Haafu as a Sum of Natural Identities

The following is a sample paper I made for an ESL project-based class using real, but very limited, research – hence the somewhat simplified writing style and dearth of citations. Still, I think it comes to a legitimate conclusion that is somewhat supported by the data, and the research is on a worthwhile topic.

Haafu Identity

Gerry thought she was white. She didn’t understand why her teacher always grouped her with Black kids in her recently integrated class. As a biracial, part Japanese and part White, she was not always identified as a member of a minority group, but institutions, including her school, sometimes forced a minority identity on her (“Beyond the Binary”).

Japanese biracials and the communities around them seem to have conflicting opinions on their ethnic identity (Morrison). Some see them (or themselves) as simply Japanese, particularly if they were raised in Japan. Others see their Japanese side as a kind of decoration on their basic Whiteness, something that adds to their basic White identity instead of conflicting with it. Others still see biracial Japanese, and sometimes other Asian ethnicities, as a unique identity of its own. Among these, “haafu” has emerged as a popular descriptor. Unlike the Hawaiian “hapa”, “haafu” refers specifically to Japanese biracials, as it originates from the Japanese word for these people (which originates in the English word “half”).

Like other identities, “haafu” identity can be formed and recognized in many different ways. The psychologist James Paul Gee has divided identities into four categories: Nature, Institution, Discourse, and Affinity. All of these categories share a basic definition of identity. As Gee describes, “When any human being acts and interacts in a given context, others recognize that person as acting and interacting as a certain ‘kind of person’ or even as several different ‘kinds’ at once” (99). I.e., being recognized, not simply being, is an essential part of an identity. The basis for this recognition creates the four categories. For instance, if one is recognized as being a certain “kind of person” because of circumstances of one’s birth, then that identity is called a Nature identity, or N-identity (Gee 102). On the other hand, people can come to have a Discourse identity (or D-identity) “through discourse and dialogue (D-Identities) without the overt sanction and support of ‘official’ institutions that come, in some sense, to ‘own’ those identities” (Gee 103). That is, a D-identity is not sanctioned by an institution (which is an I-identity) or seen as natural (which would be an N-identity). Last, an A-identity is one “shared in the practice of ‘affinity groups’” (Gee 100). Gee gives being a member of a fan club as an example to illustrate that an A-identity is not seen as natural or given by a formal institution, but is formed in practices that one participates in with other people voluntarily. It may seem obvious that a racial identity like “haafu” is an N-identity, but as we shall see, not everyone views “haafu” identity as strictly a matter of nature.

In order to study what kind of identity “haafu” is, I conducted research on people who identify as “haafu”. The methods of this research will be covered in the following section.

Methods

In order to study the definition of “haafu” identity, I was faced with several issues. Among them were how to reach a large enough number of people, which questions to ask, how to reduce bias, and whether to focus on quantitative or qualitative data.

To reach a large enough number of people, I decided to reach out to people on the Internet with a simple, short survey. To find people who identify as “haafu”, I joined several groups on Facebook that had “haafu” or “hapa” in the title. I then posted the link to my survey with a short explanation, and hoped that I would gather enough participants that way. I also reached out informally to a small number of people known to me, and told them that they may also pass the survey to anyone they know who might be interested in completing it. In this way, I was able to gather 11 surveyees. I kept the survey short because I did not want people to be afraid to spend the time to answer every question. 

I wrote the survey itself to give equal attention to each of Gee’s categories of identity. The reason for this was simply that I did not want to bias the answers by appearing to favor a particular interpretation of “haafu” identity. I assigned two questions to each of the four categories. The reason for the two separate questions was that, as Gee points out, identities are sometimes assigned despite the will of the individual, meaning that people may have different views of their own identities than the people around them. As Gee states, some identities “can be placed on a continuum in terms of how active or passive one is in ‘recruiting’ them” (104). In other words, identities can be given and they can be earned, or sometimes both. I wanted to assess how much people’s own views of their identities matched how other people saw them, so there were one question about self-identity (what one thinks of oneself) and one about other-identity (what one thinks that others think of oneself) for each category. For example, the questions about N-identity said, “People around me see ‘haafu’ identity as a matter of birth” and “I see my ‘haafu’ identity as a matter of birth.” These questions served to find whether there was a difference in whether the individual saw his or her identity as an N-identity and whether he or she thought that other people saw it that way. 

Each question had an optional “comments” section as well, which was intended for open-ended elaboration. This means that I mostly focused on quantitative data in order to calculate statistics, but I allowed for qualitative data as well. The reason that I focused on quantitative data was that I suspected that people would rate the different categories of identity differently, and I wanted to be able to find that difference easily and display it in numbers. For that reason, almost all of the questions in the survey were on a scale of 1-5 (a Likert scale). The results of this survey will be given in the next section.

