Saturday, August 17, 2019

Readability in reporting statistics

Within the past 20 years there has been a bit of a (r)evolution in the quantitative methods used in the speech sciences and linguistics. A renewed focus on experimental research in linguistics and the development of laboratory phonology as a field have contributed to this development. Though phonetics has always been an experimental field, it too has benefitted from a renewed interest in quantitative methods. The availability of free and powerful statistical analysis software, such as R, has improved access to tools. Finally, several books focusing on quantitative methods in linguistic sciences have been published, all of which improve the statistical learning curve.

Yet with any changes to a field come challenges. Since several types of linguistic data violate the assumptions of ANOVA, what method should you use instead? With several newer methods (multi-level modeling, generalized additive models, growth curve analysis, spine ANOVA, functional data analysis, Bayesian methods, etc), it is often also unclear what statistic to report. If we are concerned about replicability in our field, how do we ensure that our methods are clear enough to be replicated? And, importantly, how do we communicate these concerns to both novice and experienced researchers that might not be familiar with them? Since so many methods are new (or new to some of us), we are often tempted to include a fancier model without understanding it fully. How do we ensure we understand it enough to use it?

These issues are all very important, but we must also not lose sight of our duty as scientists to properly communicate our research. It would be great if our research could "speak for itself." It would be great if we could rely on our readers being so engaged in our results that they never got bored or frustrated reading pages and pages of statistical modeling and tests. It would great if we could assume that all readers understood the mechanics of each model too. Yet, our research seldom speaks for itself and readers can be both bored and uninformed. Unless your research findings are truly groundbreaking, you probably have to pay attention to your writing style.

I'm not an expert in writing or an expert in statistical methods. I teach a somewhat intense graduate course in quantitative methods in linguistics and have been a phonetician for about 15 years (if I include some time in grad school). My graduate education is in linguistics, not mathematical psychology or statistics. But as a researcher/phonetician I am a practitioner of statistical tests, as a reviewer I read many submitted manuscripts in phonetics, and as a professor I frequently evaluate how students talk about statistics in writing. I think that the best way to open up a discourse about how we report statistics in linguistics and whether it is readable or not is to present various strategies and to discuss their pros/cons.

I should mention that I'll be pulling examples from my own research in phonetics here as well as a few that I've seen in the literature. I am not intending to offend any particular researcher's practice. On the contrary, I feel that it's necessary to bring up some real examples in this discussion (and I've picked some good ones).

I.  The laundry list

One practice in reporting statistics is to essentially report all the effects as a list in the text itself. We've all seen this practice, but after digging for an example of it, I was happy to discover that it is not nearly as frequent as I had assumed (or perhaps we've become better writers). So, here's a made-up example:

There were significant main effects of vowel quality (F[3, 38] = 6.3, p < .001), age (F[6, 12] = 2.9, p < .01), speech style (F[2, 9] = 5.7, p < .001), and gender (F[3, 8] = 3.2, p < .01) and significant interactions of vowel quality x age (F[18, 40] = 2.7, p < .01], vowel quality x gender (F[12, 20] = 2.4, p < .05), but no significant interaction between vowel quality and gender nor between vowel quality and speech style. There were significant three-way interactions between vowel quality x gender x speech style (F[12, 120] = 2.4, p < .05) but no three-way interaction between either...  These effects are seen in the plot of the data shown in Figure 3.

Effect, stat, effect, stat, effect, stat, repeat. It almost sounds like an exercise routine. On the one hand, this method of reporting statistics is comprehensive - all our effects are reported to the reader. We also avoid the issue of tabling your statistics (more on this below). Yet, it reads like a laundry list and a reader can quickly forget (a) which effect to pay attention to and (b) what each effect means in the context of the hypothesis being explored.

If the research involves just one or two multivariate models for an entire experiment, the researcher might be forgiven for writing this way, but now let's pretend that there are eight models and you are reading the sample paragraph above eight times within the results section of a paper. Then you go on to experiments 2 and 3 and read the same type of results section two more times. By the end of reading the paper, you may have seen results indicating an effect or non-effect of gender x vowel quality twenty-four times. It truly becomes a slog to recall which effects are important in the complexity of the model and you might be forgiven for losing interest in the process.

There is an additional problem with the laundry list method - our effects have been comprehensively listed but the linkage between individual effects and an illustrative figure has not been established. It might be clear to the researcher, but it's the reader who needs to interpret just what a gender x vowel quality interaction looks like from the researcher's figure. Without connecting the specific statistic and the specific result, we both risk over-estimating the relevance of our particular effect in relation to our hypothesis (risking a Type I error) and we fail to guide our readers in interpreting our statistic the right way (producing either Type S or Type M errors). Our practice of reporting statistics can influence our statistical practice.

Tip #1: Connect the model's results to concrete distinctions in the data in the prose itself.

Now, just what does it look like to connect statistics to the data? and how might we easily accomplish this? To learn this, we need to examine additional methods.

II.  The interspersed method with summary statistics

If it's not already clear, I'm averse to the laundry list method. It's clear that we need to provide many statistical results to the reader, but how do we do this in a way that will engage them with the data/results? I think that one approach is to include summary statistics in the text of the results section immediately after or before the reported statistic. This has three advantages, in fact. First, the reader is immediately oriented to the effect to look for in a figure. Second, we avoid both type S and type M errors simultaneously. The sign and the magnitude of the effect are clear if we provide sample means alongside our statistic. Third, it breaks up the monotony found in a laundry list of statistical effects. Readers are less likely to forget about what the statistic means when it's tied to differences in the data.

