Since the issue has recently come up both in a private correspondence and on this list I looked up my old notes concerning that mysterious Romance root for 'small, little'. The most likely story goes something like this:
In Latin there was a word PŪTUS, meaning '(small) boy'. In Vulgar Latin this word had a variant *PŬTTUS, since there was a general tendency for V:C and VC: (long vowel + single consonant and short vowel + long/double consonant) to alternate with each other, and then by analogy or dialect mixture a third variant *PŪTTUS with both the stressed vowel and the following consonant long1.
First of all the well-known Latin diminutive ending -UL- was added to *PUTTUS, giving *PUTTULUS, which soon was contracted to *PUTTLUS. Now there was a very strong tendency in Vulgar Latin for an /l/ which formed a cluster with another consonant to become a palatal [ʎ] (like the older pronunciation of Spanish ll or Italian gli). As we see from the various reflexes of PLANTA in the Romance languges — It. pianta, Sp. llanta, Fr. plante, Ptg. chanta — the consonant + [ʎ] could change in various ways: it could change into consonant + /j/ (Italian) which could in turn change into /tj/ and further into various reflexes of that combination, or indeed revert to or remain consonant + l (French).2.
Next the palatal(ized) combination [cʎ] changed the preceding vowel /ʊ/ into /ɪ/. *PŬTTULUS had now become something which may be written *PĬTTJLJUS. Now it is a well known fact that the combination *TL, or rather its palatalized outcome *TJLJ [cʎ] was felt to be especially difficult to VL speakers: it usually changed to — or was indistinguishable from — CL > *KJLJ 3, and Vulgar Latin speakers of central Italy, who tended to preserve or restore the -ULUS ending, at some point retrofitted *PITTJLJUS to *PICCULUS, whence Italian piccolo.
Since the diminutive ending -UL- lost its meaning or vanished by sound change, and because people still felt the need to reinforce the sense of littleness, the thus obtained root variants *PITL/*PITTJ/*PIKKJ/*PIKKL and even the further back-formations *PITT and *PIKK, all meaning 'small' or 'small thing/being' could and usually were refurbished with further diminutive endings, giving the French variants and derivations petit, petiot, pichot, pichoun, and Sard pizeɖɖu < *PITJELLU, but above all *PITTJĪNUS, and in Iberia *PIKKINNUS 4, although Rumanian also has the suffixless back-formation pitu.5
The different reflexes show a wide variation in the sounds corresponding to the final consonant of the root: Sard pizinnus [ts], Italian dialects pizzinnu [tts] and piccinnu [ttʃ], French petit [tt] 6, pichot [tʃ] > [ʃ], Sp. pequeño, Ptg. pequeno [kk], It. piccolo, no doubt because this sound could be palatalized or depalatalized, and then either to a dental/alveolar or a velar, and lastly it could even remain or be restored to a cluster. Last but not least these palatalizations and back- formations happened at different times and places, and therefore gave different results. The absence of [ʎ] reflexes no doubt is because the consonant was usually long or double. It has also been thought that the frequent use of diminutives in talk by and to babies has led to sometimes anomalous sound developments. For this word if anyone one is prepared to believe that!
Benct Philip Jonsson bpj@melroch.se Wed, 27 Aug 2008
The most well-known example of this is the latin word for 'whole' TŌTUM. Spanish todo is a regular reflex of the Classical form with V:C while French tout goes back to *TOTTUM while Italian tutto goes back to a mixed form *TŪTTUM, which was due to the fact that 'long' Ō [o] and 'short' Ŭ [ʊ] interchanged — and indeed usually merged — in Vulgar Latin. Another example is the VL diminutive ending *-ITTUM/*-ĪTTUM, where French - et and Italian -etto must go back to the form with a short vowel while Spanish -ito must go back to a form with both a long vowel and a long consonant. Logically a form *PŪTUM with a long vowel and a short consonant must have existed, but I don't know if it is anywhere attested; it is notable that this form would have coincided with a common variant of the past participle ending! [back]
The reflex [ʎ] (which could develop on to [j] and various further devlopments of that sound, like Spanish [x]) is especially common between two vowels, as seen in the reflexes of SPEC(U)LUM: Portuguese espelho, Catalan espill, Sard ispiyu, Spanish espejo, though forms like Italian specchio and Frulan spieili also exist. [back]
Whence *VET(U)LUS > VECLUS (attested in "Appendix Probi") > It. vecchio, Fr. vieil, though Provençal espitlori, if indeed from *SPEC(U)LORIUM, forms an interesting exception and evidence that the development could go both back and forth! [back]
The suffix variants *-ĪNUS -and *-INNUS are of course a further instance of V:C/VC: variation! [back]
It should be noted that the Rumanian word puţin is an regular reflex of PŪTĪNUS, remarkable only in that it must be an early loan from another Romance variety where -ĪNUS remained productive. Those things happen too! [back]
If the Vulgar Latin diminutive ending *-ITTUS/*-ĪTTUS is itself developed out of -IC(U)LUS/ *-ICC(U)LUS the word petit shows the same type of development double, and also illustrates that at some times and places *TL and *CL indeed merged as [cʎ]. [back]