Non-canon novels are not taken into account here, neither are replaced novels (most notably 'The Sword of Aldones' which was rewritten as 'Sharra's Exile').
According to a description given by Jeff Kerwin, Darkover has at least nine languages (BS:218), but most of them do not appear prominently. The languages appearing in the stories (apart from Terran standard) are cahuenga and casta, and we do learn a few words in the language of the Dry Towns and one expression in the language of the chieri. Apart from this, various mountain and hill dialects are mentioned, but it remains a bit unclear to what degree these are different languages.
cahuenga is described as the lingua franca of the trade city (SC:99), as city language (TH:319) or as lingua franca of Darkover between mountain and valley, Dry-Town and river folk (WD:628). In the early concept of Darkover, cahuenga was the name of a hill range (SD:78), but it is not used like this in the later texts where it only refers to a language. Thus, cahuenga seems to be a language which is an amalgam of several languages spoken across Darkover. Originally, it is said to be based on a mountin dialect though (HH:87).
casta is describes as a mountain language (BS:274), as a language of the Hellers (TH:319) and as pure in Arilinn and Nevarsin (TH:557). The latter statement is a bit puzzling, as Arilinn and Nevarsin are quite far apart, and Arilinn lies in the plains (BS:197). Nevertheless, MZB is consistent on this point - Auster Ridenow who learned cahuenga in his childhood in the orphanage refuses to speak the language in Arilinn and uses only casta (BS:303,326). Furthermore, casta is the language used normally by the Comyn, the aristocracy of Darkover (HH:88). WW:189 describes casta as 'archaic and little used' but this may just be an early concept abandoned by MZB later on.
It is quite clear that casta is heavily based on Spanish and Latin and that cahuenga involves Celtic elements. This is openly acknowledged by MZB through Paul Harrell in TTC:381 where an admixture of Spanish words is mentioned or best through Robert Kadarin: Linguists studying [the] language found traces of three Terran languages. Spanish is your casta; English and Gaelic in your cahuenga, and the Dry-Town languages. The language spoken in the Hellers is a form of pure Gaelic which is no longer spoken on Terra. (HH:183).
There are several descriptions of the languages, for example SC:107f mentions different modes of inflection for a word meaning 'brother' (presumably bredu), in the impersonal inflection it refers to a friend or brother when used by a man, in the personal inflection to a brother as a family member whereas in the intimate inflection to a lover (which would make a man using the word appear a homosexual). Likewise, casta seems to possess a special form to address nobility (SC:98) and a feminine mode of address (ibid). In a similar context, a derogatory and respectful mode of address is mentioned in SC:112f and a special inflection for nobility and a feminine mode in SC:98. We will trace what of this can be found in actaul text samples below.
It seems in practice impossible to exactly discriminate samples of casta and samples of cahuenga. In the following, we will make the tentative assumption that the grammar of the languages is not vastly different and separate the vocabulary according to Spanish/Latin phonology (assumed to be casta) and to Celtic phonology (assumed to be an element of the less pure, amalgam-like cahuenga), although cahuenga as a lingua franca would involve many words from casta and casta would have some borrowings from cahuenga. Furthermore, when there is a possible discrepancy in grammar, we will point towards this may be due to the fact that we're looking at two separate languages.