I have been trying to practice this approach when I write. I include an excerpt from a co-authored paper here below (DiCanio et al. 2018). As a bit of background, we were investigating the effect of focus type on the production of words in Yoloxóchitl Mixtec, a Mixtec language spoken in Guerrero, Mexico. Here, we were discussing the combined effect of focus and stress on consonant duration.


The statistics reported here are t values from a linear mixed effects model using lmertest (Kuznetsova et al. 2017). The first statistic mentioned is the effect of focus type on onset duration. This effect is then immediately grounded in the quantitative differences in the data - a difference between 114 ms and 104 ms. Then, additional statistics are reported. This approach is avoiding Type S and Type M errors and it makes referring to Figure 2 rather easy. The reader knows that this is a small difference and they might not make much of it even though it is statistically significant. The second statistical effect is related to stress. Here, we see that the differences are more robust - 126 vs. 80 ms. Figure 2, which we referred the reader to above, is shown below.


While it is rather easy to get some summary statistics for one's data, what do you do when you need more complex tables of summary statistics? I generally use the ddply() function in the plyr package for R. This function allows one to quickly summarize one's data alongside the fixed effects that you are reporting in your research. Here's an example:

ddply(data.sample, .(Focus, Stress), summarize, Duration = mean(Duration, na.rm=TRUE))

For a given data sample, this will provide mean duration values for the fixed effects of focus and stress. One can specify different summary statistics (mean, sd, median, etc) and include additional fixed effects. While this may seem rather trivial here (it's just a 2x2 design after all), it ends up being crucially useful for larger multivariate models where there are 2-way and 3-way interactions. If each factor includes more than four levels, a two-way or three-way interaction can become harder to interpret. Leaving this interpretation open to the reader is problematic.

Now, for the person in the room saying "don't post-hoc tests address this?" I would point out that many of the statistical tests that linguists have been using more recently are less amenable to traditional post-hoc tests. (Is there an equivalent to Tukey's HSD for different multi-level models?) Also, if there are a number of multivariate models that one needs to report, the inclusion of post-hoc tests within a manuscript will weigh it down. So, even if certain types of post-hoc tests were to address this concern, they still would end up in an appendix or as article errata and essentially hide a potential Type M or Type S error.

We've now connected our statistics with our data in a clearer way to the reader and resolved the potential for Type S and M errors in the process. I think this is a pretty good approach. It also manages to treat the audience as if they need help reading the figure because the text reiterates what the figure shows. Is this "holding the reader's hand" too much? Keep in mind that you are intimately familiar with your results in a way that the reader is not and the reader has many other things on their mind, so it is always better to hold the their attention by guiding them. Also, the point is to communicate your research findings, not to engage in a competition of "whose model is more opaque?". Such one-upmanship is not an indicator of intelligence, but of insecurity.

What are the downsides though? One potential issue is that the prose can become much longer. You are writing more, so in a context where more words cost more to publish or where there is a strict word limit, this method is less good. This issue can be ameliorated by reporting summary statistics just for those effects which are relevant to the hypothesis under investigation. There is another approach here as well - why not just eliminate statistics from the results section prose altogether. If it is the statistics that get in the way of interpreting the relationship between the hypothesis and results, we could just put the statistics elsewhere.

III.  Tabling your stats

Another approach to enhancing the readability of your research is to place the results from statistical tests and models in a table. I'll admit - when I first studied statistics I was told to avoid this. Yet, I can also see the appeal of this approach. Consider that as models have gotten more complex, there are more things to report. If one is avoiding null hypothesis significance testing or if one is avoiding p values, a set of different values might need to be reported which would otherwise be clunky within the text itself. At the same time, reviewers have been demanding more replicability and transparency within statistical models themselves. This means that they may wish to see more details - many of which need to be included in a table.

A very good recent example of this is found in an interesting recent paper by Schwarz et al. (2019) where the authors investigated the phonetics of the laryngeal properties of Nepali stops. I have included a snippet of this practice from this paper below (reprinted with authors' permission).

Snippet from p,123 of Schwarz, Sonderegger, and Goad (2019), reprinted with permission of the authors.
The dependent variable in the linear mixed effects model here is VD (voicing duration). The authors refer the readers to a table of the fixed effects. They include p values and discuss the directionality and patterns found within the data by referring the readers to a figure. The paragraph here is very readable because the statistics certainly do not interfere with the prose. The authors have also avoided Type M and Type S interpretation errors by stating the effects' directionality and using adverbial qualifiers, e.g. slightly.

One general advantage of tabling statistics is that reading one's results becomes more insightful. When done in a manner similar to what Schwarz et al. do above, readers also do not forget about the statistics completely. This is accomplished by commenting on specific effects in the model even though all the statistics are in the table.

If this is not done, however, the potential problem is that the reader might forget about the statistics completely. In such a case, the risk for a Type M or Type S error is inflated. Moreover, sometimes the effect you find theoretically interesting is not what is driving improvement to statistical model fit. This is obscured if individual results are not examined in the text at all.

Tip #2Whether tabling your stats or not, always include prose discussing individual statistical effects. Include magnitude and sign (positive or negative effect) in some way in the prose.