Results

The survey results show that people who identify as “haafu” overwhelmingly consider it an N-identity. They consider their own “haafu” status as coming from natural forces, and believe that others around them also consider it natural. Full results can be seen in Appendix A.

To go into more detail, the mean responses on a scale from 1-5 for the N-identity questions were 4.4 (whether others see “haafu” as an N-identity) and 4.6 (whether the surveyee sees it as an N-identity). No other question had an average higher than 3. The lowest averages were for I-identity. This indicates that people see “haafu” status as similar to the way Gee describes being a twin: “Being an identical twin is a state that I am in, not anything that I have done or accomplished. The source of this state-the ‘power’ that determines it or to which I am ‘subject’-is a force (in this case, genes) over which I had no control” (Gee 101). It seems that the surveyees agreed that “haafu” status fits this description, as a result of nature.

Qualitative comments after the N-identity survey questions confirm surveyees’ opinions that “haafu” status was a result of genes. Two surveyees responded unambiguously that “haafu” (and also “Japanese” and “non-Japanese”) were natural, immutable identities. In their words: 

The definition of hafu only depends on whether a person is born from one parent who is non-Japanese, and another who is Japanese.

Haafu means ‘half-Japanese’, or having one Japanese parent, and one non-Japanese parent.

One respondent acknowledged the complications inherent in the topic: 

I believe that to be ‘haafu’ you need to be born as such. But identity is complicated, and individuals who fall into this category may not identify as ‘haafu’. 

Another surveyee put it quite bluntly in a comment after the D-identity survey items: 

Haafu is an absolute circumstance of birth, not a particular experience. There are no degrees of haafu. 

In sum, most of the comments reify the status of “haafu” as an N-identity.

Moreover, people who felt that “haafu” was an N-identity were especially unlikely to consider it a D-identity. Gee defines a D-identity against N-identity specifically, using his friend’s “charismatic” identity as an example: “It is not something that one just ‘is’ (‘born with’; note that one cannot be charismatic all alone by oneself on an island)” (103). Appendix B holds a correlation table for each of the quantitative answers. The N-identity questions (#1 and #2) are correlated with the D-identity questions for others (#5) more negatively than any other set of items in the table. In simpler language, people who believe “haafu” is an N-identity (according to themselves or according to others) tend not to think of “haafu” as a personal characteristic like “stubborn”, “charismatic”, or “friendly”, and those who do think of it as a personal characteristic are less likely to consider it natural, genetic, or a matter of birth (and again, most respondents thought it was a matter of birth).

The qualitative comments bring a bit more nuance to the apparent difference between those who view “haafu” as an N-identity and those who view it as a D-identity. One key distinction that surveyees made is in the audience; specifically, among a Japanese audience, “foreign” behavior is more associated with “haafu” identity than “Japanese” behavior is. One surveyee responded: 

People will always see me as Haafu no matter how Japanese I act, or even if I get better grades than the rest of the class in [Japanese national language], the assumption is that English will be my ‘thing’. Being able to behave according to Japanese norms is important to me but doesn’t change others’ views. It must be hard for Haafu who were never raised with the second language, foreign parent, or experience of ever living abroad to be viewed with these assumptions when in reality their experience is 100% Japanese.

“Hafu behavior” does seem to get pushed by the media in Japan … As a result, many people are given information from various sources on expected behavior of hafu in Japan…

Other surveyees responded that language skills might identify them more with Japan or with an English-speaking culture, but not with “haafu” in particular.

I tend to not really define myself as “haafu” since that feels kind of vague. Instead I’m more likely to think about myself in relation to the “wholes” (Japaneseness or Americanness, Whiteness) which my behavior feels like it appeals to. So speaking better Japanese makes me feel more Japanese but not more “haafu”.

I am not sure how others view this. I have heard myself being described as “more Japanese” because of behavior but never as more or less haafu.

In other words, “haafu” does not seem to have a set of behaviors associated with it; rather, the two societies that a person identified as “haafu” is presumed to inherit do. Instead of a person identified as “haafu” exhibiting “haafu” behavior, that behavior is instead attributed to Japan or the other country. 

One surveyee pointed out the difficulty in measuring identity according to Gee’s definition (recall that Gee’s definition explicitly depends on other people’s thoughts): 

It seems extremely nebulous to have people comment on what other people think. People thinking about what others are thinking. Better to ask people for their own views? Otherwise just very foggy.

This may be an inherent difficulty of studying social identity. If, at least in Gee’s conception, an identity is a view of the individual by the people surrounding them, then determining an identity for that person should involve asking many other people rather than the person himself or herself. A researcher would do well to remember that any individual has at best indirect information about their own identities and can only make partially educated guesses about what the “others” that “recognize” them are thinking. A future version of this research should probably rely less on self-assessment for determining identity under Gee’s definition.

