Collocations as a component of lexical competence in Croatian as a second and foreign language

Ordulj, Antonia. (2016). Collocations as a component of lexical competence in Croatian as a second and foreign language. PhD Thesis. Filozofski fakultet u Zagrebu, UNSPECIFIED.
(Poslijediplomski doktorski studij glotodidaktike) [mentor Cvikić, Lidija].

Preview

PDF (Croatian) Download (2MB) | Preview

Abstract

Even though syntagmatic relations were the subject of linguistic research since the beginning of the 20th century, a comprehensive investigation of collocations started only in the 1950s, primarily thanks to John Rupert Firth (1957). Firth was the first to use the term collocation in lexical semantic discussions; however, up until this day there is no clear and unambiguous definition of this lexical semantic phenomenon. The term collocation (Latin com together + locare situate) is usually defined as a co-occurrence of individual lexical items at the syntagmatic level. There are many attempts to define collocations in relevant literature. For example, Nation (2001: 516) mentions a group of words that belong together, either because, they commonly occur together, like take a chance, or because their meaning is not obvious from the meaning of their parts, as with by the way. Lewis (2000) draws attention to the natural co-occurrence of two words, therefore emphasising that not any two words that co-occur in a text should be understood as a collocation. Apart from these definitions, there are different approaches to studying collocations in relevant works; namely, the lexical composition approach (Firth, 1957; Halliday, 1966; Sinclair, 1966), the semantic approach (Mel'čuk, 2008; Katz & Fodor, 1963; Cruse, 1986) and the structural approach (Matthews, 1964; Mitchell, 1971; Greenbaum, 1974). Representatives of the lexical composition school of thought consider lexis an autonomous entity separate from grammar, and therefore define collocations as a manner in which the lexical meaning is realised, arising from the habitual co-occurrence of two lexemes along the syntagmatic axis (Firth, 1957). The semantic approach aims to predict the collocational span of a specific lexical unit by reference to its semantic features. The structural approach, on the other hand, argues for the unity of lexis and grammar and tries to establish the collocational structure. However, due to different definitions and theoretical approaches to the study of collocations, there are still no universal criteria that would clarify why certain lexical units collocate, how the meaning of the collocation is achieved, and how to differentiate collocations from other lexical combinations, especially phrasemes and free lexical combinations. Apart from theoretical linguistics, collocations have also been studied within Second Language Acquisition (SLA). In the late 1990s, Hill (1999, as cited in Martyńska 2004) introduced the notion of collocational competence, thus moving the focus from teaching individual lexical items to developing collocational competence. Even though many experts argue that collocations are a neglected area of lexical competence (Gitsaki, 1999; Koya, 2003; Martyńska, 2004), so far the most extensive research has been done on collocations in English as a Second Language (ESL). It is believed that the development of general vocabulary increases the productive and receptive collocational knowledge, which has been confirmed in research for English (Koya, 2003; Gitsaki 1999) and Italian as L2 (Bonci, 2002). Furthermore, collocational usage can be a clear indicator for differentiation of lower and higher proficiency levels of non-native speakers. Students on higher levels of ESL use more complex types of collocational structures compared to students on lower levels, who mostly rely on frequent collocations consisting of nouns, adjectives, and verbs (Gitsaki, 1999; Zhang 1993). The reasons for this could be found in the exposure to collocations, differences in the first language, the structural and semantic complexity of collocational types, but also in the educational context, i.e. the exposure to different collocational types in the studying process (Begagić, 2104; Gitsaki, 1999;). In Croatian, the situation is somewhat different. The largest number of research on collocations in Croatian linguistics is concerned with lexicographical processing of collocations, since a Croatian collocation dictionary still does not exist. The most detailed description of noun collocations in Croatian as first language so far was given by Blagus Bartolec (2014), but research work on collocational competence in Croatian as a Second Language (CSL) is sporadic (Petrović, 2007; Burić and Lasić, 2012). More extensive research on collocation acquisition in CSL does not exist, whereas only Udier (2015) discusses collocational competence in the Framework of Reference for B2 Croatian, within grammatical competence. The lack of research on collocational competence of non-native speakers of Croatian and the assumption that morphological features of words