There is, of course, another alternative here - you can always combine an interspersed method with the tabling of statistical results. This would seem to address both a frequent concern among reviewers that they be able to see specific aspects of the statistical model while also not relegating the model to an afterthought while reading. I could talk about this method in more detail, but it seems as if most of the main points have been covered.

IV. Final points
There are probably other choices that one could make in writing up statistical results and I welcome suggestions and ideas here. As phonetics (and linguistics) have grown as fields, there has been a strong focus on statistical methods but perhaps less of an overt conversation about how to discuss such methods in research effectively. One of the motivations to writing about these approaches a bit is that, when I started studying phonetics in graduate school, much of what I saw in the speech production literature seemed to follow the laundry list approach. Yet, if you have other comments, please let me know.

References:
DiCanio, C., Benn, J., and Castillo García, R. (2018). The phonetics of information structure in Yoloxóchitl Mixtec. Journal of Phonetics, 68:50–68.

Schwarz, M., Sonderegger, M., and Goad, H. (2019). Realization and representation of Nepali laryngeal contrasts: Voiced aspirates and laryngeal realism. Journal of Phonetics, 73:113–127.


Kuznetsova, A., Brockhoff, P. B., and Christensen, R. H. B. (2017). lmerTest Package: Tests in Linear Mixed Effects Models. Journal of Statistical Software, 82(13):1–26.

Monday, August 5, 2019

Is it Trique or Triqui?

Though I am a linguist who has worked on several languages over the years, one of the languages (or language groups) that I have spent the most time studying is Triqui. There are three major Triqui languages (Copala, Itunyoso, and Chicahuaxtla) and though the latter two have some degree of mutual intelligibility, the Copala dialect/language is mostly unintelligible to speakers of the other two dialects/languages.

There are all sorts of interesting things about these languages and about indigenous languages in Mexico, more generally. However, one of the persistent questions I get asked is about the name of the language itself - "is it Trique [
ˈtʰɹike] or Triqui [ˈtʰɹiki]?" The answer to this is rather simple - in Spanish used by both Triqui speakers and non-Triqui speakers in Mexico, it's [ˈtɾiki]. So, the closest equivalent in English is [ˈtʰɹiki], with a final [i] sound.

But the follow-up question is usually "Why is it spelled with an "e" then?" To understand this, it's necessary to understand a little bit about dialectal differences in the languages and linguistic practice into the 20th century. To begin, the name of the language ostensibly comes from a spanification (or castellanización) of the Triqui phrase /tʂeh³ (k)kɨh³/, 'father/padre + mountainside/monte', meaning something like 'father of the mountain' in the Chicahuaxtla dialect, though this is a bit debatable. There is another word /
tʂːeh³²/ (Itunyoso) or /tʂeh³²/ (Chicahuaxtla and Copala) meaning 'camino' or 'road/path.' So, the name itself may have come from a phrase meaning 'the path of the mountainside.'

One thing to notice is that the Chicahuaxtla dialect retains the central vowel /ɨ/ where it has merged with /i/ in the Itunyoso dialect and, in some contexts, with /u/ in the Copala dialect. 
So, the word for 'mountainside/monte' retains this vowel in Chicahuaxtla where the word is /kːih³/ in Itunyoso Triqui and /kih³/ in Copala Triqui. This vowel also exists in many Mixtec languages (Triqui is Mixtecan) and is reconstructed for Proto-Mixtec (Josserand, 1983).

The first Triqui language to be described was the Chicahuaxtla dialect (Belmar 1897) and he wrote the name of the language as Trique. Now, Belmar was not particularly adept at transcribing many of the nuanced phonetic details of many languages. His tonal transcription is non-existent and he misses many important suprasegmental contrasts. However, he chose "e" here because he heard a difference between /i/ and /
ɨ/ and it was customary at the time to transcribe this latter vowel with "e." This practice goes back to very early Mixtecan/Otomanguean philology - the Dominican friar Antonio de los Reyes (1593) used "e" to transcribe this vowel in Teposcolula Mixtec. So, the six historical Mixtec vowels are, at least in old historical sources, transcribed as /i/ "i", /e/ "ai", /a/ "a", /o/ "o", /u/ "u", /ɨ/ "e." The IPA certainly did not exist during Belmar's time and this practice is simply an extension of a Mexican philological tradition.

Incidentally, the use of 'e' for transcribing mid back unrounded vowels is not limited to languages in Mexico. The romanization of Chinese, called pinyin, uses "e" for the vowel /ɤ/, found in many Chinese languages. This practice in fact seems to go back to earlier romanizations of Chinese and, in fact, the earliest grammar of Chinese was Arte de la lengua Mandarina, written by another Dominican friar, Francisco Varo. Though, as far as I can tell, he did not use "e" in his romanization of Chinese - that came later.


The earliest work on Triqui written in English is Longacre (1952) and he must have simply taken the practice of writing the language with an "e" from Belmar and other Spanish sources. Nowadays, it is written with an "i" in Spanish. Though, due to older sources using an "e", such as all work by Hollenbach on the Copala dialect, from 1973 to 1992, the spelling with an "e" has stuck around.


References:
Belmar, F. (1897). Lenguas del Estado de Oaxaca: Ensayo sobre lengua Trique. Imprenta de Lorenzo San-Germán.

Hollenbach, B. E. (1973). La aculturación lingüística entre los triques de Copala, Oaxaca. América Indígena, 33:65–95.

