adjectives in High Valyrian. I planned to include adjectives in Astapori Valyrian as well, but this post got too long… Another time.

OH! Almost forgot. The Valyrian section of the Dothraki Wiki is live, and it looks oustanding! Take a look at the High Valyrian vocabulary page, for example. There’s tons of interlinking examples throughout the wiki and a lot of good info. Excellent work!

A lot of hands went into putting the wiki together, but there are a few people who did the most work. Hrakkar did a lot of the behind-the-scenes work (with some help from our old friend Lajaki!) to make sure the wiki works the way it ought and all the links are correct. Then the bulk of the content was generated by Esploranto (a.k.a. Najahho) and Mad_Latinist, who’s rivaling me for the most frequent commenter on this blog. Kirimvose! It looks great!


  1. Great stuff! :)

    So did the acquatic declension branch off from terrestrial in proto-Valyrian? Most of the forms look like they derive from terrestrial noun whose stem ends in {-r}, which loses the nom:sg {-on} and acquires an epenthetic vowel. Weird that some rare forms don’t have the {-r-}, though; those don’t happen to be typos…?

    1. Loc. kastot is surely a typo, but the vocatives kastos and kastas are not: it seems to be a consistent rule that aquatic paradigms drop the r in the vocative.

      1. Loc. kastot is surely a typo, but the vocatives kastos and kastas are not: it seems to be a consistent rule that aquatic paradigms drop the r in the vocative.

        Ugh. Yes, this is correct. I do hand-coded HTML for all these posts, and filling out tables in their HTML form is much less intuitive than filling in tables in, e.g., a Word document. I was typing all the forms by hand and messed up on kastrot. If that was my only typo, I can feel pretty good about that!

        But, yes, there are no r’s in the vocative for aquatic nouns. That one isn’t an error.

    1. “Sleek, smooth, slippery, fast, quick?” When someone says aderot ābrot you never know how many entendres you have!

      On a more serious note, adere will be the etymology of aderī, which you accurately (as usual) guessed meant “soon.”

  2. Well, I had intended to add all this material to the wiki right away (you might have seen that I have been working on the last couple days), but this is a bit overwhelming, and I don’t think I’ll get it done until after I give my midday Latin lesson.

    Here’s just a couple comments and questions for now:

    1. On the HV vocab page to which you linked, I’ve been carefully marking adjectives which are attested in prepositive position (after all, you told me “All adjectives can be postpositive if they want, but especially those that are more determinative in nature generally come before the noun, unless you want to sound…official?”). I guess I’m a bit miffed to learn that kasta can be prepositive, since I’ve been using it as my example of a Class I postpositive. Perhaps I should just give up on the idea of using separate adjectives in the prepositive and postpositive sections… it might be better that way anyway.

    2. I’m guessing the citation form of hobresse is *hobres? That means we now know the of zaldrīzes! Also, we can say “Ska tala ja hubre? in HV (well except that tala might not have the same sense there)

    3. On that note, is the AV word for owl *atroske? ;)

    4. “…and Mad_Latinist, who’s rivaling me for the most frequent commenter on this blog.” That is probably not something I should be proud of ;)

    Anyway, I’m sure I’ll have more questions later today when I get around to trying to codify this information.

    1. Perhaps I should just give up on the idea of using separate adjectives in the prepositive and postpositive sections… it might be better that way anyway.

      To be honest, in my own document, there’s only one table for adjective classes—I just use parentheses (e.g. kasto(ti), kastos(si), etc.). It gets more complicated with forms that change their final consonant or Class III adjectives, though, so I decided not to do it that way on this blog post. But, no, at this stage there isn’t really a difference semantically between prepositive and postpositive adjectives. The first kernel of change is the fact that determiner-like adjectives favor prepositive as opposed to postpositive position.

      3. On that note, is the AV word for owl *atroske? ;)

      Actually depending on the ordering of those two sound changes, it could be atruske. I’d have to check.

      1. My understanding was that any adjective could follow the noun, but only some could precede. Was this incorrect? (That’s the important thing)

        I guess I wasn’t assuming there was a specific semantic difference (though that might be a fun area for you to explore… like French ancien prof “former teacher” vs. prof ancien “ancient teacher,” or Classical Greek ἀγαθός ἀνήρ “good man” vs. ἀνήρ ἀγαθός “one who has proven his manhood by dying in battle”—you won’t find that latter distinction in your Greek books, but try looking at Classical literature, and you’ll see it’s almost always true!) more that the position of the adjective might change the style, emphasis, or subtle connotation.

        And yeah, occurred to me after I posted that there could be a vowel-raising there… I haven’t even tried to work out those sound changes yet.

        Some more questions about the original post:

        1. Daomȳssi is plural, right? So that confirms that daomio should be the main lemma, not daomior, which has no special idiomatic meaning.

        2. The instrumental vs. comitative thing… whoa. That’s fascinating. That means that on some level it’s not really a case merger, but a difference in case usage. Well, you did say that animacy had something to do with it, and one can certainlymakes some sense of that by looking at the declension tables.

