Word-formation complexity: a learner corpus-based study

Размещено на http://www.allbest.ru/

Размещено на http://www.allbest.ru/

Word-formation complexity: a learner corpus-based study

Olga Lyashevskaya

Julia Pyzhak

Olga Vinogradova

Abstract

This article explores the word-formation dimension of learner text complexity which indicates how skilful the non-native speakers are in using more and less complex - and varied - derivational constructions. In order to analyse the association between complexity and writing accuracy in word formation as well as interactive effects of task type, text register, and native language background, we examine the materials of the REALEC corpus of English essays written by university students with Russian L1. We present an approach to measure derivational complexity based on the classification of suffixes offered in Bauer and Nation (1993) and then compare the complexity results and the number of word formation errors annotated in the texts. Starting with the hypothesis that with increasing complexity the number of errors will decrease, we apply statistical analysis to examine the association between complexity and accuracy. We found, first, that the use of more advanced word-formation suffixes affects the number of errors in texts. Second, different levels of suffixes in the hierarchy affect derivation accuracy in different ways. In particular, the use of irregular derivational models is positively associated with the number of errors. Third, the type of examination task and expected format and register of writing should be taken into consideration. The hypothesis holds true for regular but infrequent advanced suffixal models used in more formal descriptive essays associated with an academic register. However, for less formal texts with lower academic register requirements, the hypothesis needs to be amended.

Keywords: linguistic complexity, morphological complexity, writing accuracy, word formation, English, learner corpora

Аннотация

Словообразовательная сложность и ошибки учащихся в экзаменационных эссе

О.Н. Ляшевская, Ю.В. Пыжак, О.И. Виноградова

В статье рассматривается словообразовательная сложность учебных текстов, которая трактуется как система измерений, показывающих разнообразие приемов словообразования разного уровня, от простых до продвинутых, используемых учащимся. Анализируется взаимосвязь между сложностью и ошибками, которые учащиеся допускают в словообразовании. Исследование основано на материалах REALEC - корпуса английских экзаменационных эссе, написанных студентами университета с родным русским языком. Предлагается подход к измерению словообразовательной сложности, основанный на классификации суффиксов Бауэра и Нейшена (Bauer&Nation1993), и анализируется соответствие между показателями индексов сложности и количеством ошибок словообразования, размеченных в текстах корпуса, с учетом типа экзаменационного задания. Постулируется гипотеза о том, что с увеличением сложности количество ошибок должно уменьшаться, и проводится статистический анализ параметров сложности и безошибочности. В работе показано, во-первых, что использование словообразовательных суффиксов более высокой сложности связано с количеством ошибок в текстах. Во-вторых, разные уровни иерархии сложности оказывают разнонаправленное влияние на точность: в частности, использование нерегулярных словообразовательных моделей положительно связано с количеством ошибок. В-третьих, следует учитывать тип экзаменационного задания, в том числе ожидаемые формально-регистровые особенности текста. Гипотеза была подтверждена для регулярных, но нечастотных суффиксальных моделей при их использовании в описаниях рисунков и графиков - текстах, следующих определенному формату и включающих элементы академического письма. Однако в случае аргументативных эссе выдвинутая гипотеза требует уточнения.

Ключевые слова: лингвистическая сложность, морфологическая сложность, безошибочность письма, словообразование, английский язык как иностранный, учебные корпуса

Main part

Text complexity is one indicator that can be used to assess authors' written and spoken skills and how varied and complex are the means of linguistic expressiontheyapply. Starting with the work of Norris and Ortega (2009) and two publications by Bulte and Housen (2012, 2014), researchers agree that complexity is a multifaceted phenomenon, consisting of several sub-constructs, dimensions, levels, and components, and many of these constructs and sub-areas have been independently evaluated. Consequently, the term ` `complexity'' has been widely applied to manifestation of objective properties of linguistic production. Bulte and Housen (2012) also draw the distinction between propositional complexity, discourse-interactional complexity, and linguistic complexity. Of these three, linguistic complexity of written production, will be the target of the present article.

In addition to the general consent on including lexical, syntactic, discursive, and morphological parameters of texts when looking at text complexity, it soon became clear that the majority of researchers focus on the first three areas, while morphological and phonological complexity or complexity phenomena at the interfaces between the traditional levels of linguistic analysis were largely ignored, which resulted in some appeals to the linguistic community to expand the construct of, and research on, complexity beyond the syntactic and lexical levels. In 2019, Centre for English Corpus Linguistics at UCLouvain hosted the colloquium, Broadening the Scope of L2 Complexity Research, and in a way the research in this paper is our contribution to bridging this gap. More specifically, we set aside the topic of inflectional complexity, which has already gained considerable attention in learner data analysis and has acquired its own methodology (Brezina & Pallotti 2016, Yoon 2017, Tywoniw & Crossley 2020), and focus exclusively on the diversity of word-formation models.

Roots and affixes are the building blocks of morphological competence. Acquiring the morphology of a second language can be seen as not only learning strategies for composing and decomposing forms from and to the building blocks and enhancing vocabulary. It is also learning relations between the stored word forms that facilitates processing at the morphological level (Baerman et al. 2015). Word-formation skills give L2 learners flexibility when modifying and adapting word meaning to the context, coercing a word's syntactic class to fit into an available grammatical construction, choosing the right vocabulary for a given type of discourse, or in constructing new words. Inappropriate use of derivational models may result in communication fallacy and other losses in social interactions, just as any kind of erroneous linguistic behaviour may do.