Hollenbach, B. E. (1992). A syntactic sketch of Copala Trique. In Bradley, C. H. and Hollenbach, B. E., editors, Studies in the syntax of Mixtecan Languages, volume 4. Dallas: Summer Institute of Linguistics and University of Texas at Arlington.

Josserand, J. K. (1983). Mixtec Dialect History. PhD thesis, Tulane University.

de Los Reyes, F. A. (1593). Arte en Lengua Mixteca. Casa de Pedro Balli, Mexico, Comte H. de Charencey edition.

Longacre, R. E. (1952). Five phonemic pitch levels in Trique. Acta Linguistica, 7:62–81.



Friday, May 3, 2019

Compassion in the academy

One of the difficulties I find in being an academic is the standards that you place on yourself. Many of us have gone from doing excessively well in primary school, high school, and college to excelling in graduate school and beyond. At each stage it can feel like a competition. The academic job market is also a competition - you compete for limited positions at limited universities in limited places you would like to live. Yet, if you love research and teaching, then you find yourself committing to being in the race and you try to consistently hold yourself to a high standard.

The sense of competition does not end with getting an academic job either - you compete for grants, for tenure, for papers to be accepted, and for recognition. All of this can wear your spirit down and burn you out. And, frankly, many of us do not want to be "warriors in the fight." We want to be curious and explore our interests and help students become researchers in the process.

One particularly exasperating area of academia is the article review process. Whether I wish to or not, as an author, I often take comments to heart. A criticism of a particular method or point can feel like a criticism on me as a researcher. Replying to reviewers can feel like standing up before a tribunal which is judging all of your perceived defects. It is the place where your harsh, internalized judgment appears to be validated by other people in your field.

In reality though, I have grudgingly learned that my internalized judgments are rarely accurate. Put another way, if a colleague of mine came to see me and verbalized the same self-criticisms, I would be likely to say that they were mistaken. Other people almost always see us better than we see ourselves. You might believe the internalized criticisms though if you struggle with finding self-validation in other ways, such as if you are a minority and feel left out. This can make submitting papers and responding to reviews rather scary. 

Submitting your work for publication does not have to feel this way though. So, I began to think about how some of the dynamics of the review process might be changed to be more encouraging. I've compiled these notes below as a way to encourage mindfulness and compassion in academia, as both an author reading reviews and as a reviewer.

1. Praise is just as valid as critique

Either as a reviewer or as an author receiving a review, we think of any praise as faint praise. If someone tells us "The topic is really interesting and I like the way in which you analyzed X and Y..." we almost universally are looking for a 'but....' to follow. The positive commentary is instantaneously invalidated. We believe it's inserted just to lessen the blow of the criticism to follow.

Yet, it is equally valid to point out positive aspects of the work as it is to point out areas in need of improvement. Doing so also does not need to involve lowering one's standards for scholarship. As a parallel, consider the comments you might provide on students' homework assignments. If you only ever pointed out problems on the homeworks and ignored praise for doing well, you would probably get labeled a harsh and demanding academic. 

We have come to expect that the review process will be all criticism, so we brace ourselves when we receive a review. We open it, put it down, walk away from it for weeks, and then pick it up again when our emotions have subsided. This is a sign that more compassion needs to be part of the process. Incidentally, being mindful in providing and interpreting praise in one's work are significant ways to create gratitude for the process. Wouldn't it be a paradigm shift if we saw peer review this way?

(As a side note, if you find yourself reading this and mentally dismissing the advice, consider that other people might not be as able to brush off criticism as well as you. Then, consider giving empathy a try.)

2. Your pet peeve might not be crucial

There is no perfect research in academia. Each paper that is submitted to a journal has its flaws. A large part of what makes scientific discovery move forward is addressing flaws in future studies. Regardless of how methodologically good a paper is, it is easy to find some flaw that you, as a reviewer, might interpret as a critical error. This issue can get easily blown out of proportion if there is little else to criticize in the paper. Framing personal pet peeves in relation to the aspects of the research that are sound provides a more useful perspective for the authors of the paper.

3. The when of the review and asking for help

Sometimes academics review papers when they are exhausted or unable to concentrate. A hallmark of this type of review is an excessive number of questions related to clarity. The paper may otherwise be clear, but the mental state of the reviewer has deteriorated. The tone of the review can be set badly if a reviewer gets a general feeling of unease because they read the paper while tired - even if they return to it the next day refreshed. It is the job of editors to help guide authors through such reviews, especially if questions of clarity come from just one of the reviewers.

What this means in practice is that authors should ask for help. Unless your research is simply rejected out of hand with no review (rare in my field), editors want to see the publication eventually succeed. I have always had a better experience in revising a paper if I have discussed the review with the editors explicitly than when I have skipped doing so. So, ask if the issues of clarity are serious. Ask if a small point has been blown up by a reviewer into something much bigger than it needs to be. Even if the editors agree with all the points raised by the reviewers, they will have helpful advice about how to tackle the points constructively.



Friday, April 26, 2019

You can relax now, I'm running for president


The last couple of years have been kind of tough, right? We've had to deal with a corrupt and immature president and administration that thinks more about doing what serves them than what serves the American people. It's been a real doozy for the democrats. Fortunately, I'm here to tell you that it's going to be okay. You can all calm down because I'm going to run for president.

“Another person running on the democratic ticket?!” you ask yourself. Now, now. There's no reason to be alarmed. I happen to have some experience and what Americans need now is someone confident to take on Trump and his cronies, someone who is able to set the record straight. 


