Corpus Linguistics

Today's Topics

1. Corpus-based Analysis [1]
2. Seminar presentation
3. [Lab] GATE platform

Corpus Annotation and Corpus-based Linguistic Analysis

Corpus Annotation is NOT the must

• Annotation schema presumes a prior linguistic analysis and studies.
• To be consistent and systematic, annotations are shallower/less interesting.

Corpus Annotation and Corpus-based Linguistic Analysis

Corpus Annotation is the need when no choice.

• allows to test linguistic hypotheses empirically.
• provides a pool for diffirent machine learning tasks (segmentation, categorization, relation discovery).

Corpus-based Empirical Methods

• Fundamental commitmment to empiricism
• Both corpus linguistics and other experimental linguistics study language system not directly but by observation of epiphenomena - output on the large scale, or either the blood-flow requirements or some other psysiological feature associated with it.

Corpus-based Empirical Methods

Distribution : toward Unified Empirical Linguistics [1], where evidence of all kinds - textual, psychological and neurological - is a matter of course used in concert to uncover the nature of language. In such context, corpus linguistics will reach its full potential as a methodology.

Review: A chain of works corpus data science involves

1. Pre-processing (cleaning, tokenizing, segmentation, etc)
2. Data annotation (Semi-automatic) Labeling (POS tagging) and Management
3. Exploratory Data Analysis (with workable knowledge of Statistics)
4. Hypothesis testing
5. Prediction and Statistical Modeling, etc
6. Presentation and Web application (Demo: Shiny-LexicoR)

Corpus-based Analysis: Basics

A corpus does not contain new information about language, but the software offers us a new perspective on the familiar.

• Two related processes: the production of frequency lists (either in rank order, sorted alphabetically, or according to keyness) and the generation of concordances (examples of particular items in context) -> Note Bene.: In order to compare frequency counts across corpora of different sizes, a process of normalisation is required.

Corpus-based Analysis: Basics

Now, what can concordance/frequency reveal the unknown?

Only 4 lines (lines 5, 13–15) is now clause-medial, and acting as a temporal adverb ; with remaining clause-initial and has discourse-level functions, signalling a change of focus or topic !!!

Corpus-based Analysis: Pattern recognition

• Patterns: A pattern is essentially repetition. In language, pattern is observed when words, sounds, rhythms or structures are repeated.
• The identification of pattern in a corpus implies a connection between theory, method and technique.

Digging out patterns: An Example

the black experience (taken from Obama’s speech from a 450-million-word Bank of English corpus.)

• If we search for the followed by experience we find that one set of words occurring frequently between those two refers to groups of people: American, black, human, undergraduate, customer and British.
• Replacing the by a or an gives very different results: the word preceding experience usually expresses an affective response to the experience: great, new, wonderful, bad, traumatic and pleasant.

Digging out patterns: An Example

• In short, there is a repeated phrase the black experience, but there is also a more frequently repeated pattern: the + group of people + experience !!
• You may also check Google Book and Google Web Search

Why are patterns difficult to spot?

1. repetition in naturally occurring conversation is transient, fleeting; it may have no perceptible effect, or its effects may not be ascribed to the repetition itself.
2. patterning involves the repetition of ‘things’, but those ‘things’ may be of many different kinds.

Collocations

Patterns can also be interpreted as a co-occurrence of a language form and a particular context.

collocation system demo Just-the-word

Semantic prosody and semantic preference

• Semantic prosody (a.k.a. discourse prosody) may be understood as a concept related to that of connotation in more traditional approaches to semantics.
• So for instance, the lemma 'happen' occurs more frequently than would be expected by chance alone with subjects that can be evaluated as negative.

Semantic prosody and semantic preference

• One long-standing problem with the notion of semantic prosody is discriminating it from the phenomenon of semantic preference.
• Stubbs (2001: 65–6) defines semantic preference as 'the relation, not between individual words, but between a lemma or word-form and a set of semantically related words' and notes that there is a fuzzy boundary between the two concepts.

Corpus-based Studies: Dimensions

1. 功能 Functional Linguistics
2. 計算 Computational Linguistics
3. 歷程 Psycholinguistics and Language Acquisition
4. 應用 Lexicography and Language Teaching

Functional Linguistics

Computational Linguistics

Psycholinguistics and Language Acquisition

Lexicography and Language Teaching

Homework

[1] Write an essay: How to use corpus (linguistics) to (write dictionaries | the study of health communication | language teaching and learning | ........)

[2] Annotate a text with (word sense | pos | polarity| discourse marker | named entity | ......)

Review: Annotation projects involve:

1. Definition of target phenomena and decision of annotation level
2. Drafting guidelines and annotation schemas
3. Choosing annotators (Alternative)
4. Annotation tools
5. Evaluation

Lab session: Annotation tools

Task-dependent

• Textual annotation: GATE , GLOZZ, etc.
• Speech annotation: Praat
• Multimodal annotation: ELAN, ANVIL

Lab session GATE

• In the following weeks, we will practice with some corpus annotation tools: GATE and GLOZZ

---

Lab session: Some terminologies

(Prevor, 2014)

• Annotation Schema: explicit characterization of the units/categories/...that will need to be annotated, and markup language as well (e.g., XML).

• Annotation Guidelines:

  • Instruction to ensure that annotators are doing the task correctly.
  • Have to be explicit, clear without becoming lengthy (practical aspects)
  • More focussed / restricted than the schema.