Conclusions

The N-identity definition of “haafu” includes reference to at least two other identities typically thought of as N-identities, although the one which provides a salient difference to the majority population depends on the host culture. To illustrate, in Japan, in which the majority identify as Japanese, it is the “foreign” heritage which seems to provide most of the content of a “haafu” identity. Conversely, in a society where Japanese are not the majority, it is Japanese heritage that sets “haafu” apart from other racial identities. Perhaps because most of the characteristics which might be used to identify “haafu” behaviorally are already associated with a larger, more culturally salient group (“foreign” in Japan and “Japanese” elsewhere), it is difficult to describe people as acting “haafu”; instead they act foreign or act Japanese. Characteristics that might be attributed to “haafu” identity are therefore subsumed by these larger groups. Even as an N-identity, “haafu” status is defined only by its relationships to the larger groups and is not thought to constitute an original group itself (notice that even those respondents who insisted that “haafu” was an N-identity defined it with reference to Japaneseness and non-Japaneseness). 

It may be that “haafu” does not constitute a “kind of person” to many people at all, much the same way “100 pounds of bananas” is not as much a “kind of thing” as “a banana” or “a bunch of bananas”. Being derived from a combination of the rules of logic and components that are themselves natural classes does not make the result also a natural class. It seems that although people view “haafu” status as the combination of the “kinds of person” that are “Japanese” and “non-Japanese” (which is definitely a “kind of person” in Japanese society), it may not be a “kind of person” itself.

A larger point implied especially by the qualitative responses is that “Japanese” and “non-Japanese” are considered “kinds of person” in ways that “haafu” is not. Notice that no respondent felt the need to define “Japanese” in terms of its component parts. One may object that “non-Japanese” is defined by its component parts (or rather the lack of a particular part), but it must be remembered that Japanese culture tends to treat the foreign world as comprising a homogeneous group of English speakers (Tsuneyoshi). What constitutes a “kind of person” that can be accepted as-is is beyond the scope of this paper, but it seems that where identity is concerned, 1+1 may only equal 1+1, as 2 has no definition of its own.

Appendix A

People around me see “haafu” identity as a matter of birth.4.4
I see my “haafu” identity as a matter of birth.4.6
People around me consider “haafu” identity to be derived from an institution that grants that status (e.g. a government document or school record).1.9
I regard the official institutional status of my “haafu” identity as important.2.7
People around me see “haafu” identity as a matter of behavior (e.g. “acting Japanese”) or achievement (e.g. bilingual skills).2.6
I see myself as more “haafu” or less “haafu” depending on my behavior and/or achievements.2.3
People around me regard “haafu” identity as a matter of shared appreciation and/or experience.2.7
I regard myself as more “haafu” when I have shared experiences or participate in activities with other people who identify as “haafu”.2.5

Appendix B

People around me see “haafu” identity as a matter of birth.I see my “haafu” identity as a matter of birth.People around me consider “haafu” identity to be derived from an institution that grants that status (e.g. a government document or school record).I regard the official institutional status of my “haafu” identity as important.People around me see “haafu” identity as a matter of behavior (e.g. “acting Japanese”) or achievement (e.g. bilingual skills).I see myself as more “haafu” or less “haafu” depending on my behavior and/or achievements.People around me regard “haafu” identity as a matter of shared appreciation and/or experience.I regard myself as more “haafu” when I have shared experiences or participate in activities with other people who identify as “haafu”.
1.000.980.030.02-0.300.120.210.00
1.00-0.05-0.14-0.350.020.27-0.11
1.000.500.630.08-0.020.03
1.000.690.38-0.230.36
1.000.01-0.110.00
1.000.410.91
1.000.38
1.00

Works Cited

“Beyond the Binary: An Interview with JALT2014 Plenary Speaker Gerry Yokota”. Journal and Proceedings of the Gender Awareness in Language Education Special Interest Group, Vol. 8, 2015, pp. 58-71.

Gee, James Paul. “Chapter 3: Identity as an analytic lens for research in education.” Review of research in education 25.1 (2000), pp. 99-125.

Morrison, Hanna. “Value of ‘haafu’ as a Category in Education Research.” International Journal of Japanese Sociology 28.1 (2019), pp. 170-182.

Tsuneyoshi, Ryoko. “8. Communicative English in Japan and ‘Native Speakers of English’.” Native-speakerism in Japan, edited by Stephanie Ann Houghton and Damian Rivers, Multilingual Matters, 2013, pp. 119-131.

D-identities in the context of other identities

This is a sample paragraph I wrote while my students were doing the same. It is based on this video. The prompt is “In an academic paragraph, compare one of Hetain Patel’s identities to one of yours. Come to a meaningful conclusion about identity from your comparison.”