Z'par servu, vai leronis. 'at your service, honoured leronis' (SQ:113)
S'dei par servu. '*by your leave at service' (HH:180)
Su serva, dom. '*I serve you, lord.' (FT:518)
A ves ordres, mestra. '*At your command, madam.' (HM:647)
Su serva, dom, a veis ordenes emprézi. '*I serve you, lord, at your orders forever.' (SE:677)
S'dia shaya, domna. '*You lend grace, milady.' (SQ:177)
Para servirte. '*To serve you.' (HM:683,TH:537,SE:509)
Men dia pre'zhiuro... '*From this day onward...' (TH:284)
Z'bredhyi, chiya. '*I love you, my girl.' (CS:908)
D'sperdo, vai dom alzuo '*[Oh] despair, worthy elevated lord' (BS:230)
Esa so vhalle Terranan acqualle... (a joke, possibly: '?if anyone commands, the Terran follows') (BS:235)
Laszlo, Laszlo, dors di ma main... (children's song of unknown meaning) (BS:467)
bredhiya mea, viyha mea '*my beloved, my life' (CS:1029f)
Adelandeyo. 'Ride with the gods.' (HM:630, FT:282)
In addition, there are few short genitive phrases:
Dhe Shaya 'grace of god' (HM:648)
donas amizu 'gift of friends' (TH:536,WW:341)
penta cori'yo 'bridge society' (CS:767)
nikhya mic 'daughter of' (SC:154)
A variant a ves ordras is known (TTC:208,255)
We know one phrase in the language of the chieri:
narzain-ye kui 'child of the yellow forest' (WW:134)
riyachiya (genetically modified sex slave) pl. riyachiyas (SQ:64f)
#catena '*chain' pl. catenas (TTC:217)
#dona 'gift' pl. donas (TTC:296)
christophoro (member of a Darkovan religion) christophoros (SQ:42,55)
nedestro 'bastard' pl. #nedestros (TTC:174)
chervine 'stag pony' pl. chervines (SQ:57f)
#ordre 'command' pl. ordres (HM:647)
Nouns ending in -i or -ii assume the same form in singular and plural, they do not append any special ending:
kyorebni (carrion-feeding bird) pl. kyorebni (CS:875,897)
com'ii 'friend' pl. com'ii (SD:135,BS:436)
Nouns ending in -u appear to be grammatically male and usually form a plural marked with the ending -in:
laranzu 'psi-worker (m)' laranzu'in (SQ:71,TTC:216)
gre'zu (insult) pl. gre'zuin (SC:19,21)
tenerézu 'keeper (m)' pl. #tenerézu'in(TTC:463)
But note the exceptions:
mestra 'madam' pl. mestra'in (TH:556)
cuere 'year' pl. cueru (WW:337)
Also somewhat deviating from the pattern is the plural of bredu 'sworn brother':
bredu 'sworn brother' pl. bredin (TTC:246,255)
However, this is usually seen denoting a pair of people which are referred to as bredin and might actually rather be an inflection for dual, see below for further support of this argument. Nouns ending in -is appear to be grammatically female and have a plural ending -i:
leronis 'psi-worker (f)' pl. leroni (SQ:15,TTC:213)
comhi-letziis 'renunciate' pl. comhi-letzii (TH:284,381)
tenerésteis 'keeper (f)' pl. #teneréstei (BS:430)
Finally, nouns ending in a consonant sometimes append the plural marker -y:
ombredin 'half-man' pl. ombredin-y (TTC:257,260)
In other cases however, -s is appended:
forst 'stockade' pl. forsts (SC:114,WD:627)
#var (unit of length) pl. vars (SD:90)
breda 'sister' bredu 'brother' (TTC:250,255)
chiya 'little girl' chiyu '*little boy' (SC:141,HM:576)
riyachiya (genetically modified sex slave) (f) ri'chiyu (m) (SQ:63f)
mestra 'missus' messire 'master' (HM:465,680)
leronis 'psi-worker (f)' laranzu 'psi-worker (m) (SQ:15,71)
tenerésteis 'keeper (f)' tenerézu 'keeper (m)' (BS:430,TTC:463)
dom 'lord' g. n. pl. domyn (SQ:202)
Thus, -yn denotes a plural without gender specification, and it would appear that the word comyn 'equal ones in rank' (SE:639) involves just this ending.
However, nouns are also coined from comyn such as comynara (SC:89) or comynaris (CS:1075) 'female member of the comyn' which lead to a gender-neutral plural comynari (BS:501) '*members of the comyn council'.
A similar case might be ombredin 'half-man' (TTC:257) with the meaning 'lover of men' - clearly, this seems to contain bredin, with a prefixed om- to give it a bad sense. However, ombredin in itself is actually a singular word, as exemplified by the plural ombredin-y (TTC:260). This would again point to bredin being perceived as a word in its own right, and not just a plural.
The pattern also carries through in plural, as the forms bredini '*sworn sisters' (TH:526) and the intimate form bredhyini (TH:673) show.