        3. I just realized that remembered that back in this post, following a suggestion by Dinok, I wrote “Etirí must then mean early.'” Aha! clearly what I was hearing was the AV reflex of aderī (whatever that actually is).

        1. On 2: Yeah, that’s how I’ve been reading it. Second declination is mostly for people, and you’re supposed to be with people rather than using them… whereas third and fourth seem to be more for things, where the opposite is true. So the inappropriate cases are discarded and the appropriate ones fill the gap…

          It reminds me of how Indoeuropean has NOM=ACC for neutral nouns (I figure neutral nouns aren’t expected to show much agency!), or how in Spanish the preposition {a} covers both ACC and DAT meanings for people.

          1. Third and fourth? Sorry, I mean third. I got distracted by the fact that {-or} and {-on} both belong to third.

  3. I don’t really have much to say about Valyrian adjectives in particular, but I just wanted to say I think it’s really cool that you’ve put so much effort into making these languages, and that you’re willing to share it with us all with informative blog posts. Thanks!

  4. Would “Adero kastoti zaldrizo Rhaegalo” – “To the quick greenish Dragon Rhaegal” be correct? Trying to work with the language as I would with Latin, but I’m not sure.

    Thanks for all these great descriptions and explanations!

    1. Would “Adero kastoti zaldrizo Rhaegalo” – “To the quick greenish Dragon Rhaegal” be correct? Trying to work with the language as I would with Latin, but I’m not sure.

      Not quite. Here it is:

      adero kasto zaldrīzot Rhaegalot

      1. Thank you! As a lover of languages, I think I would one day like to work on something like ‘Learn Valyrian for Latin Scholars’. I think it would be a fun way of learning it!

      2. So what declension is Rhaegal anyway? Usually anything weird turns out to be 3rd declension, but that doesn’t look possible here. And, unfortunately the dative tells us next to nothing ;)

        1. I just put it into the sixth declension since it didn’t look like anything else. It’s not really a proper High Valyrian name (looks to me like it was inspired by some offshoot, perhaps). The name itself is based on Rhaego, Dany’s unborn son. As such I just treated it like a borrowing.

            1. Not sure what you mean by “(fem.)”, but Rhaegel was male, and son of king Daeron II ( There is also princess Rhaelle (, who married a Baratheon and is the grandmother of Robert, Stannis and Renly. Since the “e” is silent, it could be considered to have the -el ending. Anyway, good that you brought it up – it would seem Dany didn’t just conjure up a new random noun ending, which would have been rather strange. But given that they are so uncommon, maybe they just a variant of some other ending, like -er and -ar. They could also be “corrupted” by Westerosi influence, but how would the names work in High Valyrian then, given that it was spoken on a somewhat regular basis among the Targaryens? Or perhaps GRRM was just inconsistent in two/three cases, and there’s no point in making sense of it.

              But while we’re on the subject of noun endings, how does a name like Aelix ( fit in to the noun declension/gender system? Since Aelix is the great-grandson of “native” Valyrians, Daenys and Gaemon, with little or no Westerosi influence on their culture having occured, it doesn’t make sense for his name not to “work” in HV.

            2. In at least some of these cases it could be a matter of the Valyrian names being given in common Westerosi form, like for instance how Marc Antony is not a Latin name, but it is a Roman name.

            3. But it is actually named after her brother, Rhaegar, which is a very native Valyrian name.

              Ah, so it was. I thought she named him after Rhaego (whom, of course, she named after Rhaegar), but I misremembered. Son’s barely cold in the ground and he’s already playing second fiddle to another dead guy…

              But while we’re on the subject of noun endings, how does a name like Aelix ( fit in to the noun declension/gender system?

              Nothing wrong with “Aelix”. Just third declension (declines like baes, but with k where baes has h).

              BTW, is there evidence that the “official” pronunciation of Rhaelle is with a silent e?

              The “e” is not silent in the audiobooks I listened to. I wouldn’t treat it as silent.

            4. You’ve mentioned baes as a paradigm noun before (you decided Lys was of the same class), but what does it mean? And was I right that the word for god declines the same way? For that matter, what about the word for old? Jaes, uēpa??

  5. Kastor – aquatic class – sounds exactly like French for ‘beaver’ = castor. Probably a coincidence, but I do enjoy little things like that. I love your posts and learning more and more about Valyrian, it’s truly awesome !

  6. …and Portuguese, and Romanian, and Italian, and Spanish, and Catalan… :) I think beaver is “castor” in all romance languages (well, maybe not Sardinian or Venetian or something). The genus is called castor in English too.

    1. Not really in any way relevant, but the Chinese name for beaver is hélí, written as 河狸。

      河 = river
      狸 = tanuki, as in the thing that Mario turns into when he gets the leaf (they’re a real animal, English name is apparently raccoon dog).

      Which I find kind of cool. I want a flying Tanuki suit.

    2. DJP bardutas:

      And the beaver is my favorite animal! Though, indeed, this was a coincidence.

      I find this hard to believe, given that beavers are both aquatic, and… um… cyan.