I've raised three sons and all were All-American football players in Delaware. Growing up, they got their fair share of stern warnings from me and my wife Molly. They listened because I am a firm believer in sitting down and talking like rational human beings. I'll do the same as your president.

We should all be looking toward the future. Many of the other candidates are stuck in the past and this really worries me. The Democrats have the better path forward than the Republicans do but, listen, an old man whose hair is frizzed out yelling about how evil corporations are in America is not gonna convince people to vote for us. Neither is a school mom doing the same. Americans know that the public and private sectors can still work together like they used to. We can do this while protecting our rights at the same time.

It's a big playing field out there and I know that you have lots of options. Some of the other candidates come from different ethnic backgrounds and they all have important things to say.

You know though, as I was brushing my pearly whites with Aquafresh this morning, I remembered something my coal-miner dad used to tell me - “Here, we're all Americans first. We all eat apple pie on the fourth of July.” So, whether you have it with vanilla or strawberry ice-cream, whether you're a same-sex couple or African-American or Hispanic, or any of the other different flavors, you're an American. My campaign is about getting beyond our differences and doing what's right.

As long as we're stuck on what divides us, we'll never move forward. This is my promise to you as a candidate - I'm going to push hard to do what's right and we're going to make progress unified as a country because we're all in this fight together. My name is Joe Biden and I am running for president.

Monday, April 8, 2019

The forgotten public universities


"Yes, in most of the world, young people go to university in the city where they grew up, but in the United States, I would explain, most young people aspire to “go away” to college, and that means that even a pre-application tour is a costly and time-consuming proposition.”

I would like to point out that this is most likely incorrect. According to a report by the National Center for Educational Statistics (https://nces.ed.gov/programs/coe/indicator_cha.asp), undergraduate enrollment at public universities in 2016 was 13.7 million students, while undergraduate enrollment at private universities was 2.7 million students. Public university students outnumber private ones by factor of 5. As a faculty member at a large public university, I can tell you that the majority of the undergraduate body is local. That is, they did not go away to university (or go away very far). So, in fact, most students in the US do indeed go to the university in the state where they grew up. Though a percentage of these students may have strived to attend private universities, most have believed public institutions to be a good deal in financial terms (they cost 1/3 the amount of private universities) and sufficiently good academically. It is the large public universities which teach most students in the US. It is also, incidentally, the large public universities that do much of the federally funded research in the US. 

The recent scandal regarding college admissions touches upon our hope in meritocratic institutions in the US. It leads us to important conversations. Yet, this criticism is itself elitist. It reflects the idea that the only educational systems worth discussing are those which are private and whether intentioned or not, it excludes at least 80% of the students attending universities and colleges in the US. 

Wednesday, January 9, 2019

Linguistic common ground as privilege

2019 was named the International Year of Indigenous Languages by UNESCO. My friends and colleagues at the recent Annual meeting of the Linguistic Society of America (LSA) have been on Facebook, Twitter, and other social media discussing what this means for Linguistics as a field. With respect to publishing, several journals have pushed to emphasize linguistic research on indigenous languages. The LSA's own flagship journal, Language, has put out a call for submissions on different indigenous languages of the world. The Journal of the Acoustical Society of America has even put out a call for submissions on under-represented languages.

There may be other journals too (which I am currently unaware of) attempting to emphasize how work on indigenous languages enhances our knowledge of language more generally, improves scholarship, and, in many cases, can promote the inclusion of ethnic minorities speaking or revitalizing these languages. This is all very positive and, as a linguist and scholar who studies indigenous languages of Mexico, I applaud the effort.

Will it be enough though? If linguists are serious about promoting the equality of indigenous languages and cultures in publishing, a greater type of paradigm shift needs to take place in what we believe is worthy of scholarship.

1. Not just a numbers game

When you read academic articles in linguistics, chances are that the topic is examined in a language that you know about. This is partly due to speaker population. There is extensive scholarship in English, Mandarin Chinese, Hindi/Urdu, Spanish, Arabic, French, Russian, and Portuguese because 4.54 billion people speak these as their first or second languages.

Where linguistic scholarship has developed has also played a strong role. There are 263 million first language speakers of Bengali and 23 million first language speakers of Dutch in the world. Bengali outnumbers Dutch by more than 11:1. Yet, a quick search on Google Scholar for "Bengali phonetics" reveals 4,980 hits, while a simultaneous "Dutch phonetics" search reveals 52,600 hits. A search for "Bengali syntax" reveals 11,800 hits while "Dutch syntax" reveals 180,000 hits. When it comes to academic articles, the numbers are reversed. Here, Dutch outnumbers Bengali by either 10:1 or 16:1.

Dutch phonetics and syntax are not inherently more interesting than Bengali phonetics and syntax. Bengali has a far more interesting consonant system (if you ask me as a phonetician). Even Bengali morphology, which is far more complex than Dutch morphology, is under-studied relative to Dutch. Dutch speakers just happen to reside in economically-advantaged countries where there has been active English-based scholarship on their language for many years. Bengali speakers do not.

2. Small phenomena in big languages, big phenomena in small languages

A consequence of studying a language that has a history of academic scholarship is that many questions have already been examined. There is a literature on very specific aspects of the sound system of English (look up "English VOT", for instance) and Dutch morphology (look up "Dutch determiners", for instance). If linguists wish to study these languages and make a contribution, they must take out their magnifying glass and zoom in on specific details of what is already a restricted area.