The pattern seems to be that a -h(i)y- is added beween stem and gender ending. If that is true, nikhya 'daughter' (SC:154) might actually be the inflected form of an impersonal *nika denoting a young relative of a daughter's generation.
A second example may be the word kiya (CS:1008) to denote any female relative of the mother generation. kiya is clearly the impersonal form, as it is used by Callista Lanart-Alton to address Leonie Hastur in FT:349. Probably the personal inflection of this word is kiha 'mother, foster-mother' (CS:823) and an intimate inflection *kihyia would not be used across generations.
Strangely enough, Danilo Syrtis is addressed by his foster-father Dyan Ardais as kiyu in SE:432 where we would have expected that it should just be the other way round. The easiest explanation would be that this is a slip by MZB, as it would contradict the whole pattern of endings seen otherwise.
Taking all together, the relevant pattern may be breda (imp.) *bredha or bredhis (pers.) bredhya (intim.); bredu (imp.) *bredhu (pers.) bredhyu (intim.); kiya (imp.) kiha (pers.) *kihyia (intim.) and *nika (imp.) *nikha (pers.) nikhya (intim.).
As the examples vai dom cario (WW:139) and vai dom alzuo (BS:230) indicate, the adjective usually seems to follow the noun. dorilys would then represent an unusual word order in a compound and cleindori a compound preserving he normal order. Comparing cario and dori, one may tentatively identify one class of adjectives with female endings in -i with a male form in -io, thus 'dear flower' might be *lys cari and 'golden lord' *dom dorio.
Maybe carya is then another form of the adjective, or it is rather the form the noun formed from the adjective takes, thus
carya 'dear one (f)' #caryu 'dear one (m)' (SQ:236)
It seems reasonably clear that adjectives in -a as raiva would be a female form. Probably they are also inflected for gender and the corresponding male form would be *raivu. But the gender marking for alzuo or #alte is significantly less obvious.
The honorific vai 'excellent, worthy' (BS:283) may be classified as an adjective. It is indeclinable and seen with male and female nouns in singular and plural without any inflectional change. It seems to be used for the Comyn rather exclusively, presumably a phrase like **vai mestra would be perceived as odd.
vai dom 'worthy lord' (SQ:65)
vai domna 'worthy lady' (SQ:24)
vai leronis 'worthy psi-worker' (SQ:113)
vai dom'yn 'worthy lords and ladies' (HM:576)
A variant with an ellision is attested as well (presumably for euphony): va' altezu 'your highness' (TTC:348)
A different variant vahi is attested in SC:19, but this is presumably intended to show the wrong use of casta by a Dry-Towner who doesn't speak the language very well.
The word tenerézu 'keeper (m)' (TTC:463) seems to involve a verb #tener 'to keep' (Spanish tener 'to hold', Latin tenere) with a male ending -zu, the connecting -e- may then be the remnant of an original infinitive form *tenere as in Latin. In a somewhat similar line of reasoning, servirte may or may not indicate a verb *servir '*to serve'.
Just like in Latin or Spanish, this points towards the existence of several different verb classes which are distinguished by their last vowel.
For reasons to be given below, we identify a few verb forms in the corpus of phrases. We take dia '*you give' for a verb *dar, dei '*you permit' for a verb *deir, serva '*I serve' for a verb *servar and bredhyi '*I love' for *bredhyir. The pattern is a bit hard to spot, but is at least consistent with the 1st person sg. being the verb without infinitive marker and the 2nd person sg. with an additional -i- inserted. Thus, *tene may be '*I keep' and tenie '*you keep'.
An imperative may be attested in latti! 'leave [it]' (BS:275), the corresponding verb may be *lattir 'to leave alone'. Tentatively, one may deduce that the imperative involves the verb form without the infinitive marker -r and is identical to the 1st person sg., although this may not be so in all verb classes (it is not so e.g. in Latin). Thus, one may find *tene! 'keep!' or *servi! 'serve!'.
