Regional Variation in Speech Rate in American English from YouTube Videos

class: inverse, center, middle
background-image: url(http://cc.oulu.fi/~scoats/oululogoRedTransparent.png);
background-repeat: no-repeat;
background-size: 80px 57px;
background-position:right top;
exclude: true

---

<br>

## Regional Variation in Speech Rate in American English from YouTube Videos

### Steven Coats

English Philology, University of Oulu, Finland<br>
<a href="mailto:steven.coats@oulu.fi">steven.coats@oulu.fi</a><br>

RDHum Conference, Oulu<br> 
August 14th, 2019<br>

---

<div class="my-footer"><span>Steven Coats  
&emsp;&emsp;&emsp;&emsp;&emsp;&emsp;
&emsp;Regional Variation in Speech Rate in American English from YouTube Videos | RDHum, Oulu</span></div>

---

<div class="my-footer"><span>Steven Coats  
&emsp;&emsp;&emsp;&emsp;&emsp;&emsp;
&emsp;Regional Variation in Speech Rate in American English from YouTube Videos | RDHum, Oulu</span></div>

## Outline

1. Corpus sociophonetics, speaking and articulation rate, previous work, research questions

2. Data collection from YouTube, transcript files, corpus creation

3. Calculation of articulation rate

4. Spatial autocorrelation: Getis-Ord G<span class='supsub'><sup>*</sup><sub>i</sub></span> for regional analysis, urban-rural diffences

5. Caveats, summary, future outlook

- New method for the calculation of articulation rate from automatic speech-to-text transcripts
- Investigation of articulation rate vs. location, articulation rate vs. locality population
- Mapping with local autocorrelation statistics

---

<div class="my-footer"><span>Steven Coats  
&emsp;&emsp;&emsp;&emsp;&emsp;&emsp;
&emsp;Regional Variation in Speech Rate in American English from YouTube Videos | RDHum, Oulu</span></div>

### Corpus sociophonetics

Prosodic features increasingly being considered as bearing indexicality in the same manner as (e.g.) lexical or grammatical variables .small[(Ray & Zahn, 1999; Kendall, 2013)]

Attitudes about regional or urban-rural differences in speech temporality are common in the U.S. .small[(Preston, 1989, 1999; Roach, 1998)]

Faster speech can be associated with

- Competence, intelligence, and expertise .small[(Smith et al., 1975; Street & Brady, 1982; Thakerar & Giles, 1981)]

- Persuasiveness .small[(Apple et al., 1979; Giles & Powesland, 1975, Miller et al., 1976)]

- Attractiveness .small[(Street et al., 1983)]</br>

compared to slower speech

---

### Example video (slow talker)

---

### Example video (fast talker)

---

### Speaking rate and articulation rate

*Speaking rate*: Sum of units of speech (e.g. phones, syllables, or words) divided by total utterance time

*Articulation rate*: Sum of units of speech divided by total utterance time, **omitting pauses between segments of unbroken speech**

- Pause duration has been shown to vary .small[(Goldman-Eisler, 1961)], also according to demographic and regional parameters .small[(Clopper & Smiljanic, 2011, 2015)]

- In this study **articulation rate**, measured in σ/sec., is compared

---

### Factors that can affect articulation rate

- Type of speech: Reading, monologue, conversation

- Conversation: Interlocutor familiarity, topic under discussion .small[(Yuan, Liberman & Cieri, 2006)]

- Utterance-internal considerations .small[(Byrd & Saltzman, 1998; Yuan, Liberman & Cieri, 2006; Oller, 1973)]

- Anatomical, physiological, or neurological parameters .small[(Tsao & Weismer, 1997; Tsao, Weismer & Iqbal, 2006)]

- Demographic, social, or **regional** identity .small[(Byrd, 1992, 1994; Jacewicz et al., 2009, 2010; Kendall, 2014)]

---

### Previous work on regional variation in speech rate in the US

.small[630 speakers reading 2 short sentences. Speaking rate measured. 
Results: **North > South** .small[(Byrd, 1992, 1994).]
Problems: Locations of speakers not noted, only regional affiliation, non-naturalistic speech, small sample size.