To a great degree, the field of linguistics respects this approach. Scholarship is enhanced by digging deeply into particular topics even in well-studied languages. Moreover, since many members of the field are familiar (at least passively) with the basic analyses of phenomena in many well-studied languages, linguists zooming in on the particular details benefit from shared common ground. Resultingly, linguists are able to give talks on very specific topics within the morphology, syntax, phonology, or pragmatics of well-studied languages. One can find dissertations focusing on specific types of constructions in English (small clause complements) or specific morphemes in Spanish (such as the reflexive clitic 'se'). This is the state of the field. Linguists all agree that such topics are worthy of scholarship.

But imagine if you were asked to review an abstract or a paper where the author chose to zoom in on the specific details of a particular syntactic construction in Seenku (a Mande language spoken by 17,000 people in Burkina Faso, see work by Laura McPherson) or how tone influences vowel lengthening in a specific Mixtec language (spoken in Mexico). These are minority and indigenous languages. Many linguists would agree that these topics are worthy of scholarship if they contribute something to our knowledge of these languages and/or to different sub-disciplines of linguistics, but where do we place the bar by which we judge?

In practice, linguists often think these topics are limited in scope - even though they are no more limited than topics focusing on the reflexive clitic 'se' in Spanish. A consequence of this is that those working on indigenous languages must seek to situate their work in a broader perspective. This might mean that the research becomes comparative within a language family or that the research is a case study within a broader survey on similar phenomena. Rather than magnifying more deeply, if they want their work to be considered by the field at large, linguists working on indigenous languages often take the "go wide" approach instead.

Note that this is not inherently negative. After all, we should all seek to situate our work in broader typologies and compare our findings to past research. It's just that the person working on the Spanish reflexive clitic is seldom asked to do the same. Their contribution to scholarship is not questioned.

3. Privilege and a way to move forward

For the most part, academic linguists believe that all languages have equal expressive power. It is possible to express any human idea in any language. Linguists also believe (or know) that language is arbitrary. De Saussure famously argued that the relation between the signified and the signifier is arbitrary. In other words, it is equally valid to express plurality on nouns with an /-s/ suffix (in English) or a vowel change (in Italian and Polish). No specific relation is better than another in a different language. If we take these ideas seriously, research on certain languages should not be more subject to scrutiny than research on other ones.

Whether intentioned or not, both people and languages can be granted privilege. Scholars working on well-studied languages benefit from a shared linguistic common ground with other scholars which allows them to delve into deep and specific questions within these languages. This is a type of academic privilege. Without this common ground, scholars working on indigenous languages can sometimes face an uphill battle in publishing. And needing to prove one's validity is a hallmark of institutional bias.

So, how do we check our linguistic privilege in the international year of indigenous languages? As a way of moving positively forward into 2019, I'd like to suggest that linguists think of the following questions when they read papers, review abstracts/papers, and attend talks which focus on indigenous languages. This list is not complete, but if it has made you pause and question your perspective, then it has been useful.

Question #1: What languages get to contribute to the development of linguistic theory? Which languages are considered synonymous with "Language"?

If you have overlooked an extensive literature on languages you are unfamiliar with and include only those you are familiar with, you might be perpetuating a bias against indigenous languages in research. "Language" is not synonymous with "the languages I have heard of." Findings in indigenous languages are often considered "interesting footnotes" that are not incorporated into our more general notions of how we believe language works.

Question #2: Which phenomena are considered "language-specific"?


There is value to exploring language-specific details, but more often than not, phenomena occurring in indigenous languages are considered exotic or strange relative to what is believed to be typical. Frequently, judgments of typicality reflect a bias towards well-studied languages.

Question #3: Do you judge linguists working on indigenous languages or articles on indigenous languages by their citation index? (h/t to Laura McPherson)

Citations of work on indigenous languages are often lower than citations of work on well-studied languages. In an academic climate where one's citation index is often considered as a marker of the value of one's work, one might reach the faulty conclusion that an article on an indigenous language with fewer citations is poor scholarship.

Question #4: Do you quantify the number of languages or the number of speakers that a linguist works with?
If a linguist studies one or two indigenous/minority languages, do you judge their knowledge of linguistics/language to be lesser than that of someone who does research on one or two well-studied languages? If so, you are privileging well-studied languages.

I'd like to specifically note that I am not a sociologist of language or a sociolinguist. There are undoubtedly others who have probably worked on this question.

Sunday, December 30, 2018

What is phonetics? A 20 minute guide for academics

As a phonetician, I often get so absorbed within my own area of study that I fail to notice other perspectives. My field is devoted to the study of speech sounds. It is important to humanity, to science, and to knowledge, but so are many other fields which I may not even recognize as distinct research areas in their own right. To get beyond this, it is important to try to educate the public and, in particular, other academics outside one's field.


Figure 1: Siri cost Apple like $300 million dollars to create and involved speech recording (phonetics), speech processing (phonetics), and speech annotation (phonetics).
Telling the public that phonetics is an important field is easy. People accept that speech sounds are important things to study. Many people have opinions about the sounds of language. Ask almost anyone their opinions about different dialects and they will immediately voice them (their opinions, that is). Tell them about technology like Siri or Alexa and it is not much of a stretch to get them to realize that people had to think about speech acoustics and analyzing speech signals in order to create these things.

