Ling435 Assessment

Variation in an individual’s realisation of /t/ across time and style

Task Overview

For this assignment, you will investigate how /t/ is realised by a single speaker at two distinct points in time or in two different speech styles. Your speaker should be a public figure whose speech is readily available online (e.g. interviews, public speeches, press conferences), ensuring you can access clear audio from both contexts. You might, for example, compare speech from the 1990s versus today, or compare a formal speech versus a casual interview from the same time period—or you can combine both an era and a style comparison if data is available.

Key Instructions

1. Focus on /t/ Realisation

o  Examine /t/ in word-medial or word-final positions after vowels or sonorants [n,l, r] (e.g. better, cat, want).

2. Common Variants of /t/

o T-glottalling [?] — butter ['bΛ?. Θ]

o T-flapping [Γ] — city ['sɪ.Γi]

Less Common Variants to Consider

o Unreleased /t/: common word-finally in American English [t ]

o T-glottalisation / Glottal reinforcement: [?t] — e.g. a glottally masked /t/ in

Tyneside English: city ['sɪ .?ti]

o T-to-r: an approximant-like realisation in certain words or phrases (e.g.“get off” →“gerroff”), though this only happens in certain words.

o T-frication: [ts] or another fricative-like realisation of /t/.

o Complete deletion: [∅]

o There may be more …

Ifyou have just a couple of less common variants cropping up,you can exclude them from your analysis andfocus on the main two variants.

3.   Beyond the envelope of variation:

o  Exclude /t/ in stressed syllable-initial position (e.g. attack) in your analysis, as stressed /t/ typically does not reduce.

o  Do not include cases of /t, d/ deletion in clusters (e.g.“best man”→“bes’ man”). This is a separate variable.

4. Dependent and Independent Variables

o Dependent variable: the realisation (or variant) of /t/.

o Independent variables:

Recording date (e.g. 1990s vs. 2020s) or style (e.g. formal vs. informal interview).

Position of /t/ in the word or phrase (word-medial, word-final, preceding a pause, preceding another word).

Following sound

Other factors you notice might be conditioning the variation (e.g. vowel length, morphological or syntactic class, preceding segment).

5. Choosing the Speaker

o  Avoid a speaker who shows very little or no variability. For instance, some

American speakers may flap /t/ in most positions, leaving almost no variation to analyse.

o  Ensure you can find at least two recordings or speech contexts with clear audio and enough tokens of /t/.

6. Data Collection Guidelines

o  Aim for enough tokens of /t/ to support a quantitative analysis—ideally, at least 30-40 tokens in each context.

o  Recall the principle of accountability: we include all variants of /t/.

o  Ideally, 10-15 minutes of continuous speech might yield a sufficient number of tokens for each context. If you find the speaker uses /t/ less frequently, you may need more time, but if you have lots of tokens, 5 minutes may even work.

o 20 minutes in each context might be more than necessary; you can sample shorter segments if you gain enough tokens.

o  Include relevant information in your methodology

Analysis and Reporting

• Data Presentation and Visualisation

o  You should produce three graphs to illustrate the variation you have found.

o  Each graph should be accompanied by a table that has the percentage of one

variant for each independent variable (e.g. casual vs. formal), as well as the overall totals.

• Notes:

o  If you have more than two variants that are productive, you can just include all variants in the table.

o  I suggest incorporating a cross-tabulation within your analysis (e.g. a table or combined graph that compares two independent variables simultaneously).

o Perform at least one statistical test (e.g. chi-square, correlation) to assess the significance of your findings.

o Graphical Guidelines:

Graph axes should typically run from 0-100% to represent the proportion or percentage of each /t/ variant clearly.

If you have only two variants in your dataset, you only need to plot the proportion of one variant (the other being the complement).

Categorical independent variables (e.g. style, region) should be displayed with bar graphs.

Numeric independent variables (e.g. year of recording, age of speaker) should be displayed with line graphs.

• Interpretation of Findings

o  Relate your results to existing literature and theoretical frameworks (e.g. research on t-glottalling, lifespan, style shifting and sound change).

o Hypothesise why your chosen speaker exhibits these patterns: consider dialect background, social identity, audience design, or any relevant factors.

o  Where possible, draw comparisons to studies of similar speakers or dialects, explaining whether your results support or challenge previous research.

• Reflection on Speaker Patterns

o  Reflect on how and why these variations may have arisen in your speaker’s repertoire.

Essay Length:

Your written report can be an absolute maximum of 5,000 words (but much shorter reports are also welcome).

Use this section as a guide to organise your results, figures, and interpretation, ensuring a clear, coherent discussion of your speaker’s /t/ realisations.

Suggested structure

Written Report:

1. Introduction:

•    Rationale for studying /t/ variation

•   Speaker background

•   Research questions

2. Literature review

•    Studies on /t/-glotalling, studies on lifespan or t-glottalling, or any relevant theory for your discussion

3. Methodology: (this can be brief)

•    Speaker selection, data sources, recording contexts

•    Transcription, coding scheme

•    Criteria for inclusion/exclusion of tokens

4. Results & discussion