Language Variation, Register Analysis, and Corpus Linguistics Study of Communication, and Principles of Corpus Linguistics

Introduction to Language Variation and Register Analysis

• Overview of Topics: The discussion covers two primary areas: language variation and registers, followed by corpus linguistics, specifically focusing on learner corpora and register-based research applications.

• Evolution of Perspective: Traditional sociolinguistic education often focuses on linguistic variation due to social factors (gender, status). Bill (the speaker) introduces register analysis as a wider social perspective on variation, emphasizing socio-situational variables over fixed social categories.

• Defining Language: Language is categorized into two main components:

  • Structure: Includes the transmission (phonetics/phonology) and core knowledge of linguistic structure (morphology, syntax).

  • Use: Includes variation over time (diachronic change), social variation (gender, socioeconomic status), regional variation (dialectal differences), and personal variation (individual style within a community).

Subfields of Linguistics: Structure and Knowledge

• Phonology (Structure/Transmission): Knowledge of sounds and rules.

  • Example: Pluralization rules in English (voicing assimilation).

    • $bats$ ends in voiceless morpheme $/s/$ due to the voiceless consonant $/t/$.

    • $bags$ ends in voiced $/z/$ due to the voiced consonant $/g/$ ($eggs$ follows this rule).

    • $buses$ adds $/iz/$ because adding an $/s/$ sound to a fricative requires a vowel.

  • Nonsense Word Test: Speakers apply these rules to words they have never heard ($one blit, two blitz$; $one blag, two blags$; $one bluss, two blusses$), proving that phonology is a core knowledge set rather than just memorization.

• Morphology (Core Knowledge): Knowledge of meaningful units.

  • Example: The prefix $un-$ changes semantic category (e.g., $happy$ vs. $unhappy$).

  • Derivational Morphology: Morphemes can change word classes (e.g., verb $agree$ to noun $agreement$ or adjective $agreeable$).

• Syntax (Core Knowledge): Knowledge of word order.

  • Constraints: Sentences such as "inevitable the of was city destruction of" are rejected because they violate syntactic order.

  • Representation: Knowledge is represented through tree diagrams, illustrating how articles and nouns form noun phrases ($NP$), which combine with prepositional phrases ($PP$) to form larger structures.

• Semantics (Word Meaning): Explores the arbitrary relationship between sounds and objects.

  • Arbitrariness: The sound "rock" has no inherent connection to the object; it is an arbitrary societal agreement.

  • Connotations vs. Denotations: Words like $cheap$ and $thrifty$ denote the same behavior (avoiding spending) but have different connotations (negative vs. positive).

• Pragmatics (Structure/Event Context): The relationship between language and context necessary to make meaning.

  • Social Expectations: A four-year-old calling their boss to say they are late is pragmatically strange because the context lacks the necessary role infrastructure.

  • Communicative Appropriateness: Choosing between "Hey Bill, give me more time" and "I was wondering if I could have an extension" is a pragmatic choice based on formality and politeness.

  • Speech Acts: Pragmatics governs how we perform specific acts (apologies, refusals, invitations). For example, a stranger bumping into someone might just say "I'm sorry," whereas a friend who forgot a commitment may need a longer, more detailed apology to be pragmatically appropriate.

Traditional Sociolinguistics vs. Register Analysis

• Traditional Sociolinguistics: Traditionally looks at variation by factors such as:

  • Region: Differences in naming objects (e.g., $sofa$, $couch$, or $Davenport$; $bag$ vs. $sack$).

  • Example (Thai Context): Papaya salad is called $Som Tum$ generally, but has different names in regions like Chiang Mai.

  • Phonological Variation: Regional differences in vowel merges/distinctions, such as $caught$ vs. $cot$, or the pronunciation of $New York$ vs. "New Yawk."

  • Social Factors: Time (age-based), Socioeconomic Status ($SES$), and Gender.

• Register Analysis: Variation based on situational characteristics. Register is defined as a variety of language characterized by its context and the pervasive linguistic features used within it.

  • Key Difference: While traditional sociolinguistics focuses on a single variable (e.g., word choice for "soda" vs. "pop"), register analysis looks at "pervasive linguistic features" (frequent, co-occurring features) that are functionally motivated by the situation.

• Socio-Situational Variables (The Register Framework):

  • Participants: Speaker, hearer, writer, reader, and onlookers.

  • Relationships: Degree of interaction (dialogic vs. monologic), social goals, and shared knowledge (e.g., linguists talking to other linguists).

  • Channel: Mode of production (speech, writing, gestures) and permanence (fleeting vs. recorded).

  • Production Circumstances: Real-time/unplanned vs. edited/planned.