Trying to educate other academics about phonetics is a rather more difficult task, however. Academics are a proud group composed of people who make a living being authorities on arcane topics. Tell them that you study a topic that they believe they know about (like language, ahem) and they will be highly motivated to voice their opinion, even though they may know as much about it as the average non-academic. Frankly, academics are terrible at admitting ignorance. I'll admit that I struggle with this too when it comes to areas that I think I know about. In response to this, I have created a short guide to phonetics as a way to tell other academics two things: (1) phonetics is an active area of research and (2) there is a lot we do not know about speech.

I. Starting from Tabla Rasa
Let's start with what phonetics is and is not. Phonetics is the study of how humans produce speech sounds (articulatory phonetics), what the acoustic properties of speech are (acoustic phonetics), and even how air and breathing are controlled in producing speech (speech aerodynamics). It has nothing to do with phonics, which is the connection between speech sounds and letters in an alphabet. In fact, it has little to do with reading whatsoever. After all, there are no letters in spoken language - just sounds (and in the case of sign languages, just gestures).


So imagine a world where you have to think about language but are unable to refer to the letters of your alphabet. This is, in fact, one of the motivations for the International Phonetic Alphabet, or the IPA. Consonant sounds are represented using the IPA and are principally defined in three ways:
  1. Voicing - whether your vocal folds (colloquially called your "vocal cords") are vibrating when you make the speech sound.
  2. Place of articulation - where you place your tongue or lips to make the speech sound.
  3. Manner of articulation - either how tight of a seal you make between your articulators in producing the speech sound or the cavity that the air flows through (your mouth or your nose being the two possibilities).
Vowel sounds are a bit harder to define, but phoneticians distinguish them in terms of (a) how open your jaw is, (b) where your tongue is, and (c) what your lips are doing.

Why define speech this way? First, is scientifically accurate/testable. After all, the same sound should be produced in a similar way by different speakers. We can measure exactly how sounds are produced by imaging the tongue moving; or by recording a person and looking at an image of the acoustics of specific sounds. The figure below shows just one method phoneticians can use to examine how speech is produced.


Figure 2: An ultrasound image of the surface of the tongue, from Mielke, Olson, Baker, and Archangeli (2011). Phoneticians can use ultrasound technology to view tongue motion over time.

Second, this way of looking at speech is also useful for understanding grammatical patterns. When we learn a language, we rely on regularities (grammar) to form coherent words and sentences. For a linguist (and a phonetician), grammar is not something learned in a book and explicitly taught to speakers. Rather, it is tacit knowledge what we, as humans, learn by listening to other humans producing language in our environment.

To illustrate this, I'll give you a quick example. In English, you are probably familiar with the plural suffix "-s." You may not have thought about it this way, but this plural can be pronounced three ways. Consider the following words:


[z] plural       [s] plural       [ɨz] plural      
drum - drum[z]       mop - mop[s]       bus - bus[ɨz]      
rib - rib[z] pot - pot[s] fuzz - fuzz[ɨz]
hand - hand[z] bath - bath[s] wish - wish[ɨz]
lie - lie[z] tack - tack[s] church - church[ɨz]

In the first column, the plural is pronounced like the "z" sound in English. In the IPA this is transcribed as [z]. In the second column, the plural is pronounced like the "s" sound in English - [s] in the IPA. In the third column, the plural is pronounced with a short vowel sound and the "z" sound again, transcribed as [ɨz] in the IPA.

Why does the plural change its pronunciation? The words in the first column all end with a speech sound that is voiced, meaning that the vocal folds are vibrating. The words in the second column all end with a speech sound that is voiceless, meaning that the vocal folds are not vibrating. If you don't believe me, touch your neck while pronouncing the "m" sound (voiced) and you will feel your vocal folds vibrating. Now, try this with while pronouncing the "th" sound in the word "bath." You will not feel anything because your vocal folds are not vibrating. In the third column, all the words end with sounds that are similar to the [s] and [z] sounds in place and manner of articulation. So, we normally add a vowel to break up these sounds. (Otherwise, we would have to pronounce things like wishs and churchs, without a vowel to break up the consonants.) What this means is that these changes are predictable; it is a pattern that must be learned. English-speaking children start to learn it between ages 3-4 (Berko, 1958).

Why does this rule happen though? To answer this question, we would need to delve further in to how speech articulations are produced and coordinated with each other. Importantly though, the choice of letters is not relevant to knowing how to pronounce the plural in English. It's the characteristics of the sounds themselves. Rules like these (phonological rules) exist throughout the world's languages, whether the language has an alphabet or not - and only about 10% of the world's languages even have a writing system (Harrison, 2007). Unless you are learning a second language in a classroom, speakers and listeners of a language learn such rules without much explicit instruction. The field of phonology focuses on how rules like these work across the different languages of the world. The basis for these grammatical rules is the phonetics of the language.

II. Open areas of research in phonetics
The examples above illustrate the utility of phonetics for well-studied problems. Yet, there are several broad areas of research in phoneticians occupy themselves with. I will focus on just a few here to give you an idea of how this field is both scientifically interesting and practically useful.

a. Acoustic phonetics and perception
When we are listening to speech, a lot is going on. Our ears and our brain (and even our eyes) have to decode a lot of information quickly and accurately. How do we know what to pay attention to in speech signal? How can we tell whether a speaker has said the word 'bed' or 'bet' to us? Speech perception concerns itself both with what characteristics of the sounds a listener must pay special attention to and how they pay attention to these sounds.