A future tense is attested in the Hastur motto permanedál 'I shall remain' (HH:20) and its variant permanedó 'here we remain' (WW:177). The verb is evidently based on Latin permanere 'to remain' and would therefore probably be *permaner 'to remain'. A further comparison with Latin verb structure suggests that -dá- is a future suffix and -l a 1st person sg. pronominal ending and á > ó is the way 1st person plural is marked in future tense. This ending is not seen for 1st person sg. verbs in present tense, but neither is 1st person sg. marked in Latin the same in all tenses. If this holds, one might perhaps form *tenedál 'I shall keep' or *lattidó 'we shall leave alone'.
Another possessive ves is found in ves ordres 'your orders' (HM:647). This can be identified with Latin vestra and would a 2nd person plural, here possibly used for politeness. The form indicates that possessives are also inflected for plural, 'your order' might then be *ve ordre where ve < vee for gender agreement, and 'my orders' possibly render as *mes ordres.
A different pronoun su is seen in su serva (FT:518). For reasons to be outlined below, this should probably not be taken as a possessive pronoun but rather as an object '[for] you' which would appear in various phrases as abbreviated s' or z'. The corresponding possessive form might be *sua. Also below, we will suggest the possibility that there might be forms *se 'you (m)' and *si 'you (f)'.
a occurs in a ves ordras (HM:647) and can readily be identified with Spanish a 'at'. This may also be the first element of the verb accandir 'to lie together' (FT:472), comparison with Latin verbs of similar shape would then suggest #candir 'to lie'.
di occurs chiefly in Names such as 'di Asturien' (SE:435) or 'di Scarp' (SC:115), but also in the phrase di catenas (TTC:475) which denotes a formal marriage. It would seem that it is used to indicate the genitive 'of', thus di catenas would be part of a longer phrase originally meaning something like '*[joining] of chains'. A variant is attested in the name 'des Trailles' - maybe for euphony this is the form used for a plural in -es - as di catenas proves the preposition is not in general different in plural. Thus, we might have *des ordres rather than ?di ordres as translation of 'of orders'.
mic is only attested in the long form of n'ha, i.e. nikhya mic (SC:154) and this must mean 'daughter of'. Thus, mic translates in all likelihood 'of' rather than 'daughter' (note that nikhya is consistent with being a female noun ending in -a). Maybe the difference to di is that mic should literally be translated 'out of' in the sense that a baby comes 'out of' the womb.
par, para is attested in para servirte (HM:683) and z'par servu (SQ:113) and can readily be identified with Spanish para 'for, in order to'.
#per is part of the verb permanedál 'I shall remain' (HH:20). Based on Latin manere 'to stay' and Latin per 'through', one can with some confidence decompose the casta verb as #per '*through' and #maner '*to stay', the Hastur motto then literally means 'I shall stay [it] through'.
#tra does not occur independently, but is seen in the name 'Tramontata' which can be decomposed into #tra and #montana. The last word agrees readily with Spanish montaña 'mountain' and the first part can be related to Latin trans 'across', thus the name literally means 'across the mountain[s]'.
An example for a proper compound may be cleindori 'golden bell' (BS:197). As a comparison with dorilys 'golden flower' (ibid) indicates, #dori is the element for 'golden' (dorado is 'golden' in Spanish), thus #clein must mean 'bell'. It is not quite clear if #lys can then be translated as 'flower' however.
The other case is presumably illustrated by the words comyn (Darkovan nobility) (TTC:355), com'ii 'friend' (SD:135), comhi 'guild' (TH:381). -yn is transparently a plural marker appended to a stem com '*together', but it would be very bold to assume one could extract **yn 'people' as an independent word. 'ii seems to denote a person, and it is suggestive to connect this with the female -iis seen in letziis 'renunciate (f)' (ibid), so one might derive *com'iis 'friend (f)' and *letzii 'renunciate (m)' from a stem #letz- 'to renounce'. The ending -hi would then perhaps denote an organization, thus com-hi '*together-organization, guild'. It is not quite clear if ?hi could stand alone as a word, but probably not.
