France Martineau presented a variationist perspective on the evolution of French negation, looking at when and how the preverbal negative ne weakens, the postverbal items pas and point emerge and the n-words go from polarity items to negatives.
Weakening of ne can be interpreted not so much as semantic or formal loss of properties, but as lower frequency. There seems to be a competition between preverbal and embracing negation until the 17th century. Ne deletion however varies according to context, with interrogatives progressively allowing dropuntil the 19th century, when it reverses. Early examples of ne deletion as in Heroard seem related to weak subjects, which would predict a loss of ne in varieties allowing morphologisation of clitic subjects as diagnosed by subject doubling. Overall, in private letters, ne is dropped mainly in the 19th century, and less so in the 20th century. The rate of deletion is similar in Continental and Canadian French, the latter consistently displaying slightly larger percentages.
The competition between pas and point went from specialisation in partitive vs. non-partitive contexts, maybe because of a paucal meaning point might have [isn't pas supposed to have paucal meaning as the minimal unit of measure?], a distinction which is blurred in the 17th century where pas is found with partitives, until point became a regional or a formal variant; in this later capacity, it hardly ever drops the ne in the 20th century. Pas seems to win out because as data from Völker (2007) indicate, pas was prevalent in higher political and social classes, and therefore used in charters related to the chancery.
The status of n-word aucun indicates a relation between negative value and concord. The existential reading equivalent to someone and lost in the 17th century in Continental French is still found in early Canadian French; the polarity reading in averidical contexts peaking at 30% of all uses in the 15th century seems to recede slowly to 15% at present. From the Middle Ages, the negative value takes up the majority of uses. [Völker provided data from his 13th century Luxemburg charters that show that aucun is largely positive (90 occurrences vs. 14 negative ones), but that nul can also be at a lesser rate (30 positive vs. 90 negative uses).] This tallies with the use in conjunction with pas, which drops sharply after the 17th century. Although low at 3%, the use of what has become a full n-word with pas must therefore be negative concord. The use of document types more likely to approximate the vernacular are essential in assessing changes of sociolinguistic and grammatical status of markers which may present a fuller history of the language. This is what is promoted by the project Martineau is directing on the history of Canadian French, which is developing new vernacular corpora.
Lene Schøsler then provided a comparative presentation of the negative items and their reinforcement in the 9 manuscripts of the 13th century chanson de geste Le Charroi de Nîmes. Over 85% of sequences did not contain reinforcement. The emergence of reinforcements was explored through phonological, pragmatic and enunciative hypotheses. The phonological hypothesis according to which postverbal particles are promoted by phrase-final stress was queried as the text might be too early to reflect the shift to phrase-final stress, and that if change of stress is determinant, it should have affected other clitics to induce clitic-doubling for instance, which isn't the case. The pragmatic hypothesis by Schwenter (2005) that postverbal negatives emerge in activated propositions that are discourse-old met with the problems of manuscript variation: where activated contexts do not have reinforcement in some of the manuscripts (in several of which preverbal non is found), where discourse-new contexts contained reinforcement in one of the manuscripts, and where two coordinated verbs have reinforcement and not. Actualisation theory would predict that change comes from spoken language and is reflected in reported speech sequences. They comprise nearly 90% of mie, two-thirds of pas, but also 78% of ne alone. On the whole, there is some regional variation, the few point being found in Anglo-Normand and mie being highest in the North-West manuscripts.
The discussion raised the importance of developing corpora to account not only for the temporal dimension, but also registers, text types [more lexical variety of reinforcements in literary than legal texts], social groups and political identity. The evolutions noted take place on extended periods. This is illustrated by the use of ne alone, which has certainly declined in frequency [Völker cites 65% in his corpora, the contemporary situation being closer to 1% (Pouder 2008)], but has not changed in terms of contexts: the list of averidical environments for modern French would apply to Middle French according to Tony Lodge, and Mair Parry noted similar distribution for some Northern Italian dialects. Ne deletion has been going on for at least five centuries, in some linguistic contexts more than in others. Changes in context of use are illustrated by the competition between pas and point, the latter extending beyond partitive environments to acquire regional and register specialisation. Changes of status are demonstrated by aucun. Why some phenomena reverse their decline in the 20th century is a question to be considered, especially as to the role of schooling, literacy and the media. How length of competition, (in)variable contexts, and frequency are to be accounted in a grammatical model is a question that remains to be defined. The idea that variation is explained by (groups of) speaker having different but categorical grammars seems at odd with individual variation; variation might be captured by the distinction between a core grammar and more or less free alternative realisations suggested by David Willis. Could that explain why in the same source of a same speaker as in the Héroard diary, ne is dropped very frequently with weak subject and less so with other?
Richard Ingham identified the emerging consensus that the evolution of French negation is one where concord becomes possible. The question is how to analyse items that may have polarity and negative readings. Different lexical entries might be used as proposed by Ana Maria Martins. An indefinite trait that can yield negative effect in non-veridical contexts at a period and a negative trait that can be absorbed in averidical contexts at a subsequent period may constitute an alternative. In any case, it still seems necessary to define diagnostics [one mentioned was that a polarised noun like mot cannot be used after intransitive verbs like dormir is an indication that it has not grammaticalised] for when a word becomes a negative ; co-occurrence might not be enough in and of itself, as the contemporary Canadian French pas aucun might well be different from the formally identical strings in Claudel. Qualitative consideration must accompany quantitative data. It nonetheless looks as though the evolution of French negation is tied to negative concord, unlike that of English for instance.
Martineau, France. A variationist perspective on the evolution of negation
Schosler, Lene. Manuscrits du Charroi de Nîmes