Abstract:

Introduction and motivation:
In the paper, we present research of the most common errors in coherence occurring in learners’ essays. Coherence is one of the pillars of a comprehensible and well-structured text. A text is not simply a set of individual sentences, but a complex of sentences interlinked by a variety of text relations. Coherence may be manifested through various language means, e.g. by discourse connectives, anaphoric devices, lexical repetition etc. Reaching a well-written coherent text is often a difficult task for language learners.

Methods and material:
We carry out corpus-based research of essays written by learners of Czech. We analyse their most common errors concerning text coherence. We focus especially on the use of discourse connectives: on their putting into a relevant context or on the most common semantic confusions. We examine whether there are some shared issues or tendencies concerning coherence errors that could be generalized into recommendations to students to which aspects they should pay attention when writing a well-structured essay.
The data come from the corpus MERLIN that contains essays written by learners of Czech, German or Italian. Concerning Czech, it contains 442 texts across the language levels defined by CEFR (Common European Framework of Reference for Languages). The corpus is annotated on several language aspects, including coherence and connective accuracy. We put under scrutiny all cases with problematic uses of connectives in the whole Czech part. We analyse all these cases in detail and divide them according to the specific language aspects they violate.

Results:
Errors in using connectives by language learners fall into three basic categories:
i) a missing connective in a place requiring a connective,
ii) an overuse of connectives (redundant, sometimes ungrammatical uses), and
iii) a wrong connective use (connective confusion). The most frequent as well as linguistically and pedagogically relevant is the last category.

The wrong uses of connectives concern several language aspects: semantic confusion, reversed direction of a discourse relation (e.g. “protože” ‘because’ vs. “proto” ‘therefore’), semantic nuances of connectives expressing the same / similar semantic category, confusion of potential synonyms used in a semantically inappropriate context, wrong uses of complex (multiword) connectives and correlative pairs, confusion of connectives with a similar form (but a different meaning), or uses of connectives in a wrong sentence position (word order).

Discussion:
In the next step, we introduce how learners of Czech may improve their writing skills using an online tool EVALD (Evaluator of Discourse). EVALD provides an automated essay scoring and gives also feedback to the users – it gives the strong and weak aspects to the text in various language fields including text coherence and discourse connectives. EVALD is a suitable tool for e-learning which can be used not only by students (language learners) but also by teachers as an assistant tool in their classes.

Keywords:

Language acquisition, e-learning, text, discourse connectives, error analysis.