can affect collocational knowledge were enough motivation for this research. This thesis consists of two main parts. The theoretical part discusses approaches to teaching collocations, lexical combinations, and collocational competence in second language. The practical part presents a research on collocational competence in CSL, which was preceded by two pre-researches; the evaluation of the associative power of collocations with native speakers and a corpus analysis of collocations. These are followed by a discussion and a conclusion. Firstly, it is necessary to define collocations and go back to the criteria for differentiating collocations from other lexical combinations in Croatian (phrasemes, multi-word lexical units, names, and free lexical combinations). WORD COMBINATIONS In Croatian, there are three types of fixed lexical combinations: phrasemes, terms, and names (onemes), and Turk (2000) draws attention to their similarities and differences. She mentions the complex structure, semantic transformation of at least one element, and productivity as similarities, and stability, connotative power, and idiomaticity as differences. Another important feature that Turk (2000) emphasises in differentiating mentioned lexical combinations is their communicative function. Phrasemes are seen as expressions of secondary nomination which indicate personal, emotive attitudes, and opinions; terms are used for communicating and defining concepts amongst experts in a scientific or professional area; whereas the primary role of names (onemes) is to identify objects. The discussion about lexical combinations would not be complete without mentioning free combinations, since together with phrasemes, they represent the largest stumbling block for linguists to clearly define collocations. Free combinations are those combinations of words that are characterized by many possibilities of combining and exchanging elements, and therefore, a lower possibility of co-occurrence (Petrović, 2007; Borić, 2002, 1996; Pritchard, 1998;). In this thesis, the difference between collocations and free combinations is set by taking into consideration the freedom of combining lexical units and the semantic criterion. Those lexical items whose elements can be combined with other lexical units without limits are considered as free lexical combinations. However, collocations are lexical combinations that arise from the syntagmatisation process, where assimilation or change in meaning of individual elements causes the creation of a new collocational meaning. The most complete definition of collocations in Croatian was given by Blagus Bartolec (2014: 80), who argues that a collocation is a „special lexical combination at the syntagmatic level based on semantic connectedness of individual lexical units, which specifies their meanings”. This definition needs further explanation. Blagus Bartolec (2014) limits collocations to two content words that formally denote a syntagm. The communicative function of language is especially important here, i.e. the usage among native and non-native speakers, which makes collocations a special type, different from other types of lexical combinations. To understand the difference between collocations and other lexical combinations, it is important to mention their semantic features. Therefore, it is should be mentioned that collocations in Croatian arise from the process of syntagmatisation, which means that they do not have a status of lexical units. This is due to the semantic adaptation arising from the semantic potential of the second word that enters into the collocational relation. Because of the semantic potential of the components, one component is modified by the other and a new, collocational meaning appears, which is “motivated by the extralinguistic phenomenon that is being referred to (sunny / rainy day, long / short hair; new / used car)” (Blagus Bartolec, 2014: 85). Additionally, the meaning is specified, i.e. narrowed down to a specific communicative situation. In this thesis, the above mentioned definition by Blagus Bartolec (2014) is accepted. However, since collocations have never been studied in the context of CSL, this research deals only with basic collocations having nouns as bases and adjectives as their collocators. Semantically, the basic collocational type refers to “the basic level of primary meaning specification” (Blagus Bartolec, 2014: 92). The division into base and collocator allows us to perform a corpus analysis of these types of collocations, which was done in this thesis. Furthermore, previous research has shown (Burić and Lasić, 2012; Duplančić Rogošić, 2007) that Noun+Verb is the most common collocational type in Croatian. RESEARCH PROBLEMS AND HYPOTHESES The main goal of this thesis is to examine the knowledge of noun collocations based on their frequency and collocational strength, and considering the CSL proficiency level. There are several specific research questions and hypotheses: Q1. To examine the receptive collocational knowledge considering the frequency of