92 speakers from Wisconsin and North Carolina reading test sentences and producing  spontaneous speech. Articulation rate measured. Results: **Wisconsin > N. Carolina**. .small[(Jacewicz et al., 2009, 2010)]. Problems: Small sample size, regional inferences based on two locations.

42 young adults from Tennessee, New York State, and Nevada read a 266-word passage. Articulation rate measured. **Nevada > New York > Tennessee** .small[(Kendall, 2013)]. Problems: Small sample size, non-naturalistic speech, regional inferences based on three locations.

159 speakers, 30,136 utterances from sociolinguistic interviews. Articulation rate measured. **Texas, Ohio > North Carolina, Washington D.C.** .small[(Kendall, 2013)]. Problems: Regional inferences based on four states.

60 undergraduate students. Articulation rate measured. **New England > Midwest > South** .small[(Clopper & Pisoni, 2006; Clopper & Smiljanic, 2011, 2015)]. Problems: Regions based on small samples.]

Trend evident, but

- Low granularity
- Non-naturalistic data
- Small samples

---

### Research questions

- Can we confirm the previous inferences on the basis of much larger data sets?

- Are there differences in the temporality of urban/rural speech in the United States?

---

### Data source

- Naturalistic language data on YouTube

- Beginning in 2009, English-language videos accompanied by automatically generated speech-to-text captions  .small[(Google, 2009)] with individual word timestamps

- Recent advances in neural-network-based speech-to-text transcription increase transcript accuracy .small[(Chiu et al., 2018)]

- High audio fidelity, standard language = accurate transcript = accurate word timings

- Word timings can be leveraged to calculate articulation rate

---

### Example video

---

### .vtt file

![](vtt_example_Bellevue.png)

---

### Focus on US local government/civic organization meetings

Public meetings of elected representatives at town/city/county/state level: advantages in terms of representativeness and comparability

- Speaker place of residence (cf. videos collected based on place-name search alone)

- Topical contents comparable

- Communicative contexts comparable

- Audio quality often high

---

### Data collection and corpus

Script to search YouTube API for channels:
- Substrings `county of`, `city of`, `municipal`, `town meeting`, `city council`, `county supervisors`, `board of supervisors`, `government`, and `official government` + names/abbreviations 50 U.S. states or names of the 312 municipalities and 100 counties by population in the United States + corresponding state names/abbreviation
- E.g. `county of Alabama`, `city council CA`, `official government Chicago, Illinois`, `official government Los Angeles County, California`
- 1,680 channel matches, remove false positives and duplicates → 579 channels
- Download all captions files (53,743) in .vtt format
- Script to extract text and timings and assign exact latitude-longitude coordinates
- Filtering: Remove channels with aggregate transcripts < 1,000 words, geographically delimit to 48 contiguous states
- 49,345 videos; 28,166.77 hours of video; 235,824,795 words .small[(Coats, 2019)]

---

### Calculation of articulation rate

- Words and their timing metadata are arranged sequentially in a captions block

- Timings within a block do not overlap

- If speech is continuous, approximate word durations (and articulation rate) can be calculated by subtracting a word's start time from that of the following word.

&emsp;

BUT: the annotation does not explicitly indicate word ending, so silences within utterances or between utterances by different speakers = loooooong words

AND: Utterance-initial words and those after long pauses are assigned a length of one second

---

### Calculation of articulation rate 2

`it's<00:17:31.820> a<00:17:31.940> 3:3<00:17:32.660> vote<00:17:49.000> because<00:17:50.000> it's<00:17:50.150> a<00:17:50.240> personnel`]

---

### Calculation of articulation rate 3

Word          | Start Time    | End Time     | Duration     
------------- | ------------- | ------------ | ------------ 
it's          | 17:30.820  | 17:31.820 | 1.000 
a             | 17:31.820  | 17:31.940 | .120 
3:3           | 17:31.940  | 17:32.660 | .720 
vote          | 17:32.660  | 17:49.000 | 16.340 
because       | 17:49.000  | 17:50.000 | 1.000 
it's          | 17:50.000  | 17:50.150 | .150 
a             | 17:50.150  | 17:50.240 | .090 
personnel     | 17:50.240  | 17:50.680 | .440

