Let's take a look.
First the references (and if you're interested in Quenya, this is a great site), I get everything from Ardalambion and this dictionary.
Your first word "Caitan" seems correct. "Caita-" is the aorist form of "lie" and the "-n" is the pronominal suffix for "I".
Here's the definition of "Caita-":
caita- vb. "lie" (= lie down, not "tell something untrue"), aorist
tense "lies" in the sentences sindanóriello caita mornië "out of a
grey land darkness lies" (Nam, RGEO:67)
And here is the construction:
We also see -i- before all pronominal endings; indeed
Tolkien very often cites Quenya primary verbs as aorist forms with the
ending -n "I" attached (e.g. carin "I make", LR:362, tulin "I come",
LR:395). A-stem verbs show no variation, but end in -a whether or not
any further ending follows (e.g. lanta "falls", lantar "fall" with a
plural subject, lantan "I fall", etc.)
Next word "nu" is also correct.
nu prep. "under" (LR:56, Markirya, Nam, RGEO: 66, MC:214; the
Etymologies alone gives no [q.v.] instead). In Mar-nu-Falmar,
nuhuinenna, q.v. Prefix nú- in nútil, q.v
So then we have "i" which is Quenya for "the". I don't think you need the hyphen here. The hyphen is usually used when a compound word is created, and that is not the case. (For example Mar-nu-Falmar, "Land (lit. Home) under Waves" is a compound word, a name for Númenor after it sinks)
"isilmë" is moonlight, so that's correct!
isilmë (þ) noun "moonlight", occurring in Markirya; free translation
"the moon" in MC:215 (isilmë ilcalassë, literally "moonlight
gleaming-in" = "in the moon gleaming"). Isilmë also appears as the
name of a Númenorean woman (UT:210).
So far so good! Let's look at the next line.
"Ápan" is not quite right. It should be Appan "I touch" with Appa- being the aorist form and the "-n" the pronominal suffix. (I don't see any reference with the accent over the a.)
appa- vb. "touch" (in the literal sense; contrast #ap-, q.v.) (VT44:26)
"Talan" is ground.
talan (talam-, e.g. pl. talami) noun "floor, base, ground" (TALAM)
This next one is a little tricky, but I think you got it. "rosséva" is "of dew", with "rossé" being "dew" and the suffix "-va" meaning "of". I am not sure of the accent. It looks right, but I couldn't find the specific rule.
Here's the definition:
rossë noun "fine rain, dew" (ROS1, PM:371)
And the construction:
Then there is the possessive, by some called the "associative" or
"adjectival case"; Tolkien himself speaks of it as a
"possessive-adjectival...genitive" in WJ:369. This case has the ending
-va (-wa on nouns ending in a consonant). Its general function is like the English genitive, to express ownership: Mindon Eldaliéva "Tower of
the Eldalië". The function of the possessive was long poorly
understood. In Namárië it occurs in the phrase yuldar...miruvóreva,
And finally, line three.
"Ólan" is also correct. "Óla-" is the verb dream and again "-n" is the pronominal suffix.
óla- vb. "to dream" (said to be "impersonal", probably meaning that
the dreamer is mentioned in the dative rather than the nominative)
The last word is very difficult, and I think you nailed it. "ranculyat" breaks down like so: "rancu-lya-t", rancu-=arm, lya=your and t=dual plural. Quenya has a special plural for natural pairs, like arms or eyes, and you got that suffix correct. Also, the "a" in "lya" does not get a stress because the dual plural suffix is "-t". (The first "u" may need to be stressed, but I couldn't find a reference for it.)
So...here's the definition:
ranco ("k") noun "arm", stem *rancu- given the primitive form ¤ranku,
hence also pl. ranqui ("q") (RAK)
and the construction:
2nd person sg. formal/polite: -lya "your, thy"
The plural suffix of a pair:
Like the nominative plural, the nominative dual is formed with one of
two endings. Most nouns take the ending -t, as in the word máryat "her
hands" (two hands, a pair of hands) in Namárië. "Two ships, a couple
of ships" is likewise ciryat (cirya "ship")
And finally, the stress:
When stressed, most of these pronouns have long vowels: ní, lyé, tyé,
sé, sá etc. The vowel however remains short in the dual forms in -t
So, the "a" in "-lya" stays short!
My translation would be very similar to yours:
Caitan nu i isilmë
Appan talan rosséva
You have done amazingly well, as far as I'm concerned. I think you'd like this course from Ardalambion...it's where I learned everything I know.
Also, it would be great to see it in Tengwar!
Hope this is helpful!