collocations and CSL proficiency level. H1. Subjects of higher CSL proficiency will have a better receptive knowledge than subjects of lower CSL proficiency. Q2. To examine the productive knowledge of noun collocations considering the associative strength of collocational elements and the frequency of collocations for both morphologically unmarked collocations in the nominative case and morphologically marked collocations in oblique cases considering the CSL proficiency level. H2a. Subjects of lower CSL proficiency will have more correct answers for morphologically unmarked collocations in the nominative case compared to morphologically marked collocations in oblique cases. H2b. Subjects of higher CSL proficiency will have an equal knowledge of morphologically unmarked collocations in the nominative case and morphologically marked collocations in oblique cases. Q3. To examine the productive knowledge of noun collocations considering the associative strength of collocational elements, the frequency of collocations in a wider context, and the CSL proficiency level. H3. Subjects of higher CSL proficiency will have more correct answers than subjects of lower CSL proficiency. Q4. To examine the answer types in receptive knowledge tasks about noun collocations considering their frequency and subjects’ CSL proficiency level. Q5. To examine the answer types in receptive knowledge tasks about noun collocations considering their frequency, associative strength, and subjects’ CSL proficiency level. For this research it was necessary to collect several hundreds of collocations that represented a working corpus and were used to create the research instruments. Collocations were collected from essays written by students of Language exercises in Croatian as L2 on levels B1 and B2. The validity of collected collocations was verified in the Large dictionary of Croatian language (Anić, 2003) and the hrWaC corpus (Ljubešić and Erjavec, 2011), which was also used to count the frequency of collocations. Finally, the associative strength of collocational elements was assessed by native speakers of Croatian. Based on their frequency and associative strength, collocations were divided into two groups: those of lower (LF) and higher frequency (HF) and those of weak and strong associative strength. The sample comprised of 70 students of Language exercises in Croatian as L2 on B1, B1+, B2, B2+ and C1 levels. The sample structure is very heterogeneous and implies a diverse representation of gender, language (pre)knowledge levels, and first languages. INSTRUMENTS 1. THE EVALUATION OF COLLOCATIONAL ASSOCIATIVE STRENGTH In order to determine the associative strength between collocational elements, an assessment was made on a sample of about 50 native speakers of Croatian. Subjects estimated the strength of association between collocational elements on a scale from 1 to 5. After the evaluation of the associative strength, descriptive data (M, SD) that represent the average associative strength was calculated for each collocation. 2. MULTIPLE CHOICE TEST In order to examine the receptive knowledge of collocations in CSL, subjects completed a multiple choice task (Task B) that consisted of 16 sentences. Each sentence, apart from the correct answer, had three distractors. Distractors comprised false pairs from subjects’ first languages (English, Polish, Spanish, and German) and Croatian adjectives of similar formation. 3. BLANK-FILLING TESTS In order to examine the influence of morphological features on collocational knowledge, subjects completed two blank-filling tasks that comprised forty isolated sentences, i.e. sentences without a wider context. In the first task, collocations were in nominative (Task A), i.e. the unmarked case, whereas in the second task (Task D), sentences were in oblique cases. In the third part of the test subjects solved a task in which collocations were put into more informative, wider contexts (Task C). General distribution of collocations in the instruments comprised an even number of collocations considering the frequency and associative strength of collocational elements. In every task, the collocation base was given (a noun) and subjects showed their receptive and productive knowledge of the collocator (adjective). The results were analysed both qualitatively and quantitatively (answer type typology for collocator choice). Statistica analytic package was used to analyse the results. RESULT ANALYSIS AND DISCUSSION The answer type analysis will be presented first, followed by the response accuracy analysis for receptive and productive tasks on lower and higher CSL proficiency. ANSWER TYPE ANALYSIS OF THE RECEPTIVE LEVEL TASKS As expected, the percentage of correct answers was higher at a higher CSL proficiency level for both collocation types considering their frequency (80% for HF and 76% for LF) (Table 11: 103). However, among the rest of the answers, contrary to expectations, on lower proficiency levels a higher percentage of distractors similar in formation was chosen (15% for HF and 14% for LF), whereas