Both my and Hetain Patel’s D-identities are reevaluated as the opposite of their initial states based on our I-identities, which shows that I-identity can provide a context for the assignment of other identities. Both Patel and I have D-identities that at first seem to contrast with our I-identities. Specifically, I am unserious and jokey in class, which contrasts with my I-identity as “teacher”; and Patel’s D-identity as serious and intelligent is made to contrast with his I-identity as an artist. In both cases, the apparent contrast forces people to reevaluate our D-identities: me as a serious teacher, and Patel as a skilled artist. In my case, my apparent casualness is seen in the context of my classes as attentiveness to students and a modern, unpedantic teaching style, which are valued by other teachers as signs of modern pedagogical thinking. This makes me seem to be an even more dedicated teacher precisely because I seem to be so unserious in class. On the other hand, Patel’s apparent seriousness is eventually revealed during his act to be an intentionally shallow façade, which because it is part of an artistic performance, shows him to be a humorous artist with an eye for the ironic. That is, his seriousness increases the impact of his comedic performance and the audience’s esteem for him as an artist because the audience knows it to be a skilled performance. In both cases, a quickly formed D-identity changes to an opposite D-identity when seen in the context of an I-identity.

When it’s not passive, but it *feels* passive

I did a bit of very informal research last night – almost the least formal kind possible, just above whatever anti-maskers are referring to when they say “do your own research”.

It involved searching for “passive voice” on Twitter and making notes of what people were calling “passive voice” when it wasn’t what grammar books call passive voice – that is, something other than what is described here.

It probably doesn’t help that many descriptions of passive voice say things like “put the verb in the passive voice” when, like many manipulations of the verb phrase in English, passive voice often involves 2 or 3 verbs (like “it has been completed”), and is often described with its semantic, aesthetic, or strategic characteristics (“unclear”, “wordy”, “obfuscatory”, etc.) rather than its more formal, nuts-and-bolts characteristics that set it apart from other “unclear” phrasings. Is it my imagination, or are ESL explainers a lot clearer about what passive voice is than ones written for native English speakers?

People seem to have taken certain of those semantic, aesthetic, and strategic characteristics and extended the meaning of “passive voice” to anything with those characteristics – or so I thought. It is actually mostly just about particular types of intransitive verbs, plus a few others with similar characteristics, as we shall see below.

I’ve written before about how certain terms of art in the grammar teaching/copy editing/annoying people on Twitter world have made the leap into popular discourse with some metaphorical extension. I just wanted to verify that my instincts about which features of passive voice were the basis for the extension were correct.

My methodology was extremely simple: search Twitter for “passive voice” and note the first 20 unique instances of people incorrectly identifying it. I didn’t include people identifying it correctly (which seemed like about half the time) or separate instances of people calling out the same piece of writing, as happened with this selection from the New York Times:

Then, I counted the characteristics of the verb phrases that were incorrectly identified as “passive voice”. Three major categories revealed themselves: unaccusatives, transitives, and others. The breakdown and some examples of each are below.

Unaccusatives

Just to recap, an unaccusative verb has a subject, but the subject is not an agent. In the example below, the cause of the explosion is not clear – the agent isn’t identified, and “explode” doesn’t imply that the building had a hand in its own destruction, unlike agentive, intransitive verbs like “swim” or “proceed”.

Called out as “passive voice” in the replies

Unaccusative verbs, like many passive voice constructions, are missing an agent, leading people to confused the two.

Whenever I found someone misidentifying passive voice, I replied:

And most people probably assumed I was a troll or a pedant (a venn diagram with a lot of overlap), but a few replied. Most who replied implied that it was the lack of agent that led to their judgment:

While a few backtracked and said that while it wasn’t technically passive voice, it shared some characteristics with it:

And one instructed me that I was wrong about passive voice in the first place:

So basically, the reasons that people have for extending the meaning of the technical term “passive voice” to unaccusatives fall into three categories:

  • Some extend the meaning in casual discourse while acknowledging that the technical meaning of “passive voice” is separate,
  • Some extend the meaning of the technical term because they aren’t sure of the technical meaning, and
  • Some have a concrete notion of the technical meaning that deviates from that of most orthodox English grammars.

Transitives

Here’s where I was initially confused, because I assumed that transitive verbs in the active voice are about as far as you can get from passive voice, both semantically and gramatically. I only found two examples of people calling transitive active constructions “passive”, but two is a surprising number when you’re basically counting the number of people who confuse a crocodile not with an alligator but with a flamingo. As we shall see, however, some transitive constructions are actually not far off from unaccusatives in obfuscating the agent.

The characteristic of “Israeli airstrikes hit Gaza” that the author thinks is passive voice is probably that, as with unaccusatives, the subject isn’t an agent here. The “airstrikes” aren’t the cause of their own “hitting” in the same sense as a structurally similar sentence like “Bob hits Jim” (“hit” polysemously indicating either intentional striking or mere physical contact). It’s not what a style manual would call passive voice, but like passive voice often does, this sentence conceals the agent. I would need to study this for real to find out, but I imagine people like the author of the tweet above would not call “Israel hits Gaza” passive, since it heavily implies intent (inaccurate as it may be to attribute motives to an entire nation), even though it has the same active and transitive structure as “Israeli airstrikes hit Gaza”.

