Help for the search

You can search the MERLIN corpus with the help of the search and visualisation software ANNIS.

Getting started: Open the search. Choose an example search from ↘ Help/Examples to get an impression how the ANNIS search works. Now, you can modify the query. Choose and copy an annotation (see section 2) or use the ↘ Query Builder as described in section 3 to search for a specific L2 feature.

1 Explanation of your search output

1. Search field displaying your query in the query language
2. Options: export search results or perform a frequency analysis
3. Choose L2 corpus 
4. Number of tokens displayed left and right from the key word
5. Metadata, i.  e. information on the learner and the ratings, as well as statistical information
6. Detail from the learner text (L2 text) that contains the word or the feature you are looking for 
7. TH1 = minimally corrected, i. e. orthographically and grammatically acceptable version of the learner texts; TH2 = sociolinguistically acceptable version of the learner text; TH1Diff and TH2Diff = description of the deviation between the L2 text and the target hypothesis
8. Categorical description of the error or the L2 feature [EA_category] 
9. Actual manifestation of the feature/error [_type]
10. displays the complete learner text

For automatic annotations displayed under "automatic grid" (POS annotations, lemmas, t-units, sentences) see  MERLIN for research.

2 Search MERLIN for L2 features

In the following section, all annotated learner language features are listed according to their categorical description [EA_category] and specific manifestation [_type]. Copy the annotation names (tags) into the ANNIS search window to start a search.

For concrete examples of annotated learner language features, see MERLIN Annotations, and for a detailed description of annotations and their scope (tag span) and annotation rules, see the MERLIN Annotation scheme.

G_ Grammar

O_ Orthography

G_ Intelligibility**

V_ Vocabulary**

C_ Coherence/Cohesion**

S_ Sociolinguistic appropriateness**

P_ Pragmatics**

** Note: these error categories are only accessible for a subset of MERLIN texts. See MERLIN Annotations / Annotation structure.

Further specification of error categories

3 Narrowing the search using metadata

Use ANNIS's Query Builder to search for features or a combination of features while narrowing the query based on specific metadata.

1. Open the ↘ Query Builder in ANNIS. 
2. Choose ↘ Word sequences and meta information. 
3. Select the corresponding feature and its attribute under Linguistic sequence ↘ Initialize ↘ Add.
4. In the Toolbar, click on ↘ Create AQL Query to paste the query into the search field.

To restrict the query to a specific group of learners (e. g. by L1 or age) or a specific CEFR level (fair rating), select a metadata category before pasting the query into the search field (step 4) under  ↘  Meta information ↘ Add and tick the required attribute, e. g.:

Alternatively you can copy the feature you are searching for from the feature list under section 2 and paste it into the ANNIS search field. Then, add the metadata using the following scheme, to restrict your search to specific texts:

• & meta::_rating_fair_cefr="B1"  [A1, A2, B1+, B2]
• & meta::_author_L1="German"  [English, Russian, Arabic, etc.]

The ANNIS User Guide offers a thorough introduction to using the ANNIS interface.

4 Retrieve statistical information

To get an indication of the frequency of certain L2 features use the ANNIS search.

1. Search for specific L2 features as described in section 2 or use global error categories.
2. Then, click on ↘ Frequency Analysis [2] and subsequently on the right on ↘ Perform Frequency Analysis. You will retrieve a statistical analysis of the annotated features within the category in question.
3. Amend your query according to the following scheme, to restrict the search to a certain CEFR level, e. g. B1: & meta::_rating_fair_cefr="B1".

Global error categories