Articulation rate: .610 σ/sec. = Not accurate

---

### Calculation of articulation rate 4

Word          | Start Time     | End Time       | Duration     
------------- | -------------- | -------------- | ------------ 
**it's**      |**17:30.820**|**17:31.820**|**1.000**
a             | 17:31.820   | 17:31.940   | .120 
3:3           | 17:31.940   | 17:32.660   | .720 
**vote**      |**17:32.660**|**17:49.000**|**16.340**
**because**   |**17:49.000**|**17:50.000**|**1.000**
it's          | 17:50.000   | 17:50.150   | .150 
a             | 17:50.150   | 17:50.240   | .090 
personnel     | 17:50.240   | 17:50.680   | .440

Articulation rate: 5.26 σ/sec. = Reasonable

---

### Calculation of articulation rate 5

- Filter out word tokens with long durations (utterance-initial words and words spoken immediately before or following longer pauses)

#### Intra-utterance Continuous Articulation Rate

- Articulation rate, in syllables per second, for all word tokens in a captions file whose sequential duration is less than 1 second

- Validation of method: Use Praat script *speechrate* .small[(De Jong & Wempe, 2009)] to calculate articulation rate directly from audio files for random sample of 20 videos from corpus, Pearson's `$r = 0.83$`

---

### Validation of method

- Count syllables in Praat using *speechrate* .small[(De Jong & Wempe, 2009)]

![](praat_cK3CXpoH0qg_new.png)

---

### Regional analysis

- For each channel location: Calculate the mean intra-utterance continuous articulation rate, based on all videos from that channel

- Use the local spatial autocorrelation statistic Getis-Ord G<span class='supsub'><sup>*</sup><sub>i</sub></span> to infer large-scale patterns of difference or similarity

- Map the variate into a Voronoi tesselation

---

### <font>Getis-Ord G<span class='supsub'><sup>*</sup><sub>i</sub></span></font>&nbsp;&nbsp;<font size='3.5'>(Ord & Getis, 1992; Getis & Ord, 1995)</font>

Local spatial autocorrelation statistic used in geography and recently in dialectology .small[(e.g. Grieve, 2016)]

For each point in spatially distributed data: Positive value ➜ in a cluster of high values, negative value ➜ in a cluster of low values, zero ➜ not in a cluster

`$$G_i^* = \frac{\sum_{j=1}^{n} w_{ij}x_j- \bar{x}\sum_{j=1}^{n} w_{ij}}{{\sqrt{\frac{\sum_{j=1}^{n} x_j^2} {n}-\bar{x}^2}}\sqrt{\frac{n\sum_{j=1}^{n} w_{ij}^2 -{(\sum_{j=1}^{n} w_{ij})^2}}{n-1}}}$$`

.small[
`$n$` = number of locations, `$x_{j}$` =  value of variable at location `$j$`, `$w_{ij}$` = value of spatial weights matrix for locations `$i$` and `$j$`, `$\bar{x}$` = mean of `$x$` at all locations
]

Result is a standard deviate (significant at `$p = 0.05$` for `$G_i^*\geq\pm1.645$`)

.small[
- **Spatial weights matrix** can be binary based on choropleth contiguity, nearest neighbors, or a cutoff distance, or continuous based on inverse distance or some other function]

---

### Varying spatial weights matrices

- Binary matrices based on choropleth contiguity, fewer nearest neighbors, or a small cutoff distance tend to reveal **more local patterns**, as do continuous matrices based on small inverse distances

- Binary matrices based on more nearest neighbors or a larger cutoff distance tend to reveal **broader, supra-regional patterns**, as do continuous matrices based on large inverse distances

---

<div class="my-footer"><span>Steven Coats  
&emsp;&emsp;&emsp;&emsp;&emsp;&emsp;
&emsp;Regional Variation in Speech Rate in American English from YouTube Videos | RDHum, Oulu</span></div>