Others

I found a few other cases of people calling out passive voice where it wasn’t that didn’t fall into one of the two larger categories. One simply had too many verb phrases for me to identify what the author was trying to call out:

And one where a prepositional phrase is called out as “passive voice”:

Note that it is technically possible for a prepositional phrase to contain passive voice (e.g. “in being attacked”), but not here.

Neither of these tweet authors replied to my formulaic query, so I don’t know what they considered passive about these passages/phrases, but my instinct is the lack of a blameable agent.

Conclusions

My previous assumption is partly verified here: People identify the “passive voice” where the agent is not obviously identified. This tends to happen most often:

  1. In sentences with unaccusative verbs
    • Examples: “die”, “explode”, “collapse”, etc.
    • But not unergative (i.e. agentive but intransitive) verbs like “retaliate” or “invade”
  2. In sentences with unagentive transitive verbs
    • Examples: “hit” (as in “the ball hit the ground”), “include”, “take (place)” etc.
  3. In sentences where agency is unclear
    • Examples: “My country is at war”, “I was in the middle of a crime influence network”

To me at least, this is more evidence for a “folk” definition of passive voice that is distinct from its technical grammatical definition. And although it may have seemed like I was pulling rank on some of the people above by questioning what they thought was passive voice (the implication being that it really wasn’t passive voice), I think we may need to get to a place where “passive voice” just means different things in technical and non-technical contexts and everyone generally recognizes that. After all, we don’t tolerate neuroscientists butting in on our telling funny anecdotes to insist that we aren’t technically “recalling” a memory, or classical voice teachers interrupting American Idol to take issue with Randy Jackson’s use of the word “pitchy” (note: I am old).

It is a bit of an issue if people use the term “passive voice” with its non-technical definition thinking that they are in fact using it with its technical definition (as in the tweet author who corrected my correction above), but perhaps each piece of jargon that undergoes this secularization has to go through a stage like that. Maybe there was a time in the lifetime of the verb “to short-circuit” when people weren’t aware that a person getting suddenly confused and an electrical circuit being shorted weren’t literally the same thing, and it took many years of the casual use in sentences like “he totally short-circuited” before people began to “rediscover” the technical definition.

I try to take a linguist’s perspective on such changes: notice but don’t judge.

What I have learned by looking at grades with basic statistics

At the behest of some congenial podcast hosts, I’ve been reflecting on how I got started analyzing the grades I give this way and why I continue to do it, almost compulsively, at the end of every semester.

I started using a few basic tools of analysis in Excel for my MA studies, and realized pretty quickly that although you could spend a lifetime exploring multivariate statistics, you can also do a lot right now with a half dozen or so functions on Google Sheets (mostly AVERAGE, CORREL, STDEV, TTEST, assisted by FILTER) and no more knowledge of statistics than you gained in your first year of undergrad.

I am decidedly not an expert on statistics, and especially not for someone who works at a university. I just started the habit of throwing the .CSV file that Canvas optionally spits out at you at the a semester into a Google Sheet that computes a bunch of things automatically, and keeping my eyes peeled for anything extremely odd or patterns across multiple semesters. In the interview that spurred this post (listen here), I compare my way of working to Black Sabbath in that I make the most of a small palette of tricks*, painting as much as I can in primary colors.

So just to provide a hook in case you want to start looking at your own course grades a bit more rigorously, below is a list of changes I’ve made in my courses based on things I learned by applying basic statistics to grades.

