• Nie Znaleziono Wyników

3. Complexity, Accuracy and Fluency in a Second Language

3.1. Complexity

The diversity of definitions and approaches focusing on complexity can best be understood through a  non-linguistic explanation of the phenomenon: “A criterion of the complexity of a  phenomenon is the amount of information necessary to define it and the practical difficulty in obtaining that information” (Favre et al. 1988, p. 7). In other words, the deeper the insight the more elements we need in order to describe it accurately.

Complexity in second language development is considered to be the most difficult component to handle, both because of the ambiguous way it is defined in several studies and also owing to its multidimensional character. We can find more general definitions in the literature, which characterize complexity, such as “the extent to which learners produce elaborated language” (Ellis & Barkhuizen, 2005, p. 139), or more fine-grained descriptions, involving more definientia, and focusing on selec-ted aspects of language, such as in the definition of linguistic complexity referring to “the intrinsic formal or semantic-functional properties of L2-elements (e.g. forms, meanings, and form-meaning mappings) or to properties of (sub-)systems of L2-elements” (Housen, Kuiken & Vedder, 2012, p. 4). The latter view understands complexity to be two-dimension-al in character, being either cognitive or linguistic, where cognitive com-plexity, sometimes even called difficulty, expresses the relative difficulty a L2-learner has in processing second language elements. Such a view has been forwarded by DeKeyser (1998), Housen and Kuiken (2009), as well as by Williams and Evans (1998), and is paralleled by the taxonomy de-vised by Dahl (2004) and Miestamo (2008), consisting of absolute and relative complexity, where absolute complexity corresponds to linguistic complexity, and relative complexity to cognitive complexity.

It is not seldom the case that complexity is associated with difficulty, which we can see on the one hand in the view of cognitive complexity as being on a par with difficulty. A similar approach has also been adopted by Pallotti, who distinguishes between the complexity of tasks, which he identifies as difficulty in an objective sense, and the complexity of per-formance, which has a subjective character (Pallotti, 2009). The difficul-ty-factor of complexity seems to be one of its most distinctive features:

Skehan, for example, juxtaposes it with “challenging language” (Skehan, 2009, p. 511), while Trudgill identifies it with “difficulty of learning for adults” (Trudgill, 2001, p. 371). However, Dahl suggests that complexi-ty should not be treated as equivalent to difficulcomplexi-ty. Beginning with the assumption that relative complexity can be associated with difficulty he argues that “what an individual finds difficult obviously depends not only on the complexity of the object of learning but also on the individual’s previous knowledge” (Dahl, 2004, p. 282). Therefore, the subjective and objective aspects of complexity should be treated as two distinct entities so as to avoid any misunderstanding both in theoretical considerations and empirical investigations.

Pallotti distinguishes between three different meanings of complexity in linguistic research: structural, cognitive and developmental. The first reflects the formal range of linguistic elements and their relational pat-terns, the second is associated with the effort needed to process linguistic structures and the last, i.e. developmental complexity, refers to the pro-cess of mastering linguistic structures when the first or second language is being acquired (Pallotti, 2015). Structural complexity is very often treated as a representative of complex systems, where there are “many different elements each with a number of degrees of freedom” (Nichols, 2009, p. 111), “the number of discrete components that a language fea-ture or a  language system consists of, and the number of connections between the different components” (Bulté & Housen, 2012, p. 24) or “the number of different elements and their interconnections” (Pallotti, 2015, p. 120)

The most common classification proposed in most studies on struc-tural complexity involves a distinction between lexical complexity, on the one hand, and grammatical complexity, on the other. Lexical complexity is often understood as the diversity of lexical items used in a text. The more the vocabulary varies the more complex the utterance is perceived to be. Lexical diversity can be measured using, for example, type/token ratios, the Giraud Index or the D-index, while the type/token ratio has

42

very often been criticized due to its unreliability in view of its negative correlation with text length. Another way to investigate lexical complexity is to look inside the word-level, i.e. to measure the number of lexemes in a word or the number of derivational affixes, which, however, tends to il-lustrates grammatical rather than lexical complexity, as Bulté and Housen (2012) point out. Yet another view of lexical complexity, and one that has also been criticised, entails looking at the density of words, where lexical words should reveal themselves to be more complex than function words.