Recent discussion on the status of word(-formation) families has brought to light many derivation-related issues important for L2 research and education (Brown et al. 2020, Laufer et al. 2021, Nation 2021). To put it in a nutshell, the researchers emphasise that the derived forms cover a significant portion of texts and hence learners' word-formation competence impacts text comprehension. Affix knowledge develops with general proficiency and facilitates vocabulary learning. However, many advanced learners have limited or patchy knowledge of affixes and find it challenging to identify derivational forms of known headwords given in context, even in structures with top-frequent affixes. Brown et al. (2020) present some evidence that the learners are concentrated on the recognition of the affix meaning rather than its grammatical function. Although the evidence provided largely concerns the receptive aspects of acquisition, the discussion has important implications for the theories of L2 production.

In the task of examination writing, learners are expected to achieve a balance in the interaction of the performance areas such as complexity, accuracy, and fluency in choosing preferable derivational strategies. Although the lack of high - level and complex skills is not the only source of word-formation errors and that low writing quality may also be related to, for example, time and stress management, genre of the task and familiarity of the topic, and learners' individual effects, learner corpora and annotated research datasets provide the basis for empirical studies focused on the relationship between complexity and accuracy (Skehan 2009, Bardovi-Harlig & Bofman 1989, Plakans et al. 2019, Lahuerta 2018, among others).

Our study is based on the learner corpus REALEC (Vinogradova et al. 2017), which includes examination writings of students with Russian L1. We examine the use of word-formation constructions with the focus on suffixal ones, since their number in the examination texts significantly exceeds the number of prefixes and other word-formation units. We will test the following hypothesis: the higher the parameter of derivational morphological complexity, the less often errors occur in word formation, since the scale for measuring morphological complexity is based on the order in which students studying English as a foreign language learn derivational affixes. Accordingly, the more the student uses advanced suffixes, the higher their language proficiency.

Word-formation complexity

There are several ways to define complexity, among which relative complexity (difficulty) and absolute (structural) complexity are most discussed (de la Torre Garcia et al. 2021). Relative complexity refers to the cognitive difficulty of the task, the amount of effort and resources that a speaker has to employ in order to process and make use of a linguistic structure. In contrast to this, absolute complexity is defined as the numerical characteristics of a text based on the quantity of encoded and encoding linguistic units and the number of connections between these components. Morphological complexity belongs to the area of formal parameters of absolute complexity, according to the classification of parameters critical for measuring the acquisition of a target language (Bulte & Housen 2012), see Fig. 1. These are features which can be measured objectively at the word level in a text.

Figure 1 shows that the criteria of morphological complexity are divided into two groups: inflectional and derivational. The first type refers to the use of grammatical forms, for example, the frequency of tense forms, the frequency of modals, the number of different verb forms, the variety of past tense forms, and MCI (Morphological Complexity Index) (Brezina & Pallotti 2019). The second group of criteria deals with the use of derivational affixes, composites (multi-root words), and conversives, and refers to the variability of word-formation models and the size of word families.

Word-formation complexity criteria have gained much less attention compared to inflectional complexity criteria. A possible explanation for this is that there is low agreement among theoretical approaches to this level of linguistic description, the coverage of derivational models, affixes, and that their taxonomy in lexical resources is sketchy and inconsistent (cf. Table 1), and it is problematic to use (a very few) tools for automatic morpheme segmentation of non-standard texts. It is indicative that Biber used only one derivation-related feature Nominalizations (words ending in - tion, - ment, - ness, - ity) among the other 67 linguistic features in his multidimensional analysis of English register (Biber 1988). A slightly longer list of regular affixes (-able, - er, - ish, - less, - ly, - ness, - th, - y, non-, and un-) was used in the calculation of a types-per-family ratio in a study of vocabulary structure (Horst & Collins 2006), yet no distinction was drawn between inflection and derivation in the construct of word families.

Recently, Tywoniw and Crossley (2020) introduced the TAMMI method, in which the density of derived and non-derivational words is measured in the text, among other metrics of (inflectional) morphological complexity. To calculate the Derived Word Tokens per Word index, they divide the number of word tokens with any derivational affixes by the number of words in a text. The Derived Word Types per Type index is the number of distinct word types with any derivational affixes divided by the number of types in a text. Similar calculations apply to the measurement of Non-Derivational Word Tokens per Word and Non-Derivational Word Types per Type indices.

Tywoniw and Crossley's approach is a generalisation over statistical measures developed for tokens (N) and types (V) of a given morphological class in corpora - based research (Plag et al. 1999). Other derivational complexity indicators recommended in the literature include:

- the length of the affix;

- the complexity-based rank, constructed on the affix ordering principle: suffixes with a rank greater than the rank of a given affix may follow that suffix in a word, while suffixes with rank lower than this will never follow it;

- the derivative's junctural phonotactics: the probability of the sequence of sounds spanning the juncture between the morphemes (e.g. «nh», which is highly unlikely to occur within morphemes and more likely to create morphological boundaries, as in inhumane) (see Baayen 2009 for an overview).

Table 1. Numbers of distinct affixes attested in English L1 corpora, dictionaries, and other resources according to (Laws & Ryder 2014: 6, 10)