Modifying Language for a Higher Goal: Investigating Quantitative Features of Apple’s Launch Event Speech from 2016 to 2022

ABSTRACT

In response to the COVID-19 pandemic, Apple Inc. switched its product launch from onsite events to pre-recorded online broadcasts. Yet this form has been carried on even though the pandemic has faded. It is unclear if this shift was accompanied by a move in language complexity and how the communication effect was influenced. This study analysed Apple’s product launch events from 2016 to 2022 using quantitative linguistics metrics such as lexical complexity, word token length, entropy, readability, and sentence length distribution. The analysis adopted the reach-effect-durability framework for interpreting the effects of strategic communication. The results showed that Apple’s online product launch events featured richer and longer words, increased text difficulty, and a decrease in the proportion of short sentences. Such changes happened in a balanced manner that can be interpreted in view of the driving and restraining forces resulting from the mode shift of the launch event. This study demonstrates the feasibility of assessing efforts and strategies used in a communication campaign through quantitative text analysis.

Acknowledgments

We sincerely thank the anonymous reviewers and the editorial team for helping us improve this work.

Disclosure Statement

No potential conflict of interest was reported by the author(s).

Notes

1. The iPhone release events during 2020 and 2022 (Oct 14, 2020, Sept 15, 2021, and Sept 8, 2022) had an average of more than 3,840 million views on YouTube: https://www.youtube.com/@Apple/streams.

2. https://www.businessofapps.com/data/apple-statistics/.

3. Get supplements only for academic use: https://github.com/eddiezhou99/supplements-for-the-strategic-communication-study.

4. The data presented herein were accessed on Mar 13, 2023, by querying search indices over the past five years in the United States for iPhone, Mac, iPad, and Apple Watch, separately treated as independent thematic keywords, via https://trends.google.com/.

5. On YouTube, if the channel uploads subtitles (sometimes in several versions of different languages), the viewers can choose the subtitle by clicking the setting button. If no subtitle was uploaded, YouTube also provides an automatically generated version of the subtitle that lacks punctuation. Both the prepared and automatically generated subtitles can be downloaded by pasting the video URL to a third-party tool: https://downsub.com/.

6. In 2020, the new generation of iPhone was launched in October. As we aim to make a longitudinal comparison of the introduction of this product line in priority over the years, we chose the launch event in October rather than September 2020. In this table, duration is for the whole video. It does not exactly measure the length of the speech since the product launch is often aided with other materials (ads videos, customers’ storytelling, etc.).

7. To see how Wordless calculates Token Length in Syllables, refer to the code: https://github.com/BLKSerene/Wordless/blob/3.4.0/wordless/wl_nlp/wl_syl_tokenization.py.

8. Z_skewness (=$\frac{\mathit{Skew}\mathit{value}}{\mathit{S}{\mathit{E}}_{\mathit{skewness}}}$) and Z_kurtosis (=$\frac{\mathit{Kurtosis}\mathit{value}}{\mathit{S}{\mathit{E}}_{\mathit{kurtosis}}}$) are the key indices for z-tests under the null hypothesis of normal distribution. For small samples ($\mathit{n}<50$), as in this study, the absolute z-scores no larger than 1.96 stand for the acceptability of the null hypothesis (Field, 2017, p. 443). When executing the independent samples t-test, offline keynote speeches (2016–2019) subtitles were set as Group 1, the online counterparts Group 2, i.e. the Mean Difference was the indices’ Mean of Group 1 minus that of Group 2. Same for Table 3.

9. The launch event in September 2017 was excessively longer than other events (Table 1). And official subtitle was unavailable for this event. When analysing sentence length distribution, we decided to leave out this event to exclude the possible accumulated error of manually added punctuation.

10. Significance codes: 0.05*, 0.01**.

Additional information

Funding

The work was partly supported by the MOE Project of Key Research Institute of Humanities and Social Sciences at Universities in China, “Data-driven Studies of the Development of Foreign Language Capacity” [22JJD740018].