Even this perspective, however, has no reasonable explanation when we base our reasoning on the assumption that complexity means diversity of items AND their interconnectedness. In investigations of lexical density the mutual interconnections of lexical items are completely ignored.

Grammatical complexity can be analysed as morphological or syntactic complexity. However, morphological complexity seen as, for example, the frequency of inflectional morphology of words, is not often the subject of SLA-research. Syntactic complexity, on the other hand, has attracted much more attention and appears in a variety of studies on complexity.

Norris and Ortega investigated syntactic complexity from the point of view of the reliability of the adopted measures and their correspondence with the multifaceted characteristics of this proficiency skill. They dis-cussed the most commonly used measures, based on: (i) length, (ii) the amount of subordination and coordination, (iii) the variety, sophistication and acquisitional timing of grammatical forms and (iv) the frequency of certain grammatical structures treated as more sophisticated and (there-fore) later acquired (modals, passives or infinitival phrases). On the one hand, they point out that some measures are redundant and should not be used together in one study – especially those concerned with subordi-nation as the main indicator of syntactic complexity. On the other hand, they call for the use of complementary measures that can refer to other dimensions of complexity, such as, for example, the variety of grammat-ical constructions. In an analysis of 16 studies they severely criticized the incomplete view of complexity that emerges when applying only one measure, which turned out to be the most common practice among re-searchers. The most common measures either focused on global complex-ity, such as the mean length of T-unit, or, which was even more common, subordination, which Norris and Ortega treated as providing a too limit-ed understanding of complexity. What they observlimit-ed was the increasing-ly common view of syntactic complexity as a form of structural variety and sophistication. They thus argue that a reasonable way to investigate

complexity is to analyse it both generally and more specifically, i.e. using length-based measures in combination with measures exploring subordi-nation and phrasal elaboration (Norris & Ortega, 2009).

Following on from earlier approaches, Bulté and Housen presented a detailed overview of several aspects of L2-complexity that focused on grammatical and lexical complexity, and linked them together with differ-ent levels of construct specifications (theoretical, observational, and op-erational) and different measurements. They referred to forty studies on linguistic complexity, in which 40 measures were used. The most frequent-ly employed measures turned out to be those that are general in character, such as the mean number of words per T-unit and the mean number of clauses per T-(c- or AS-, respectively) unit for grammatical complexity, and the number of different words per total number of words (TTR) or the Guiraud Index (WT/√W) for lexical complexity, rather than measures that take into account more specific features, such as the frequency of specific linguistic constructions, e.g., passive forms or infinitival phrases (Bulté &

Housen, 2012).

If we view complexity as a  property of complex, dynamic systems we can assume that the development of this dimension will involve the self-organisation of several subsystems, such as lexicon, morphology and syntax. This self-organisation can proceed in the form of “parts separat-ed” to “parts joinseparat-ed” (see Ashby, 1962), which, for example, can be ex-pressed in the broader use of extended, complex phrases instead of sim-ple phrases, consisting only of the necessary constituting elements, or in the increasingly extended use of subordinated structures, such as sub-ordinated clauses or clause-like constructions used in order to package information, instead of adding linguistic elements in the form of coordi-nate constructions. A complex and dynamic view of the development of lexical and syntactic complexity should thus cover more than one aspect of it. Based on the definition that complexity in general refers to a variety of elements and their mutual interconnection we will thus investigate lexical complexity as lexical diversity, on the one hand, and phrasal elab-oration, on the other. Syntactic complexity, in turn, expresses itself not solely in subordination, but also in a variety of syntactic constructions.

And all these features are in ongoing interplay with one another during the entire developmental process.

44