1 Parasynthesis in Romance

Abstract

Parasynthesis is a word-formation process that Romance languages have inherited from Latin. It is characterised by the simultaneous and joint attachment of two affixes (a prefix and a suffix) to a lexical base. In order to define the concept of parasynthesis, several theoretical tenets (e.g., the transcategorisation power of prefixes, the binary branching hypothesis, etc.) must be taken into account. In Romance languages, verbs are the most representative cases of this morphological process; there are, however, other non-verbal formations that have been included in this category.

1 Introduction

Parasynthesis is a linguistic term which goes back to the Greek grammarians, who called pa?as???es?? the process of forming derivatives, occasionally also compounds, on the basis of compounds (cf. Lindner 2011: 17-19). The result of the process was called a pa?as???et??. In Latin grammar, this latter term was rendered as decompositum. While this terminology continues to be used - parsimoniously - in the same way in most philologies (cf., for example, Henzen 1957: 222 on German "Dekomposita oder Parasyntheta"), in Romance philology it has undergone a semantic change, denoting nowadays a word-formation process whereby a prefix and a suffix are simultaneously attached to a lexical base (cf. Darmesteter 1875; Elliot 1884; Malkiel 1941; Allen 1981; Crocco Galèas and Iacobini 1993; Brachet 1999; Iacobini 2010). The standard example is constituted by Fr. embarquer 'to load, board', a verb consisting of a base barque 'ship', a prefix em- and a suffix -er, which is the infinitive ending. Neither *embarqu-, nor *barquer are actual, or even possible French words.

This terminological tradition in Romance philology goes back to Darmesteter (1875: 79-80):

Cette sorte de composition est très-riche: les mots qu'elle forme, et que l'on désigne du nom de parasynthétiques, offrent ce remarquable caractère d'être le résultat d'une composition et d'une dérivation agissant ensemble sur un même radical, de telle sorte que l'une ou l'autre ne peut être suprimée sans amener la perte du mot. C'est ainsi que de barque l'on fait em-barqu-er, dé-barqu-er, deux composés absolument uns et dans lesquels on ne retrouve ni les composés débarque, embarque, ni le dérivé barquer, mais le radical barque. La langue tire les deux composés immédiatement du radical, sans l'aide d'aucun intermédiaire. [This type of compounding is very rich: the words that it forms, which are called parasynthetic, show the special characteristics of being the result of compounding and derivation on the basis of the same root, in a way that neither of them could be eliminated without the word disappearing as such. In this way, from barque one creates em-barqu-er, dé-barqu-er, two compounds completely indivisible, and in which it is not possible to find either the compounds débarque, embarque, or the derived word barquer, only the root barque. The language forms those two compounds directly from the root, without the help of any go-between.]

As one can see, Darmesteter's use of parasynthesis still follows the Classical tradition, since he considered the em- of embarquer and the dé- of débarquer to be prepositions (i.e. Fr. en 'in' and de 'from') and not prefixes, and hence the combination of em- and dé- plus barque as cases of compounding. However, when during the 20th century elements such as em- and dé- came to be viewed as prefixes, the term parasynthesis was retained and reinterpreted as the simultaneous use of prefixation and suffixation, instead of compounding and suffixation.

This terminological tradition has been consolidated among early 20th-century Romance linguists from different origins, among them Menéndez Pidal (1904: 130), Nyrop (1908: 206) and Thorn (1909: 8). Most Romance linguists have agreed on the need to treat simultaneous double affixation as a necessary condition for parasynthesis, and tried to differentiate these formations from others where the presence of prefixes and suffixes is the result of consecutive processes (cf. Alemany Bolufer 1920: 152; Brøndal 1943: 125; Badía Margarit 1962: 394-395; Lloyd 1964: 736; Malkiel 1966: 314; Tekavcic 1968: 145; Reinheimer-Rîpeanu 1973: 487; Brea 1977: 127-128, etc.).

There were few exceptions to this restrictive view of parasynthesis. Tollemache (1945: 110) starts from a wider perspective, but eventually keeps the name parasynthetic for those cases where the affixing processes take place at the same time ("parasinteti simultanei", simultaneous parasynthetics, in his terminology). A real exception is Asan's (1965) proposal who defends a wide definition of parasynthesis that comprises not only formations resulting from attaching a prefix and a suffix simultaneously, but also those where prefixation and suffixation are consecutive processes. Still other scholars, as we will see, have been led to call into question the very concept of parasynthesis (cf. Serrano-Dolader 1995) due to the numerous problems of delimitation that it gives rise to.

Outside Romance linguistics, this special use of parasynthesis/parasynthetic is rare, which does not mean that the phenomena referred to do not exist. They are simply treated under different headings such as "prefixal-suffixal formations", etc.

2 Problems concerning the delimitation of parasynthesis

2.1 Meaning

A proper description of the concept of parasynthesis requires the formal analysis to be complemented with semantic considerations. The fact that one of the intermediate stages ([p + X] or [X + s]) is attested is not enough to justify the rejection of the parasynthetic character of a formation. It is necessary to verify that such a formation shows the proper meaning that allows one to compositionally derive the meaning of the whole word. Thus, for example, the non-parasynthetic analysis of Sp. embaldosar 'to tile' would be shaky if it were only based on the existence of baldosar 'to tile', which formally corresponds to the stage [X + s]. It is the semantic analysis of all the members of the word family embaldosar/baldosar/baldosa 'to tile/to tile/tile' that allows us to state that the form baldosar 'to tile' cannot be considered as the base for the creation of embaldosar 'to tile', but that both forms are derived from the noun baldosa 'tile', at least from a synchronic perspective. It is irrelevant that etymologists might establish a different derivation process from a diachronic point of view.

On the other hand, the semantic peculiarities of a given formation can allow it to be analysed in different ways. For instance, desnivelar 'to unbalance' has two semantic interpretations that correspond to two different morphological analyses: desnivelar1 'to cause to lose balance' (derived from nivel 'level' by parasynthesis) vs. desnivelar2 'to reverse the action of levelling' (derived from nivelar 'to balance, to level' by prefixation).

2.2 Actual and possible words

Aronoff's (1976: 21) word based hypothesis holds that "[a]ll regular word-formation processes are word-based. A new word is formed by applying a regular rule to a single already existing word". According to this view, it would be sufficient for a complex word to contain an intermediate stage which is not attested to qualify as a case of parasynthesis. Corbin (1980) takes issue with this requirement for intermediate stages to be attested. Her basic starting point is that notions such as "existence" and "non-existence" are ambiguous, and therefore unreliable. For Corbin, attested words and possible, but non-attested words should have the same status in grammar. The fact that an intermediate stage is not attested, therefore, is not considered to be a sound argument in favour of a parasynthetic analysis (cf. Corbin 1980: 191).

The idea of possible intermediate stages is exploited by several authors. In Scalise (1984: 204), for example, parasynthetic forms are formations in two stages: at the first stage (suffixation), a possible, but not necessarily actual word is generated, while at the second stage (prefixation), the whole word is actually generated. According to this account, the derivation of a Spanish word like engordar 'to fatten' (from gordo 'fat') would go through a non-existent intermediate stage *gordar.

2.3 Binarism

The binary branching hypothesis, which goes back to the structuralist principle...