  • Setting: Public vs. private; shared time/space vs. separation (e.g., email).

  • Communicative Purpose: To inform, entertain, argue, narrate, or persuade.

Comparative Registers: Linguistic Evidence

• Case Study: Emails:

  • Email to Boss: Long, full sentences, formal greetings ("I hope my email finds you well"), complex syntax ($therefore$), and lack of contractions/emojis.

  • Email to Friend: Slang/rhymes ("What's shaking, bacon?"), contractions ($it's$), emojis, exclamation points, and elliptical sentences.

• Case Study: News Writing vs. News Talk Show:

  • News Talk Show: Highly interactive, featuring a massive frequency of personal pronouns ($I, you, we, they$). Includes fillers like "you know" and shorter, fragmented sentences.

  • News Writing: High density of nouns (informational focus) and longer noun phrases. Pronouns are rare except in reported speech. Verbs are more specific ($hikes$, $injects$).

• Case Study: Research Article vs. Academic Lecture:

  • Academic Lecture: Encourages interaction through "stance" markers (e.g., "this might be boring for you"). Uses many pronouns and focuses on engaging the audience.

  • Research Article: Dense, noun-heavy language for information density. One whole noun phrase can be roughly $20$ words long (e.g., "An extensive literature on agriculture in Roman times which maintained a preeminent position until comparatively recently").

Genre, Register, and Style

• Genre: Identifying a text type through specific conventions or markers, even if they occur only once (e.g., "Dear Kim" or "Sincerely").

• Register: Focuses on frequent and pervasive linguistic features that co-occur throughout the whole text due to situational constraints.

• Style: Individual variation within a register. A speaker's personality or specific emotional state affects their language choices (e.g., a mother using different language for an order based on being in a "good mood").

Introduction to Corpus Linguistics

• Definition: A method (not a theory) for analyzing language use by collecting large bodies of electronically stored text ($corpora$) analyzed with computer programs.

• Text vs. Corpus:

  • Text: Read horizontally (left to right), for content and meaning, as a single communicative event.

  • Corpus: Read vertically (fragmented patterns across many texts), for frequent patterns/concordances.

• Key Terms:

  • Keyword in Context (KWIC): Seeing a word (e.g., "that") with its surrounding fragments to identify usage patterns.

  • Corpus-based: Research with a pre-determined idea/feature to look for.

  • Corpus-driven: Using the computer to find what is frequent without pre-conceptions.

• Limitations: Corpora show what people do say, but not what is impossible (ungrammatical) in a language; they provide evidence, not explanations.

Case Studies in Learner Corpora

• Usage of "That" in L1 vs. L2 Argumentative Essays:

  • Findings: L2 writers used the demonstrative pronoun "that" ($200$ times per $100,000$ words) much more frequently than L1 writers ($150$ times per $100,000$ words).

  • Interpretation: This may reflect writing experience rather than native language background.

• Functional Analysis of "Now":

  • Functions: Present time, transition marker, and discourse marker.

  • Results: L1 and L2 student writers used the "present time" function almost identical amounts. Professional academic writers, however, used "now" as a transition marker much more frequently ($process-focused development$).

• Collaboration in Speaking Tasks:

  • Task: Two students argue for whether "Amelia" or "Joe" needs a weight loss clinic.

  • Corpus Size: $778$ tasks, $270,000$ words.

  • Proficiency Levels (TOEFL Equivalent):

    • Level 1: $32-44$

    • Level 2: $45-56$

    • Level 3: $57-69$

  • High Collaboration Traits: More turns ($16.4$), fewer words per turn ($9.9$), high use of first/second person pronouns and questions. Interactive, involved register.

  • Low Collaboration Traits: Fewer turns ($9.9$), longer turns ($27.4$ words per turn), high noun density, nominalizations ($obesity$), and factual language. Informational/monologic register.

Questions & Discussion

• Q: Is texting language (textese) a register?

  • A: Yes. It is defined by situational variables: thumbs-only communication, time constraints, and a mix of interactional/informational purposes. Features like $LOL$ or $IDK$ are markers of this situational context.

• Q: Can personality affect register?

  • A: The speaker defines this as "Style" variation rather than register. Style accounts for individual decisions and perceptions (e.g., speaking differently to a father vs. a mother because of their specific moods).

• Q: Is video game interaction a register?

  • A: Yes. It involves a three-way interaction (Human-Human-Game) and high levels of shared knowledge between players, which strongly dictates the linguistic features used.

• Advice for Graduate Research:

  • Focus on the process over the product.

  • Ask questions that are interesting regardless of the finding.

  • Choose a topic you love; research is like a long trip, and if you realize mid-trip you don't love the topic, the experience is miserable.