  • Most recently, I’ve decided to decrease the number of small assignments at the beginning of the course in favor of mini-projects mostly done during class time after I found that the number of 0s for graded assignments(the most common cause of failing my classes) have increased since our switch to remote teaching, and that they predict failure even on relatively lightweight assignments near the beginning of the term. I’m not sure that my approach will result in more completed assignments and fewer failures, but I did notice recently (anecdotally) that groupwork was more likely to be to be completed, and I would prefer that students who fail do so for poor work (and that they get the practice that even poor work requires, in order to do better next time) rather than because they just don’t check Canvas on weekends.
  • I’ve mostly eliminated extra credit in favor of flexible deadlines and accepting late work after finding that extra credit assignments didn’t correlate much at all with final grades. The negative correlation that I expected would have been due to students with lower final grades using extra credit to go from D to C, but I found that just as often, it was being used to go from A- to A. In either case, extra credit assignments are usually not designed with as much rigor as my usual assignments, so if they’re not being used for what I intended them for in the first place, I figured I should just replace them with the original missed assignment and save myself a bit of time, especially when lateness doesn’t seem to indicate problems with planning or time management as much as I had assumed.
  • Adding a bit of extra data gathering on top of my usual spreadsheets that Canvas spits out for you, I found that fasting during Ramadan does seem to adversely affect students’ test grades in some cases. I no longer work at the institution that required these high-stakes reading tests, but if I did, and I still taught evening classes during April or May, I would replace those exams with applied projects of some kind that could be completed at a more flexible pace.
  • I stopped making essays due in class on the first day of the week after finding that students who edit essays the day that they are due tend to do worse in the class overall, if not on those essays in particular. The flipside of that coin is that students who edit essays more overall and spread their edits across many days tend to do better, even on assignments that aren’t directly affected by the editing (In short, the timing and number of student edits on an essay shows more about their general tendency to do well in the class than their grade on that specific essay). I now put much more of the writing process during class hours and give more of the reflection process as homework.
  • On a related note, I give much more support now during the drafting and editing process and give explicit points for showing me how they’ve consulted a corpus during the editing process, left comments on each others’ drafts, reflected on those comments, etc., since I have consistently found that rough drafts predict final grades almost as well as final drafts do although rough drafts are worth far fewer points. This indicated a redundancy in my grading – I was essentially giving “A” writers two As for the work of one. Instead of focusing on perfect final drafts, I now focus more on the process of editing, checking, rewriting, and asking for help. (Not to mention that in the age of Google Translate, you have far less control over what forces have combined to produce the final drafts that you’re grading, and focusing on whether or not using Grammarly constitutes academic dishonesty takes the focus off of much more important skills.)

Last, I’ve updated the Google Sheet I dump all my classes’ grades into, and you can make copies of it for your own use here: https://docs.google.com/spreadsheets/d/1KuoWIlE1EQX-ZW6Mo6BUOgFm88kyT8g5YI544Lm_lUk/edit?usp=sharing

And here is a short video introduction:

*unfair to Black Sabbath, by the way – they wrote quite a spectrum of tunes, including, unexpectedly, the theme to Big Mouth.

Editorial copy of forthcoming guidebook The Cultures of Japan – Ancient Roots, Modern Blossoms (with markups)

Placeholder for jacket image

Yamanaka Village 山中村

The New Year is celebrated across Japan, but one village’s 大忘祭り (oo-wasure matsuri) is sure to surprise even jaded expats. Actually, they’ll need to come in consecutive years to see the real mind-bending charm of this festival, as it’s different every year. Scratch that – not only is the festival new every year, but the culture of the entire village gets rewritten every January 1, and the festival along with it. You see, the dwellers of this remote mountain village – its name literally means “the middle of the mountains” – get a whole new set of beliefs, customs, and even a new dialect to go with the new calendar to hang on the wall.

This won’t be obvious to a visitor at any time other than from December 31st to January 1st, because along with the new beliefs comes the idea that the people of this village have always believed those things, and along with their new customs comes the idea that they have been practiced since time immemorial. Unlike the practices in neighboring villages, or those of Japan as a nation, these are self-consciously reinvented every year, but like those of any place, they are considered ancient and immutable traits of the unique culture of the village and its people for the entire year that they are in effect.

This correspondent questioned this “always changing, always permanent” feature of the culture of the village to one local tea kettle artisan, and his response was at once frustratingly opaque and crystal clear. Indicating one of his kettles, he said, “A kettle needs certain parts to be a kettle. It needs a spout and a handle and it needs room for water inside. A tradition is the same — it needs commonality, uniformity, and antiquity. That is what we make when we make a new tradition.” It didn’t seem to make sense to him to think of “antiquity” as a characteristic of things that really are old rather than just one more feature that is subject to tinkering like a piece of handicraft.

NOTE FROM EDITOR:

Quirky, but maybe cut this one since it doesn’t seem part of any larger pattern / hard to categorize? Also doesn’t jibe with the “ancient roots” angle of the volume – JB, 3/22/2021

Kami-Sakai Town 上坂井町

Rural Japan is rich in dialects, and even experienced visitors often find themselves at a loss in conversation with locals outside of the major metropolitan areas of Kantō and Kansai. The rural town of Kami-Sakai (and here the word “town” is a bit of grandstanding – the entire county whose seat is Kami-Sakai has a population of just 1,532) boasts a dialect with some features to thwart even the most linguistically adventurous traveler.

The main distinguishing feature which gives rise to the others is a complete lack of negation. That’s right, negation, the crucial difference between “I know” and I do not know”, is missing in the local speech in Kami-Sakai, leading to some grand circumlocutions as visitors and locals alike try to dance around as simple an idea as “no”.

What follows is an example of a conversation that this correspondent heard just outside the train station:

Customer: “Do you have the evening edition?”