It is interesting to compare comhi 'guild' (ibid) with cori'yo 'society' (CS:767). The stem #cori apparently denotes a localized social entity (in contrast to the guild which has houses in every major place in Darkover) and is in fact long enough to be tentatively identified with a word #cori '*estate'. -yo would then be an abstract suffix. The element #cori occurs also in cori-dom '*estate-lord, estate manager' (TTC:289).
Hali'imyn (HM:453), used by the mountain-folk to denote inhabitants of all the lowlands, but literally probably rather 'Hali-people' appears to be a proper compound. The second element #imyn shows a gender-neutral plural -yn. The remaining stem would then mean 'someone, a person'. Such an element may in fact be seen in the old name El Haleine (TTC:273) where the second word would involve Hali. Thus, Haleine < *Hali-in and #imyn < *inyn by dissimilation. If so, one could deduce *in 'someone, a person'. In all likelihood, one might also form *Thendara'imyn 'the Thendara people', although other such forms are in fact not seen.
In general, there are not many sound shifts seen in compounding, but there are a few interesting exceptions. leronis 'psi-worker (f), sorceress' (SQ:15) is evidently derived from laran 'psi-power', but shows internal i-affection which shifts the first a > e and the second a > o. Such internal i-affection is frequent in Celtic languages. However, the formation pattern does not seem to be active in modern casta any more, as the example terranis 'Terran (f)' (CS:1075) indicates which does not come out as **terronis.
gre'zu and its plural gre'zuin appear in SC:19,21 as apparently insults with a sexual connotation. Given that grezalis is 'whore, prostitute' (SC:107) one might take gre'zu as the male form and guess that it denotes a male prostitute.
bre'sui is likewise an insult with some hint of dirtiness. A variant 'little bre'suin' is found in SD:111, but this variant may be driven by the suffix *-in 'person' described above. The insult suggests a small, dirty, sneaky animal, quite possible something like a rat.
cagavrezu is used in HM:541 for Rakhal Hastur who punished a man cheering for his opponent Carolin severely. The last part is the suffix -zu, this leaves #cagavre. Since cagada is 'shit' in Spanish, this should probably be decomposed as caga-vre-zu. The second element might be related to bre'sui above - if so, the whole word would literally mean something like '*shit-rat-guy'.
viyha is attested as an endearment in the phrase viyha mea (CS:1029f). The closeness to Spanish vida 'life' suggests that this is the meaning of the word, and this also fits the context very well.
lemvirizi us used in connection with 'filthy' as an insult to the renunciates in (TH:546). Presumably the singular form is #lemvirizis. Presumably, it refers to the perceived indecency of independent women and must mean something along the lines of '*slut'.
s'dia shaya is translated in BS:217 as 'you lend me grace'. shaya can clearly be identified as 'grace' by comparing with Dhe Shaya 'grace of god' (HM:648) where it is the common element. This leaves s'dia as somehow translating 'you lend me'. A comparison with su serva (FT:518) and the possessive pronoun mea 'my' and would suggest that s' is an abbreviated form of perhaps su here and translates 'you' rather than 'me' in the phrase. dia finally may be an inflected for of the verb 'to give' (dare in Latin, dar in Spanish, maybe also *dar in casta). Thus, dia would translate '[you] give' (allowing for a pronominal verb inflection as in Spanish) and 'me' could be omitted alltogether, thus the literal meaning of the phrase would be '*you give grace'.
z'par servu is translated 'at your service' (WD:548). servu can most easily be rendered as 'service' and par seems to be a short form of para seen in para servirte (HM:683) and presumably corresponds to Spanish para 'for, in order to'. z' is probably just a variant form of s' seen above and hence is an abbreviated pronoun '[for] you'. Literally, the phrase would thus be '*[for] you for service'.
s'dei par servu (HH:180) is untranslated. The last part would be '*for service' (see above). s'dei reminds structurally of s'dia which we have translated as 'you give' above, thus we're probably looking for a verb. Spanish dejar 'to leave, to allow' suggests itself which would need to take the form *deir in casta. If so, the phrase probably renders literally '*you permit [me] for service'. This has certainly the ironic ring which befits the scene in which Robert Kadarin introduces himself.
