I have always liked clitics, these curiously-named morphemes that appear to live at the fringes of our usual linguistic classifications. Takes on clitics range from adopting them as linguistic primitives to considering them entirely epiphenomenal.
My direct experience working on clitics is limited, but I have encountered them throughout my whole life as a linguist: they’re not only one of the big messy parts of the grammatical and orthographic norms of my native Catalan language, but they’re also a pain in the neck when you try to figure out Medieval French sentence structure, or the alignment split in an Amazonian language.
As a note on notation: I will use the traditional = symbol to indicate that an element is a clitic.
Clitics fall into roughly two big groupings, largely based on the approach taken: those that operate at the level of sentences, and those that operate at the level of verbs. The first type of clitics were classically discussed by Jacob Wackernagel (1892), among others. These clitics are a group of words in Indo-european languages that have the characteristic of appearing in the second position of a sentence, before the first phrase or stressed phrase. These clitics include pronominal elements, but also adverbs or auxiliaries, like in
(1) Polees = te = min eresanto hippees phoreein.
many = and = it prayed riders carry
‘And many riders prayed to carry it.’ (Ancient Greek)
where “te” is the second-position clitic. Wackernagel went on to propose an interesting morphophonological explanation of the verb-second phenomenon (which means that in languages like German or Dutch the inflected verb may only appear as the second element in the sentence) that is connected with his observations on second-position clitics, discussed extensively in Anderson (1993).
The other type of clitics is what we would call pronominal clitics, or verb-level clitics. These include not only proclitics (which “lean forward”) but also enclitics (which “lean backwards”), and the term is usually restricted to elements that have a pronominal nature, as opposed to the open class of Wackernagel’s clitics:
(2) No la = trobo.
no she/it = find
‘I’m not finding her/it.’
(3) Va mirar = los.
aux look = them
‘He/she looked at them.’ (Examples of an enclitic (2) and a proclitic (3) in Catalan.)
Pronominal clitics are usually classified as weak pronouns. Pronouns because they substitute (or double) nouns, pronouns or other phrases, and weak because their distribution is limited to a position next to some specific element in the sentence, usually a verb, rather than appearing in the places where other strong pronouns show up (4), like a free position in the sentence or in a prepositional phrase (5):
(4) Puc anar amb ells.
can go with them
‘I can go with them.’
(5) *Puc anar amb = els/los.
can go with = them(CL)
‘I can go with them’. (Examples of Catalan.)
It’s widely accepted that inflected verbs attract clitics because of their intonational strength. Clitics are then pronouns that can’t be stressed, and their need to lean against something that’s solidly stressed leads them to show up next to that element, typically a main verb or auxiliary in the sentence. This idealized story is usually not as pristine when it comes down to looking at the way clitics actually behave, also in Romance and Slavic languages, where the model of pronominal clitics comes from. We can now understand why something like stressed clitics sounds like a paradox. But, of course, they exist, and they’re pretty cool. A few of those exotic clitics are showcased in the rest of this post.
As a good Romance language, Catalan presents pronominal clitics in all of its varieties. There is, however, the Catalan spoken in the Balearic and Pityusic islands, as well as the city of Alghero in Sardinia. In these varietes, elements that otherwise look just like your usual clitics are the syllables on which the stress falls:
(6) Vull fer = hó.
want do = CL
‘I want to do it.’
(7) Donau = la = mè.
give = she/it = me
‘Give it to me.’
In Pityusic Catalan the phenomenon is slightly more limited, but this description is still true of all insular varieties of Catalan. In all of them, stressed clitics appear with imperatives and infinitives, and the phenomenon is restricted to enclitics. If there are two clitics, stress falls on the rightmost one (7). Unsurprisingly, if the clitic is reduced to a vowel-less form, the stress falls on the last syllable of the verb:
(8) Mirà = l ràpid.
look = him/it quick
‘Look at him/it quickly.’
In the Spanish spoken in Argentina and Uruguay, enclitics are stressed when they appear with imperatives or infinitives.
(9) Mira = ló rápido.
look = him/it quick
‘Look at him/it quickly.’
The phenomenon seems quite similar to what we find in insular Catalan. Once again, what is interesting is that this happens in languages that don’t assign stress to the last syllable in a word or constituent.
Gascon is a dialect of Occitan spoken mostly in the mountaneous regions of Gascony and Bearn in Southern France. It also has elements that look like empowered weak pronouns:
(10) Porto = la = mú.
bring = she/it = me
‘Bring her/it to me.’
So far, it seems that only enclitics can behave like stressed clitics. This is also the case in Campidanese Sardinian (Repetti and Kim 2013), spoken in the island of Sardinia.
(11) Nara = mí.
tell = me
Other varieties of Sardinian, a Romance language closely related to Italian and Corsican, present a phenomenon that’s not unrelated to that of stressed clitics: penultimate stress, in which the second-to-last syllable in a word carries the intonational stress. However, in the Campiadnese variety we also find weak pronominal elements that are capable of getting the stress.
Romance languages are not the only ones where we see rebellious clitics. In Slovenian (Franks 2000) we also find this type of not-so-weak clitics.
(12) Predvčerajšnjim še nè, včeraj pa sem = gà.
day-before-yesterday still yesterday but aux it
‘The day before yesterday I didn’t, but yesterday I did (finish it).’
In all of the cases above, the intuition is that these clitics are slightly closer to being affixes than standard non-stressed clitics, and somehow form a unit together with their verbal host in which accent falls on the last syllable. Oddly enough, this is the case even when said accent doesn’t get moved around when those same verbs appear without a “strong clitic”.
Of course, various theoretical analyses have been proposed to account for some of these rebellious clitics. But they remain really interesting particles that challenge our usual take on pronominal clitics and the division between word-level phenomena and pure morphological processes.
Anderson, Stephen R. 1993. Wackernagel’s revenge: clitics, morphology, and the syntax of second position. Language 69:68-98.
Franks, Steve. Clitics at the Interface. 2000, In Clitic Systems in European Languages, d. by F. Beukema & M. Den Dikken, pp.1-46.
Ordonez, Francisco. 2005. “Stressed Enclitics”. in Montreuil, ed. New perspectives on Romance Linguistics. Benjamins: 167-183.
Repetti, Lori and Mirian Kim. 2013. Bitonal pitch accent and phonological alignment in Sardinian” Probus25: 267-300.
Wackernagel, Jacob. 1892. Uber ein Gesetz der indogermanischen Wortstellung. Indogermanische Forschungen 1:333-436.