      Chickenduck bardutas:

      …and Portuguese, and Romanian, and Italian, and Spanish, and Catalan… :) I think beaver is “castor” in all romance languages (well, maybe not Sardinian or Venetian or something). The genus is called castor in English too.

      And not surprisingly, the word is already in use in Classical Latin. But it’s borrowed from Greek κάστωρ and the more down-to-earth word is fiber, later beber.

      The French castor must be a learned borrowing too, since it doesn’t show the characteristic French soundchanges (cf. castellumchâteau).

      Well, this is way off topic. David, anything you’d like to share to solve this problem? ;)

  7. Emilia Clarke , Emmy nomination.
    First time for a actress speaking a language from a parallel universe?
    David you should be at the ceremony !

  8. Under the HV vocab page, zyha is translated to ‘your’. Shouldn’t it be translated to ‘ his’ or ‘her’?

  9. Maybe it is a bit late to ask, but I keep wondering how the examples “valosi ēlȳse ‘with the first man'” and “ēlios valosi ‘with the first man'” are built up. I see the differences between the prepositive and postpositive adjective; I also understand the ȳ–io alteration; but not the word “valosi”. I suppose it is a declined form of “vala”, yet in the corresponding declension table—which is already known—I can’t find such a form. As far as I understand, the comitative “valoma” should be used here, or maybe the instrumental “valosa”, should it not? Or is there something else I don’t know of?

  10. Names with Vh- (Vhagar, Vhassar) are from vernacular High Valyrian, dialectal or foreign words? Volantene triarch surnames are High or Low Valyrian words? What language did the Rhoynar speak?

    1. Well, probably those should be proper High Valyrian and maybe even archaic High Valyrian names, since it is said they were named after the gods of the Valyrians. The Rhoynar spoke their own language.

      1. Words with rh are also very few in the Vocabulary List, although they are common in Targaryen’s names. Curiously, Rh- is very common in Dothraki. What would be the closest relatives of Valyrians? Qarthenes, rhoynars, andals? I found no Targaryen’s name with explicit meanings. Could their personal names be borrowed from foreign languages. Rhoynar toponyms, hydronyms and anthroponyms have a lot of examples with N-, sounding like a kind of prefix.

        1. Whatever the overall rarity of is meant to reflect, I doubt DJP intends it as a statement as to the relation of the Valyrian language family to any other language. Nor of the “pure-blooded” Targaryens to anyone else.

          Note to DJP: for your next #conlangfc, or if you want to go back to giving us a word of the week, like we used to do on IRC, you might want to consider giving us some more rh-. It appears we have only one that is not a proper noun!

    2. This has come up before, and Najahho here came up with the brilliant suggestion that ‹vh› is a fanciful way of representing the sound [f], though that requires some deeper explanation, and it will almost certainly not be the way DJP goes. On this topic, he’s said:

      I figure it’s an affectation of a different spelling system, or just a misapprehension of how the system works. We’ll see, though. It’s all academic until (or unless) I’m able to do a HV orthography.</blockquote?

    1. That does appear to be the case, or at least if it is not from Valyrian, HV treats it as if it were: Vesteros, for instance, appears to decline just like ēngos.

  11. Andalos can be related to endie “western”. Pre-Andalos substratum is “ibbenoid”, most of Free Cities’ substrata can be Rhoynar. Qohor seems similar to Qhoyne.

    1. Of course you need to draw a difference between GRR Martin’s names and what his ideas were and David’s ideas and historical reconstruction for the languages he’s been creating. These are not the same thing, the stories were not created with all the details about the languages in mind.

  12. How to explain different ordinal suffixes -nie and -lie? Why saelie “third” instead of *hārelie or *hārie?

    1. We know from a very old facebook post that saelie is a suppletive form. As for -nie/-lie, that could just be assimilation. I believe we have other examples of l>n assimilation in HV.

      1. My problem in identifying assimilation is to detect original suffix.
        izula > izunnie: we can expect izul(a)+nie>*izulnie>izunnie

        But all the another numerals show a -lie ending.A solution would be a dissimilation, instead of assimilation, with *izul(a)lie >*izul(a)nie>*-lnie>*nnie

    2. Yeah, it’s just suppletive, like english “one” → “first” (not ˣoneth) or “two” → “second” (not ˣtwoth.)

      As for the distribution -nie vs. -lie, I’m really not sure, but since -nie only occurs in izula → izunnie, perhaps it’s dissimilation? Of course you also have kūrie with -rie, but who knows what the original cardinal form of that was. I suppose tȳne might also be a variation though.

      1. bȳre : byllie
        vōre : vollie

        Both show l-assimilation and long>short vowel.

        *by:re-lie > *by:rlie > *by:llie >byllie

        gār : gallie

  13. In pair sIkuda – sIglie it seems that middle syllables collapsed after syncope. *sIkuda-lie>*sIkdlie>*sIgllie>*sIglie

  14. I’m late to the party, but I was wondering whether Valyrian has any rules about the ordering of adjectives, for example, whether it would be “round big house” or “big round house”?

Leave a comment

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.