### Regional analysis .verysmall[https://stcoats.github.io/artic_rate_new.html]

<div class="midcenter">
<iframe src="https://cc.oulu.fi/~scoats/artic_rate_rdhum.html"
    
    sandbox="allow-same-origin allow-scripts"
    width="960px"
    height="550px"
    scrolling="no"
    seamless="seamless"
    frameborder="0"
    align="top">
</iframe>
.vsup[(Weights matrices: polygon continuity binary; 5, 10, 25, and 50 nearest-neighbor binary; and inverse distance with cutoffs of 200km and 100km)]
</div>

---

### Regional findings

- Spatial autocorrelation analysis suggests lower articulation rates in the South (Mississippi, Alabama, Tennessee) and higher rates in the Upper Midwest (Wisconsin, Minnesota, Dakotas), the Mountain West, and parts of Florida

---

### Urban-rural difference
.small[
- Script to extract place names associated with the latitude-longitude coordinates for each channel, population estimates in data from the U.S. Census Bureau (U.S. Census Bureau, 2017) 
- Linear regression of articulation rate and log population]

---

### Caveats

- Many different video genres represented in channels (not only meetings)
- Meetings of local government not representative of speech in general?
- Measure dependent on quality of transcript ( = quality of audio)
- Large degree of variation within small regions (patterns only emerge using spatial autocorrelation statistic)

---

### Summary and outlook

- Large corpus of automatic speech-to-text transcripts from YouTube channels of local governments
- New method to calculate articulation rate from word timing information
- Spatial autocorrelation/visualization shows that
  - Southerners speak slightly more slowly
  - People in cities speak slightly faster

--
***

- Indexicality of articulation rate
  - Look in the transcripts for lexical items known to index regional identity (dialect words), regress frequencies with articulation rate
- Pause frequency and duration analysis
- Automatic identification of higher-quality (more accurate) transcripts
- Automatic annotation of speaker variables?

---

#Thank you!

---

### References

Apple, W., Streeter, L. A., & Krauss, R. M. 1979. Effects of pitch and speech rate on personal attributions. *Journal of Personality and Social Psychology* 37, 715–27.

Byrd, D. 1992. Preliminary results on speaker-dependent variation in the TIMIT database. *Journal of the Acoustical Society of America* 92, 593–596.

Byrd, D. 1994. Relations of sex and dialect to reduction. *Speech Communication* 15, 39–54.

Byrd, D., & Saltzman, E. 1998. Intragestural dynamics of multiple phrasal boundaries. *Journal of Phonetics* 26, 173–199.

Chiu, C.-C., Sainath, T., Wu, Y., Prabhavalkar, R., Nguyen, P., Chen, Z., Kannan, A., Weiss, R. J., Rao, K., Gonina, E., Jaitly, N., Li, B., Chorowski, J., & Bacchiani, M. 2018. State-of-the-art speech recognition with sequence-to-sequence models. [arXiv:1712.01769v6 [cs.CL]](https://arxiv.org/pdf/1712.01769.pdf).

Clopper, C. G., & Pisoni, D. B. 2006. The Nationwide Speech Project: A new corpus of American English dialects. *Speech Communication* 48, 633–644.

Clopper, C. G., & Smiljanic, R. 2011. Effects of gender and regional dialect on prosodic patterns in American English. *Journal of Phonetics* 39, 237–245.

Clopper, C. G., & Smiljanic, R. 2015. Regional variation in temporal organization in American English. *Journal of Phonetics* 49, 1–15.

Coats, S. 2019. A Corpus of regional American language from YouTube. In C. Navarretta et al. (eds.), *Proceedings of the 4th Digital Humanities in the Nordic Countries Conference, Copenhagen, Denmark, March 6–8, 2019*. Aachen, Germany: CEUR, 79–91.

De Jong, N.H., & Wempe, T. 2009. Praat script to detect syllable nuclei and measure speech rate automatically. *Behavior research methods* 41(2), 385–390.

Getis, A., & Ord, J. K. 1992. The Analysis of Spatial Association by Use of Distance Statistics. *Geographical Analysis* 24(7), 189–206.