Kiosk cashier (nods): “We still have the morning edition, but…”

Customer: “Right. Will the evening edition be here when I get back? In an hour.”

Kiosk cashier (nods): “If so, it will be good, but…”

Customer: “You hope, but… I see.”

Kiosk cashier: “Perhaps you would prefer the morning edition.”

Customer: “Ah, a good choice for a Saturday. You see, the young remember these things so well.”

There are quite a few routine ways of negating-without-negating, as the reader can get a taste for above, and experienced speakers get quite skilled at “reading the air” around an unexpectedly oblique comment that redirects, rather than rejects, one or another proposition. In the conversation above, the newspaper in question does not publish an evening edition on the weekend, something that the cashier was able to make known to the customer without rudely (from her perspective) denying the existence of an evening edition on that particular day.

As a local oyaji told to this correspondent over a few chūhai at a local pub, this feature dates back to the Edo period as a way of currying favor with a particularly strict feudal lord. The peasantry made it a habit to avoid negation as a way to stay in his favor, which that lord and his successors grew accustomed to and even at one point enshrined in law. When the feudal system ended (interesting side note: most of the former peasantry chose a last name, 佐野 Sano, that includes a character from the feudal lord’s last name, suggesting that perhaps they had learned to appreciate his contributions to the local culture in hindsight), the dialect persisted. Local residents understand the larger regional dialect and what our readers may know as “textbook Japanese”, but they consider negation unthinkably confrontational for themselves and even a bit sneaky. As my new friend at the pub told me,

“Why do you want to put an idea in someone’s head and then at the last minute, take it away again? It’s like you’re running to fly a kite, and once it’s in the air so people can see it, you crash it into the ground. Negation [this word fell from his mouth with obvious discomfort] is a bait-and-switch. It’s a scam.”

When I asked if he thought Kami-Sakai’s unique dialect – which had come through the radio and television ages intact – would survive the Internet era, he cocked his head and simply said, “I hope, but…”

Shimogōri Town 下郡町

Nestled within the Iya Valley, Shimogōri Town is well-protected from both the literal winds as well as the winds of change. The culture of this area is among Japan’s oldest, in a national that prides itself on the continuity of its modern customs with antiquity. It has achieved this feat through means that will unfortunately be unavailable for outsiders to witness, as it involves draconian restrictions on contact with the surrounding world (including not only tourists from other parts of the world but even neighboring villages), and strict controls on the behavior of residents of municipality. Inhabitants of the town are, per the word of one recent former resident, not allowed newspapers, television, telephones, mail, or even food from the outside. The town is entirely self-sustaining in both physical and cultural resources. Children are raised in temple schools (Buddhism being the most recent cultural import deemed not morally corrosive to the community) rather than the school system seen elsewhere in the country, and the county and prefectural governments seem to have written off enforcing mandatory education, despite the protestations of human rights groups who decry the (presumably) low educational completion rate and (also presumably) poor hygiene of children born into the town. Indeed, according to the same former resident, the town is in all respects a time warp back to the closed-off sakoku period.

It is not clear how an independent observer can verify some of the claims made about the culture of Shimogōri, as the townspeople shrink from contact with outsiders, uniformily retreating into their homes and not answering shouts of gomen kudasai from the outside (the sliding doors lack surfaces sturdy enough to withstand a knock, not to mention doorbells). An attempt to visit Japan’s oldest local culture may highlight a feature of Japanese culture that is universal: fear of outsiders. Japan’s insularity may have the deepest roots of all.

NOTE FROM EDITOR:

One of our editorial staff, who shall remain nameless, had the opportunity to visit Shimogōri Town this past year, and found nothing as our “former resident” described. Instead, children attended school as normal, men and women alike worked in jobs typical of the countryside, and people consumed media as normal. Revise this section? Were we sold a bill of goods? – PK, 5/2020

NOTE FROM EDITOR:

I see we decided not to revise last year. A freelance correspondent visited the same town and found it not at all typical for the area, but an exporter of “knowledge work”, with men and women of all generations telecommuting from kibbutz-like group homes while their children attended holistic schools. Are we sure this is the same town we’re visiting every year? Let’s make sure before putting this in the 2022 edition – JB, 4/2021

When 0s really matter

I was having a conversation with a co-worker about students in the higher levels of our IEP, and how she can tell pretty early on in the term which ones are not destined to do well. I had the opposite feeling – I tend not to be able to tell how well students will do until the 0s start piling up, since as I have mentioned before, an accumulation of 0s on assignments for simply not turning them in rather than poor performance on completed ones is what usually sinks people in my classes. I don’t find that my students who fail the class are reliably less skilled at English (at least under certain conditions) than the ones who pass, which makes it hard to tell who will struggle from the first week or so. Instead, my students who fail tend to do OK on the assignments that they turn in, but turn in fewer of them, getting 0s for missing assignments. They usually have classmates with similar skill levels who get similar scores but turn everything in and pass.