This topic is hard enough when you think about all the different types of sounds that one could examine. It is even harder when you consider how multilingual speakers do it (switching between languages) or the fact that we perceive speech pretty well even in noisy environments. Right now, we know a bit about how humans perceive speech sounds in laboratory settings, but much less so in more natural environments. Moreover, most of the world is multilingual, but most of our research on speech perception has focused on people who speak just one language (often English).


Figure 3: A speech waveform and spectrogram. Here we see the phrase "to go without water for" spoken by a native English
speaker reading from a text. The words are labelled below the spectrogram along with the sounds using the IPA. There are no pauses in the speech signal but humans are able to pull out individual words when listening to speech.

There is also a fun fact relevant to acoustics and perception - there are no pauses around most words in speech! Yet, we are able to pull out and identify individual words without much difficulty. To do this, we must rely on phonetic cues to tell us when words begin and end. An example of this is given in Figure 3. Between these five words there are no pauses but we are aware of when one word ends and another begins.

How are humans able to do all of this so seamlessly though? and how do they learn it? Acoustic phonetics examines questions in each of these areas and is itself a broad sub-field. Phoneticians must be able to examine and manipulate the acoustic signal to do this research.

Is this research useful though? Consider that when humans lose hearing or suffer from conditions which impact their language abilities, they sometimes lose the ability to perceive certain speech sounds. Phoneticians can investigate the specific acoustic properties of speech that are most affected. Moreover, as I mentioned above, the speech signal has no pauses. Knowing what acoustic characteristics humans use to pick apart words (parse words) can help to create software that recognizes speech. These are among a few of the many practical uses of research in acoustic phonetics and speech perception.

b. Speech articulation and production
When we articulate different speech sounds, there is a lot that is going on inside of our mouths (and in the case of sign languages, many different manual and facial gestures to coordinate). When we speak slowly we are producing 6-10 different sounds per second. When we speak quickly, we can easily produce twice this number. Each consonant involves adjusting your manner of articulation, place of articulation, and voicing. Each vowel involves adjusting jaw height, tongue height, tongue retraction, and other features as well. The fact that we can do this means that we must be able to carefully and quickly coordinate different articulators with each other.

To conceptualize this, imagine playing a piano sonata that requires a long sequence of different notes to be played over a short time window. The fastest piano player can produce something like 20 notes per second (see this video if you want to see what this sounds like). Yet, producing 20 sounds per second, while fast, is not that exceptional for the human vocal tract. How do speakers coordinate their speech articulators with each other?


Figure 4: Articulatory movement from electromagnetic articulography, which involves gluing sensors on the articulators and tracking their motion in real time. Waveforms of the acoustic signal are shown above, followed by an acoustic spectrogram. The three lower panels reflect vertical movement of the back of a speaker's tongue (TB - top), the front region of the a speaker's tongue (TL - middle), and the lower lip (LL - bottom).

Phoneticians that look at speech articulation and production investigate both how articulators move and what this looks like in the acoustic signal. Your articulators are the various parts of your tongue, your lips, your jaw, and the various valves in the posterior cavity. The way in which these articulators move and how they are coordinated with one another is important both for understanding how speech works from a scientific perspective and extremely useful for clinical purposes. One of the reasons that this is important is that movements overlap quite a bit.

Since we are familiar with writing, we like to think that sounds are produced in a clear sequence, one after the other, like beads on a string. After all, our writing and even phonetic transcription reflects this. Yet, it's not the truth. Your articulators overlap all the time and moving your lips for the "m" sound in a word like "Amy" overlaps with moving your lips in a different way for the vowels.

To provide an example, in Figure 4, a Korean speaker is producing the (made-up) word /tɕapa/. The lower panels show just when the tongue and lips are moving in pronouncing this word. If you look at the spectrogram (the large white, black, and grey speckled figure in the middle), you can observe what looks like a gap right in the middle of the image. This is the "p" sound. Now, if you look at the lowest panel, we observe the lower lip moving upward for making this sound. This movement for the "p" happens much earlier than what we hear as the [p] sound, during the vowel itself. Where is the "p" then? Isn't it after the /a/ vowel (sounds like "ah")? Not exactly. Parts of it overlap with the preceding and following vowels, but parts of those vowels also overlap with the "p." In the panel labelled TLy, we are observing how high the tongue is raised. It stays lowered throughout this word because it needs to stay lowered for the vowel /a/. So, the "a" is also overlapping with the "p" here.

Overlap in speech is the norm and sometimes speakers move their articulators in ways that are unexpected. You might struggle to coordinate your articulators in a particular way when you are learning new sounds in a first language (as a child) or new sounds in a second language (as a child or adult). You also might have difficulty producing sequences of sounds due to a range of physical or cognitive disorders. By looking at speech articulation, phoneticians are able to examine what is typical in speech and also what is atypical.

One fun way to examine what speakers can do is to have them speak really quickly or give them tongue twisters. As mentioned earlier, speech can be really fast. The Korean speaker above produced 5 speech sounds in just 400 milliseconds (12 sounds per second) and she was speaking carefully. When speakers speed up, phoneticians can both determine where difficulties arise and how different movements must be adjusted relative to one another.

References:
Berko, J. (1958). The child's learning of English morphology. Word, 14(2-3), 150-177.
Harrison, K. D. (2007). When languages die. Oxford University Press.
Mielke, J, Olson, K, Baker, A, and Archangeli D (2011) Articulation of the Kagayanen interdental approximant: An ultrasound study. Journal of Phonetics 39:403-412.