Volume 65, Issue 3 - Special Issue: Non-Canonical Grammar / Guest Editors: Claudia Lange and Tanja Rütten

The present paper focuses on non-canonical syntactic constructions in the language of evaluation. It explores in which ways and under which conditions certain syntactic constructions deviating from the unmarked SVX pattern can support the expression of the writer’s/speaker’s attitude. The study is based on a corpus of academic and non-academic book reviews (published online and in print in scientific journals of different disciplines and on the cataloguing website Goodreads respectively) . It -extrapositions, existential there sentences, and wh -clefts contained in the corpus were analysed not only as to the evaluations expressed within them but also as to other parameters, including the occurrence of different forms of evaluation, the mention of specific sources or participating roles of evaluation as well as information structure.

In the present paper, I present a pilot study on fronting in six South Asian Varieties of English, i.e. Indian English (IndE), Bangladeshi English (BgE), Sri Lankan English (SLE), Nepali English (NpE) and Pakistani English (PkE) as compared to their historical input variety, i.e. British English (BrE). For each of these varieties, based on the South Asian Varieties of English (SAVE) newspaper corpus and the news section of the British National Corpus , 500 sentences per variety were manually parsed and the sentence-initial elements were annotated for their information status. Methodologically, I apply regression and CART classification tree analysis to test for variety-specific as well as universal features of fronting in SAVEs. The analysis reveals that constituents are mainly likely to be fronted when the information they represent is given (regardless of variety). Within this category, there are also clear variety-specific differences between IndE, PkE, SLE and BdE on the one hand (showing a generally higher frequency of object fronting), and BrE, MdE and NpE on the other (with a higher frequency of fronted adjuncts).

Topicalization refers to the sentence-initial placement of constituents other than the subject and is often listed as a non-canonical construction [cf. Ward, Gregory, Betty J. Birner and Rodney Huddleston (2002). “Information Packaging.” Rodney Huddleston and Geoffrey K. Pullum, eds. The Cambridge Grammar of the English Language . Cambridge: Cambridge University Press, 1363–1447.]. In this paper, tokens of topicalization in the direct conversations in the International Corpus of English for Hong Kong and India and, for comparison, Great Britain are analysed. In order to find out if topicalization is a contact-induced feature, typological profiles with regard to topic-prominence [Li, Charles N. and Sandra A. Thompson (1976). “Subject and Topic: A New Typology of Language.” Charles N. Li, ed. Subject and Topic . New York: Academic Press, 457–489.] are created for three Indo-Aryan, three Dravidian and two Sinitic languages. I suggest that the low frequencies of topicalization in Hong Kong English and the high frequencies of topicalization in Indian English are primarily due to differences in intensity of contact [Thomason, Sarah G. (2001). Language Contact . Washington, D.C.: Georgetown University Press.] and variety development [Schneider, Edgar W. (2007). Postcolonial English. Varieties Around the World . Cambridge: Cambridge University Press.]. Typological interference at the level of information structure is assumed to only come to the fore in further developed varieties and after prolonged contact.

From a pragmatic perspective, speech acts can be seen as non-canonical if they reflect perceptions of politeness and face that differ from an accepted norm. This paper traces the canonical or non-canonical status of boasting and apologising in Anglo-Saxon society. The data suggest that boasting developed from a fairly canonical to a more or less non-canonical speech act, depending on the relative influence that Germanic or Christian values had on Anglo-Saxon society. Apologising, on the other hand, was most likely non-canonical in the Anglo-Saxon warrior society since it appears to be incommensurable to the spirit of Germanic heroes. In the context of spreading Christianity, however, acts of penitence (which are here seen as ‘pre-apologies’) are increasingly advocated as canonical behaviour. Thus, (pre)-apologising (or showing penitence) was on its way from a non-canonical to a canonical status. While the exact development of these two speech acts after Old English has still to be traced in more detail, this article shows that it is the underlying set of ideals and values associated with a society (in this case Germanic self-assertion and retribution and Christian humilitas ) that shape the canonical and non-canonical status of expressive speech-acts.

In this paper, the contractions shoulda, coulda, woulda are compared with their respective full forms should have , would have , and could have . Although the full forms are used much more frequently and are, therefore, considered canonical, the non-canonical forms have increased in frequency throughout the better part of the twentieth century. They are predominantly used in American English – in conversation as well as in fictional writing to imitate speech. With respect to their syntactic environment, shoulda, coulda , and woulda behave differently than their full counterparts since they are often used without subjects and without lexical verbs. Some of these uses can be explained by the fact that shoulda, coulda , and woulda are not always used as verbal items but also as nouns, adjectives, and interjections. Due to their overall low frequency and their restriction to a particular register, however, it appears they will keep their non-canonical status for the foreseeable future.

The adjectival use of the word chosen is in the focus of this study. Based on data taken from the Corpus of Historical American English , the adjectival characteristics of chosen are investigated first. As chosen is a participial adjective, its uses with reference to its participial character are analysed. The third line of investigation outlines the co-occurrence patterns of chosen with nouns. The results from the corpus-linguistic analyses will be related to the theoretically deduced notions ‘canonical’ and ‘non-canonical’. A discussion in how far the notions ‘canonical’ and ‘non-canonical’ can be related to the corpus-linguistic results follows.