• No results found

In the next chapters I will sometimes use sentences found on the World Wide Web to illustrate certain word order phenomena. On some occasions I may use results from Web searches in addition to examples from the NDC, or, when a construction was not attested in the corpus, as the sole source of “non-introspective evidence”. My use of Web searches is purely qualitative; I will not make any claims based on the number of search hits. I will not take a search that did not return any results (i.e. the non-occurrence of a specific construction on the Web) as an indication of ill-formedness. The example sentences found on the Web are used only to confirm my own (positive) intuitions. Despite the uncertainties (some of which are mentioned below) connected to these sentences, I still think they form a

72 Chapter 4 non-negligible body of empirical support.

The dialect data from Web searches were typically found searching for strings of words in double quotes (to ensure that the exact word order was retrieved) using the search engine GoogleTM. In contrast to the contents of a corpus like the NDC, the contents of the Web are not in any way grammatically tagged, which means that you can neither search for parts of speech nor lemma forms. Given that I have mostly been interested in search results written in dialect, I have had to type the words accordingly. Dialectal spelling is not always predictable, as single words may have multiple spelling options. As an example, the wh-word ‘why’ that in Norwegian Bokmål is written hvorfor, comes in at least nine flavors when written in any of the Rogaland dialects: Korfor, koffor, koffår, kaffår, kaffor, kafor, kafår, kefår, and kefor. Shorter words are also subject to massive orthographic variation;

the negationikke ‘not’ may be written as ikkje, ikje, issje, isje, iche, ittje, itje, kje, sje, che, or tje. In effect, this means that a seemingly simple combination such as hvorfor ikke ‘why not’ in principle has ninety-nine different possible spellings.

Needless to say, trying out all of them clearly complicates the search process. At the same time, the disadvantage of this unpredictable and more phonetic-like spelling is a natural consequence of a greater advantage, namely that the material written in dialect resembles spoken language, and as such useful and relevant here.

It has often been claimed that a drawback of using sentences found on the Web is that one rarely knows their true origin, that is, who or what produced them. (Neither can one with certainty say anything about their grammatical status.) As discussed in Schütze (2009), any sentence found on the Web may have been produced by a non-native speaker, or, by a native speaker, but not with its intended meaning due to an error of any imaginable kind. In fact, it may not even be produced by a person, but rather automatically generated by machine translation. In short, “merely having found instances of a construction of interest should not be construed as evidence of anything ipso facto” (Schütze 2009, p.

152). The examples that I have collected on the Web, however, are taken from blogs, online discussion groups, etc, and, importantly, almost without exception accompanied with information regarding where the authors live and come from, along with other personal facts such as age and sex—information that eliminates at least the possibility of a non-native speaker or non-human origin of a sentence. I

Methods and material 73 will supply such information in a footnote for each example taken from the Web in the following chapters (to the degree such details have been available). In addition, the URLs of the Web pages from which examples are taken are numbered and listed in appendix A. Each Web example in the text is followed by its URL number in parentheses.

On the positive side, the Web provides arenas for a type of language that does not exist elsewhere: Online forums, blogs, discussion groups, social network sites, and so forth. In addition, and of particular significance here, much of its contents is written in dialect. This type of written material is to a much lesser degree subject to the before-mentioned norms and prescriptive standards specific to written language compared to writing that conforms to any of the written standards Bokmål or Nynorsk. And, as pinpointed by Schütze (2011, p. 209), “[s]ome subset of it [the material on the Web] blurs the line between traditional spoken and written language in the sense that, while it is generated from a computer keyboard, it is part of a nearly real-time conversation and undergoes minimal editing or self-correction.”

In sum, the Web can be a useful source when searching for dialect data, and to a certain degree it shares some of the properties of speech corpora (oral style, semi-spontaneous, unbiased, not produced specifically for linguistic research), its main advantages being its ease of access and vast mass of data.

Chapter 5

Non-V2 in the Rogaland dialects:

wh -questions

5.1 Introduction

This chapter has two main goals, namely (i) to show that there is no complexity constraint on non-V2 in the Rogaland dialects, and (ii) to present and analyze the so-called “wh+nå/då construction”.

The chapter is organized as follows. In 5.2 I return to the complexity constraint on non-V2 dicussed in 3.2.5. In 5.3 I introduce one of the central themes of my thesis: Wh-questions with V4 word order caused exclusively by the sentence adverbs ‘now’ and ‘then’ appearing between the wh-phrase and the subject or the complementizer. In 5.4 different analyses of these questions are proposed.