Giles, H., & Powesland, P. 1975. *Speech Style and Social Evaluation*. London/New York: Academic Press.

Goldman-Eisler,  F. 1961. The  significance  of  changes in  the  rate  of  articulation. *Language  and Speech*  4(4), 171–174.

Google. 2009. [Automatic captions in YouTube](https://googleblog.blogspot.com/2009/11/automatic-captions-in-youtube.html).

Grieve, J. 2016. *Regional variation in written American English*. Cambridge, UK: Cambridge University Press.

Jacewicz, E., Fox, R. A., O'Neill, C., & Salmons, J. 2009. Articulation rate across dialect, age, and gender. *Language Variation and Change* 21, 233–256

Jacewicz, E., Fox, R. A., & Wei, L. 2010. Between-speaker and within-speaker variation in speech tempo of American English. *Journal of the Acoustical Society of America* 128(2): 839–50.

Kendall, T. 2013. *Speech rate, pause, and sociolinguistic variation: Studies in corpus sociophonetics*. London: Palgrave-Macmillan.

]]

---

### References II

Miller, N., Maruyama, G., Beaber, R. J., & Valone, K. 1976. Speed of speech and persuasion. *Journal of Personality and Social Psychology* 34, 615–25.

Oller, D. K. 1973. The effect of position in utterance on speech segment duration in English. *Journal of the Acoustical Society of America* 54, 1235–1247.

Ord, J. K., & Getis, A. 1995. Local spatial autocorrelation statistics: Distributional issues and application. *Geographical Analysis* 27(4), 286–306.

Preston, D. 1989. *Perceptual dialectology: Nonlinguists' views of areal linguistics*. Dordrecht: Foris.

Preston, D. 1999. A language attitude approach to the perception of regional variation. In D. Preston (ed.), *The handbook of perceptual dialectology, vol. 1*. Amsterdam: John Benjamins, 359–73.

Ray, G., & Zahn, C. 1990. Regional speech rates in the United States: a preliminary analysis. *Communication Research Reports* 7, 34–7.

Ray, G., & Zahn, C. 1999. Language attitudes and speech behavior: New Zealand English and Standard American English. *Journal of Language and Social Psychology* 18(3), 310–319.

Roach, P. 1998. Myth 18: Some languages are spoken more quickly than others. In L. Bauer & P. Trudgill (eds.), *Language myths*. London/New York: Penguin, 150–158.

Smith, B. L., Brown, B., Strong, W. J., & Rencher, A. C. 1975. Effects of speech rate on personality perception. *Language and Speech* 18(2), 145–52.

Street, R. L., Jr., & Brady, R. M. 1982. Speech rate acceptance ranges as a function of evaluative domain, listener speech rate, and communication context. *Communication Monographs* 49(4), 290–308.

Street, R. L., Jr., Brady, R. M., & Putman, W. B. 1983. The influence of speech rate stereotypes and rate similarity on listeners' evaluations of speakers. *Journal of Language and Social Psychology* 2(1), 37–56.

Thakerar, J. N., & Giles, H. 1981. They are – so they speak: Noncontent speech stereotypes. *Language and Communication* 1, 251–256.

Tsao, Y.-C., & Weismer, G. 1997. Interspeaker variation in habitual speaking rate: Evidence for a neuromuscular component. *Journal of Speech, Language, and Hearing Research* 40, 858–866.

Tsao, Y.-C., Weismer, G., & Iqbal, K. 2006. Interspeaker variation in habitual speaking rate: Additional evidence. *Journal of Speech, Language, and Hearing Research* 49, 1156–1164.

United States Census Bureau. 2017. Subcounty Resident Population Estimates: April 1, 2010 to July 1, 2017. [Data set]. https://www2.census.gov/programs-surveys/popest/datasets/2010-2017/cities/totals/sub-est2017_all.csv

Yuan, J., Cieri, C., & Liberman, M. 2006. Towards an integrated understanding of speaking rate in conversation. *Proceedings of Interspeech 2006, Pittsburgh, PA*, 541–544.

]]