su serva tempts the identification of serva as 'servant' (serva is the Latin word for a female slave), so the phrase would be translated as 'your servant'. However, there are a number of reasons which argue against it. The typical male-female ending pattern would suggest that serva is a female noun with a male counterpart servu. However, su serva is attested to be spoken by men to men (BS:236) and z'par servu is translated as 'at your service' (WD:548), thus servu might rather be an abstractum 'service' similar to amizu '*friendship' than a male servant. The usage of other known possessive pronouns, e.g. mea in e.g. carya mea (SQ:258) or breda mea (CS:933) suggests that possessives show gender agreement. This is not in contradiction to ves ordres '*your orders' (HM:647), so this would suggest that we ought to find ?serva sua if serva would be translated as 'servant'. The alternative would be that *servar is a verb 'to serve' and su serva renders '*I serve you'. If su stands for 'you' as in the previous phrase, casta must be a language in which pronouns are expressed by the verbal ending (in fact, just like Latin or Spanish), otherwiese the phrase could only mean 'you serve'. In this case, serva alone must be '*I serve' and su serva literally '*you I serve'.
z'bredhyi, chiya. (CS:908) quite evidently means '*I love you, my girl.' and is replied to as such. chiya is attested elsewhere as 'little girl' (SC:141) but can be used as an endearment. Thus, z'bredhyi must translate '*I love you'. As above, taking z' as 'you' leaves bredhyi as '*I love' with the pronoun 'I' expressed in the verb form as assumed above for serva. The 1st person in both cases is consistent with being the unmodified verbal stem assuming *bredhyir 'to love'. The alternative hypothesis would be to assume that z' translates 'I' and 'you' is omitted. z'par servu must then be translated as '?I [am] for service'. However, this does not only contradict the attested translation but also is unsatisfactory here - note that in English 'love you...' can be used readily as abbreviation for 'I love you', but not 'I love.'. Thus, most probably the phrase is literally '*you I love'.
adelandeyo is given the translation 'ride on with the gods' in HM:630 and 'ride with the gods' in FT:282. It is clear that the phrase is a contracted sentence (in the same way as 'hocuspocus' is a contraction of 'hoc est corpus meum'), but it is surprisingly easy to identify the elements. The verb 'to ride' appears to be the Spanish adelantar 'to move fast'. Thus we may try the whole phrase in Spanish. For the preposition 'with' me may guess Spanish con and 'gods' is dios, hence we'd find adelanta con [los] dios. But we know that 'god' is rather dhe in casta - should we assume an irregular plural deyo or rather put it down to a variation in dialect - after all, the same devocalization and softening of the consonants seen in Esteban > Istvan could explain deyo > dhe. We have to leave the question open, but suggest here that the phrase is a contraction of *adelanta con deyo 'ride with [the] god[s]'.
a ves ordres (HM:647) is readily translated given the work above. We have linked a with the Spanish preposition 'at' and ves with Latin vestras 'your (pl.)' which leaves ordres. This must mean '*orders' or in plural '*command', thus the whole phrase would be '*at your orders' and this indeed fits the context well.
a veis ordenes emprézi. '*at your orders forever.' (SE:677) is evidently a variant of the above. veis seems to be a variant of ves, the difference is probably not significant. orden is the Spanish word for 'order', thus #ordene would be '*order' in casta. The context of the phrase allows to translate the last element as '*forever, now and forever'.
men dia pre'zhiuro... (TH:284) is said to translate the first word of the oath of the renunciates. These would be 'from this day henceforth...', and indeed that seems to fit the meaning of the phrase as it is used in various other occasions. The only element which can readily be identified is dia due to its similarity to Spanish día 'day'. pre'zhiuro seems to involve the element #pre which can be linked with Latin prae 'before, in front', thus one might tentatively translate pre'zhiuro as something like 'forward', so it would translate the 'henceforth'. The remaining men would then translate either 'from' or 'this' (in principle it could do both, but it seems a little short for this combined function). As 'this day henceforth' seems more complete as to the amount of information than 'from a day henceforth', it would seem that men is the demonstrative 'this' rather than a preposition.
donas amizu is translated as 'gift of friends' in WW:341, but this is puzzling as a literal translation - we'd expect based on patterns seen elsewhere that donas is a plural and amizu a singular. Maybe the phrase should literally rather be '*gifts of friendship' which fits more easily into the attested patterns.
