The Latin word for "report" is refero, referre, rettuli, relatus. At first, I thought that rettuli, with 2 t's, was a typo, but it actually appears that retuli, with just 1 t, is the uncommon form in comparison.
This seems rather strange to me, because most of the compounds of fero do not seem to double the t. For example:

defero, deferre, detuli, delatus
effero, efferre, extuli, elatus
offero, offerre, obtuli, oblatus
infero, inferre, intuli, illatus

Why is refero an exception? Additionally, what instances are there of Latin writers using the more uncommon form of retuli instead of rettuli?

  • 1
    As far as I know, it doesn't come from tuli but from re(d)-. Compare with redire. I hope someone can give more detail and explain why the D was lost in the other forms. – Joonas Ilmavirta Mar 1 '17 at 21:10
  • 6
    Sihler's New comparative grammar of Greek and Latin (§523) suggests that it's actually because of the loss of vowel from what was originally a reduplicated perfect form (*retetuli > rettuli; cf. *repeperi > repperi). – cnread Mar 1 '17 at 21:18
  • Red- assimilated in some other verbs as well; compare relligio, reccido. Don't know why this happened only sometimes though. – Draconis Mar 1 '17 at 22:50
  • Cf. reddo. <filler text> – Cerberus Mar 1 '17 at 23:07
  • 1
    related latin.stackexchange.com/a/618/39 – Alex B. Mar 3 '17 at 5:32

As @cnread's comment indicates, the geminate t of rettuli is thought to be a remnant of Indo-European perfect reduplication.

The Proto-Indo-European perfect tense was formed with reduplication of the first consonant of the root; a few Latin perfects still do this, e.g. cecidi, peperi, cucurri. Rettuli would then come from an older form re-tetuli; the short e, which would have been unaccented in early Latin (in which accent was on the initial syllable), was lost by syncope and the two t's fell together. The same explanation accounts for other such forms, e.g. reccidi, repperi. (The geminate d of the present reddo is probably due to the same phenomenon, except that in this case the reduplication was a present-stem formant rather than a perfect marker, cf. Greek δί-δωμι.)

  • 2
    This makes sense, but why does it only happen in the prefixed form? We don't see *tutulī, for example, even though the initial stress would have preserved the first vowel. – Draconis Mar 2 '17 at 0:09
  • @Draconis, that's quite true -- most perfects that were reduplicated in PIE lost their reduplication in Latin. I don't know of any good explanation for why the handful of verbs that retained it did so (which is often the case with morphological change). – TKR Mar 2 '17 at 0:53
  • Interesting answer. There are a few things I still don't understand: 1. why no dettuli, praettuli, prottuli? Does it have something to do with vowel length? 2. Why do you think reddo is reduplication, too, rather than the -d that is often part of the prefix re(d)-? – Cerberus Mar 2 '17 at 15:04
  • The Latin perfect continues both the IE perfect and the IE aorist. That is why it has lots of different forms. – fdb Mar 2 '17 at 15:07
  • 1
    @Cerberus, 1 is hard to answer because (so to speak) regularization is irregular -- in the course of language change some forms get regularized and others don't; that said, it's possible vowel length was relevant -- Latin does have something of an aversion to superlong syllables (long vowel + coda consonant), though not an absolute one. (Note that the red- account runs into the same problem -- why red- in this form but re- normally?) 2. This verb shows reduplication in the present in other IE languages; and red- is a prevocalic alternant of re-, like prod- for pro-. – TKR Mar 2 '17 at 22:43

The prefix re- also appears as red-, an older form which mostly disappeared along with other final -ds (personal med, preposition extrad, imperative estod, ablative marid...). You can see this d in forms like red-īre and red-dere, and remnants of it in religiō/relligiō and redūcō/rēdūcō (assimilation and compensatory lengthening).

While most -ds disappeared, some words (like your rettulī) kept their older forms...sometimes. As L&S put it, "[t]he orthography and quantity of words compounded with re are in general somewhat arbitrary, especially in the ante- and post-class[ical] poets". So while rettulī was the most common form, retulī and rētulī were by no means unheard of especially in Pliny, and Lucretius also used forms like rellatus/rēlatus (showing the -d in the fourth stem).

  • 4
    AFAIK, red- is only prevocalic; rettuli, reccidi, reddo all reflect original reduplicated forms with vowel loss. Not sure about relligio, but it was explained as metrical lengthening already in antiquity (see Servius Honoratus quote here). De Vaan thinks the d of red- is not inherited but secondary, based on an analogy like pro(d)-. For rellatus/rēlatus, I suppose they may be etymologically explicable, from re-tlatus – TKR Mar 1 '17 at 23:25

A brief comment re: geminate vs. non-geminate perfect forms of refero.

enter image description here

enter image description here

  • 1
    Where are you getting this data from? You probably have access to more sophisticated tools, but a naive corpus search shows 628 hits for #rettul and 148 hits for #retul, which is certainly less conclusive than this chart. – brianpck Mar 3 '17 at 22:49
  • @brianpck the corpus is called Opera Latina (University of Liege), web.philo.ulg.ac.be/lasla It lets you search for specific morphological forms, so I got data for perfectum only. – Alex B. Mar 3 '17 at 23:15
  • 1
    The discrepancy between these numbers and those from PHI (@brianpck) is strange -- I wonder what accounts for it. – TKR Mar 4 '17 at 5:31
  • @TKR I just searched another POS tagged corpus (LatinISE historical corpus v2), and I got this: Query rettuli.*, Romana Antiqua, Romana Classica 163 (12.54 per million); Query retuli.*, Romana Antiqua, Romana Classica 12 (0.92 per million). – Alex B. Mar 5 '17 at 4:16
  • @brianpck I just looked more closely at your data. 47 occurrences of retul# (out of 148) come from the 4-5 century AD grammarian Maurus Servius Honoratus. – Alex B. Mar 5 '17 at 4:59

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.