an almost equal percentage of false pairs was chosen on both CSL proficiency levels, but only with collocations of lower frequency (14% for lower and 13% for higher language proficiency). But since this was a multiple choice type task, relying on a distractor type could be explained by the lack of collocational acquisition in general. On the other hand, relying on formative distractors shows that subjects are aware of morphology and are able to recognise already acquired formative elements and use those to recognize and thus produce new lexical items. ANSWER TYPE ANALYSIS OF THE PRODUCTIVE LEVEL TASKS The answer type analysis comprised three tasks of the productive level: 1) isolated sentences with collocations in the morphologically unmarked nominative case (Task A), 2) isolated sentences with morphologically marked collocations in oblique cases (Task D), and 3) collocations in a wider context (Task C). Responses were grouped into three categories (lexical, grammatical, and overlapping answers), which were then divided into different types or deviations. In the category of lexical answers, the following deviations were noticed: 1. phonological and orthographical deviations (FP) – affect the change in the phonological status of the lexical unit 2. neologisms (NT) – lexical units that arise based on word formation rules in Croatian 3. collocators out of context (KIK) – existing lexical combinations in Croatian, but ut of the given context 4. unknown motivation of derivation (NM) – unclear criteria in collocator choice In the category of grammatical answers, the following two types of deviations were noticed: 5. incorrect case form of the collocator (NOK) 6. incorrect part of speech of the collocator (VR) In the category of overlapping answers, the following combinations of above mentioned answer types appeared: 7. collocators out of context + incorrect collocator form (KIK+NOK) 8. collocators out of context + phonological and orthographical deviations (KIK + FP) 9. collocators out of context + phonological and orthographical deviations + incorrect case form of the collocator (KIK + FP + NOK) It should be noted that only answers that were correct both orthographically and semantically qualified for the category of collocators out of context (KIK), and additionally grammatically correct for the KIK category. It can generally be said that there is a high percentage of incorrect answers and answer type collcators out of context (KIK) on both levels of CSL proficiency in productive level tasks (A, D, C). However, the percentage of correct answers by subjects of higher CSL proficiency is higher than by those with lower CSL proficiency in all productive tasks, regardless of frequency and the associative strength of collocations, which shows that higher CSL proficiency level means better collocational knowledge. A more detailed analysis showed that the most common answer types by subjects of lower CSL proficiency are correct answers (22%) and neologisms (11%) (Table 14: 108) for collocations of higher frequency and associative strength, but only in tasks in the nominative case (Task A). Additionally, 28% of answers (Table 18: 112) were collocators out of context (KIK) for collocations of a lower frequency and higher associative strength by subjects of higher CSL proficiency. In all productive level tasks (A, D, C) the category of lexical answers prevails, especially collocations out of context (KIK), which signals to earlier acquired lexical units that are possible to combine with a given base in order to define it more precisely in a given context. The usage of neologisms signals to acquired formative morphemes and the fact that non-native speakers have the ability to form new lexical items. In general, the analysis shows that the percentage of correct answers in receptive level tasks is higher than in productive level tasks, on both CSL proficiency levels. Moreover, the frequency of collocations appears to be a very important factor since the highest percentage of correct answers (80%) (Table 11: 103) in receptive level tasks is noticed for collocations of higher frequency and on higher CSL proficiency levels. In summary, it can be said that the knowledge of noun collocations on the receptive level is better for higher than for lower CSL proficiency levels, considering collocational frequency. RESPONSE ACCURACY ANALYSIS OF THE RECEPTIVE LEVEL TASKS The results show that subjects of higher language proficiency have better receptive knowledge of more frequent noun collocations than subjects of lower CSL proficiency (Image 1). The latter result was already confirmed in research on native speakers of Croatian (Ordulj and Cvikić, in print), showing that they have a better receptive knowledge of collocations of higher frequency, which are more common in language use. RESPONSE ACCURACY ANALYSIS OF THE PRODUCTIVE LEVEL TASKS In order to answer research questions and hypotheses, the accuracy of answers in productive level tasks was analysed for isolated sentences with morphologically unmarked collocations