Leaving the realm of what can be translated with some certainty, let us turn to a few more speculative examples:
d'sperdo, vai dom alzuo (BS:130) is untranslated and uttered by a bar woman who is utterly devastated by the words of Jeff Kerwin. d'sperdo suggests Latin desperare 'to despair', but it doesn't fit the pattern of identified verbs, so let's take it for a noun meaning 'despair, grief'. vai dom clearly is an address to Jeff Kerwin, 'worthy lord'. alzuo is maybe close to Spanish alzada 'height' or alzadura 'elevation', so just possibly it is a qualifying addition to vai dom. If so, it would mean '*high, great, elevated'. Thus, the whole phrase might translate '*[Oh] despair, worthy elevated lord'.
para servirte (HM:683) tempts to translate as '?for service'. But the longer variants para servirte, vai dom (SE:509) and para servirti, vai domna (SE:747) are attested and suggest that servirte/servirti is distinguished by a gender-marking inflection and can't be a gender-neutral 'service'. It could still mean 'servant', but another idea seems more likely: It could be a compound of the verb #servir 'to serve' and a gender-specific pronoun. This would yield ?te 'you (m)' and ?ti 'you (f)' and introduce a third 2nd person consonant next to s- and v-. But in fact the forms could well be *se 'you (m)' and *si 'you (f)' with assimilation servirte < *servirse. Maybe then su is a gender-neutral 2nd person form - su serva is attested spoken by a man to a man (BS:236), by a woman to a man (FT:518) and by a man to a woman (SE:690) and never changes form. Or the phrase must be a possessive '?your servant' after all with no gender-marking for noun or possessive.
narzain-ye kui (WW:134) is chieri and means 'child of the yellow forest'. Quite likely, the first word narzain means 'yellow forest' as it carries the inflection -ye which would then express genitive. Further support comes from the fact that kui seems to short to express a compound idea like 'yellow forest' but well suited to express a simple word like 'child'.
Esa so vhalle Terranan acqualle... (BS:235) is spoken by a chuckling Darkovan after a communicator announcement leads to an exodus of Terrans out of a bar. Quite likely the joke is about the lack of independence of the Terrans. terranan is the only known element - it means 'Terran'. The last word could be related to Latin adsequor 'to follow', thus Terranan acqualle '*the Terran follows' or '*the Terrans follow'. If so, vhalle (with a very similar ending) must also be a verb in 3rd person. The whole phrase might therefore mean something like '*if anyone commands, the Terran follows', but admittedly this is little more than guessing.
Case relationships are expressed by prepositions or context. In particular, genitive relationships are more often than not expressed by loose compounding rather than the preposition di, cf. penta cori'yo 'bridge-society' (CS:767) rather than **cori'yo di penta 'society of the bridge' or dhe shaya '*god-grace' (HM:648) rather than **shaya di dhe 'grace of god'. Accusative objects are apparently marked by word order or other context only.
Verbs are inflected for at least person and tense. In all likelihood, they are also inflected for politeness, but we just don't know enough of the language to tell how this is done.
In general, the language relies a lot on context - omission of pronouns which are understood is found frequently in phrases. The evidence suggests that also pronounciation has a large role (cf. e.g. SC:107f), but again, or knowledge of details is simply insufficient.
Latti firi, nikhillu! 'Hands off he firi, little foster-son.'
Men dia adelantadál con teniente me tra montanas par Armida. 'On this day I shall ride with my deputy through the mountains for Armida.'
V'letzi y catenas di Arilinn. 'I renounce you and the chains of Arilinn.'
S'tenie kirian a Neskaya. M'da par leroni meas! 'You keep kirian in Neskaya. Give it to me for my leroni!
Pre'zhiuro tenedál ordres des cortes. 'Henceforth I shall keep the commands of the council.'
Dei me maner con su, bredhya mea. 'Permit me to remain with you, my beloved.'
To Parma Tyelpelassiva