in the nominative case (Task A) and morphologically marked collocations in oblique cases (Task D) with subjects of lower (N=38) and higher (N=32) CSL proficiency levels. Response accuracy analysis for tasks with collocations in the nominative case (A) and oblique cases (D) for lower CSL proficiency Hypothesis H2a was confirmed for lower CSL proficiency level. However, results imply several particularities considering the frequency and the associative strength of collocations and the fact that subjects know collocations of higher frequency better, but only in morphologically unmarked case tasks (Task A), where the associative strength of collocational units did not affect the collocational knowledge. Furthermore, subjects of lower CSL proficiency show a very weak knowledge of morphologically marked collocations in oblique cases (Task D), regardless of their frequency and the associative strength of collocational units (Table 18; Image 2). It can be concluded that the answer accuracy both tasks is very low for subjects of lower CSL proficiency, which shows that their collocational competence is still not developed. The reasons for this could be the morphological complexity of Croatian, language exposure, and structural and sematic characteristics of collocations. Response accuracy analysis for tasks with collocations in the nominative case (A) and oblique cases (D) for higher CSL proficiency Hypothesis H2b was fully confirmed and all subjects of higher CSL proficiency showed an equal knowledge of noun collocations in the nominative case (Task A) and in oblique cases (Task D). Moreover, the frequency and the associative strength arose as important factors for better collocational knowledge (Table 19; Image 3). The analysis showed that subjects knew best those collocations of higher frequency and associative strength. Furthermore, the response accuracy for collocations of higher frequency and weaker associative strength is also satisfying, probably due to more frequent language usage of high-frequency collocations, predictability of collocational elements, and creation of stronger mental representations between lexical units in the mental lexicon. Compared to subjects of lower proficiency levels, who show a low knowledge of noun collocations in oblique cases and a slightly better knowledge in the nominative case, subjects of higher proficiency level show an equal knowledge of noun collocations in the nominative and oblique cases, which means that morphological features did not significantly affect collocational knowledge. These results imply that subjects of higher CSL proficiency have better collocational knowledge. Furthermore, it should be mentioned that these results disagree with Udier (2015: 270), who argues that B2 level users have a highly developed collocational competence. In accordance with these results, we can say that subjects of higher CSL proficiency have better developed collocational competence than subjects of lower CSL proficiency, but not highly developed. This is in accordance with research done in ESL (Gitsaki, 1999; Bonci (2002) for Italian as SL), which shows that subjects of intermediate language proficiency still show a weak knowledge of lexical collocations, whereas thir usage of grammatical collocations is rising. The rise of lexical collocation usage can be noticed only at the advanced ESL proficiency level, which also signals to the previously acquired extensive vocabulary. Response accuracy analysis for tasks with collocations in a wider context for lower and higher CSL proficiency The analysis partially confirmed hypothesis H3, because subjects of higher CSL proficiency had more correct answers, but only for collocations of higher frequency and associative strength and those of lower frequency and week associative strength, when compared to answers to all questions by subjects of lpwer proficiency. Frequency and the associative strength thus emerge as important factors for productive collocational knowledge in a wider context. Answer type analysis for tasks of receptive and productive levels To sum up, the answer type analysis shows that the percentage of correct answers is higher in receptive than productive level tasks for both CSL proficiency levels. Collocational frequency appeared to be an important factor since the highest percentage of correct answers (80%) in receptive level tasks was noticed for collocations of higher frequency and for higher CSL proficiency level. It can be sad that, considering collocational frequency, the knowledge of noun collocations is better on higher than on lower CSL proficiency level. In all productive level tasks (A, D, C), the lexical answer category, namely collocations out of context (KIK) dominates, which signals to already acquired lexical units that can easily be combined with the given base in order to specify them more precisely in a given context. The usage of neologisms signals to acquisition of formative morphemes and the fact that non-native speakers are able to form new lexical items.

Actions (login required)

View Item