(I should point out that I accept late work, and I don’t give 0s for trivial reasons like forgetting to write your name or sharing a document on Google Drive but not on Canvas, although I find those as annoying as everyone else. I do give 0s for plagiarism.)

But the question had me wondering if I should be paying more attention to 0s as they occur early on as signs of trouble. Are the 0s spread throughout the term, or in clumps here and there, or concentrated in high-value assignments around finals week? If the pattern of 0s tends to begin early in the term, I should probably pick up quicker on who is likely to pass, like my colleague, although based on scores rather than on direct observation. On the other hand, if 0s are spread more or less randomly throughout the class until really high-value assignments are due in the last 1/3 of the term, then I should continue to reserve judgment.

And when I say “reserve judgment”, of course I’m not planning on pigeonholing students as successes or failures based on a pattern observed in the first weeks of a course. I want everyone to pass, whether because they turn in assignments that reflect their ability and/or because they work hard to increase their ability, and hope I can distribute my attention in a way that facilitates that.

I decided to look at the numbers of 0s in the first 2 weeks, 2nd set of 2 weeks, and 3rd set of 2 weeks of my 6-week courses over the past academic year and see 1) if there are patterns in the accumulation of 0s that differentiate passing and failing students, and 2) how many 0s in any of those 3 periods predict success and failure.

I thought this question would be easy to answer, since Canvas allows you to view assignments in order of due date, but unfortunately the exported .csv files don’t follow that order. So instead of using a few functions in my already-created Google Sheets to just count the number of 0s between certain columns, I had to go into Canvas and count 0s for given time periods manually. For some reason, I enjoyed the drudgery of this.

First, here are the patterns of 0s across the term in both WL (written language) and OL (oral language) classes.

n=30
n=23
n=4
n=7

First, a pattern that I observed before continues to hold: all of the passing students missed <10% of their assignments, and all of the failing students missed >10%. It’s not shocking that students who missed more got lower grades, but it is interesting that 10% seems to be the dividing line – in theory, a student could miss 10% of assignments and still get an A- if they missed assignments with average point values and got perfect scores on everything else.

And something new that I can say thanks to this data is that this pattern applies even during the first 2 weeks. Students who miss more than 10% of assignments during the first 2 weeks are much less likely to pass, mostly because the pattern of missing assignments only gets worse from there. So to answer one of the questions that spurred this post, I should pay attention to this pattern early in the term.

I did a few t-tests to see if students who got 0s on certain percentages of assignments during the first 2 weeks had lower final grades, and the results are below, with means for the over/under groups.

WL 1-2WL 3-4WL 5-6OL 1-2OL 3-4OL 5-6
>0% 0s predicts final grades? (T-TEST)0.0340.0170.0530.0000.3060.462
no 0s (mean final grade)84.889.489.989.691.290.0
>0 0s (mean final grade)73.476.177.665.083.383.9
>5% 0s predicts final grades? (T-TEST)0.0240.0170.0590.0000.0240.245
<=5% 0s (mean final grade)84.889.485.989.692.790.4
>5% 0s (mean final grade)72.476.175.865.079.382.6
>10% 0s predicts final grades? (T-TEST)0.0000.0060.0010.0000.0010.046
<=10% 0s (mean final grade)84.784.984.488.990.588.1
>10% 0s (mean final grade)62.369.463.564.174.779.8
Values produced by t-test are in italics if significant at p=0.05 and bold italics if significant at p=0.01

To put the above in more digestible form, missing more than 10% of assignments predicts failure in almost every week in either written language or oral language.

Interestingly, missing any assignments during the first 2 weeks seems to make one quite a bit more likely to fail, or at least get a much lower grade. As you can see, the average final grade for a student in the written language class getting a non-zero number of 0s during the first 2 weeks is 11% lower than that of a student who turns in every assignment. For the oral language class, it’s 24% lower, the difference between a B+ and a D.

If you’re wondering if the numbers of 0s are simply a product of my giving more assignments at particular times in the course (it being easier to forget 20% of 5 assignments than 20% of 20 assignments), my numbers of assignments are pretty stable from week to week throughout the term.

And just to drive home the point that it’s not purely missed points that’s producing the above effect, here is a chart for just one of my classes of numbers of 0s and the numbers of points lost from 0s on average (that is, counting 0/10 as a loss of 10 points, but not counting 5/10 or 9/10).

As you can see, student miss more assignments and miss more points from those missed assignments in weeks 5 and 6, but again, missed assignments from the first 2 weeks better predict final scores.

Now that that’s out of my system, I will try to enjoy my break.

Teaching Linguistics for the First Time

I just finished my first semester teaching Linguistics to undergrads. My university (like many others) has no permanent Linguistics faculty, and my name apparently came up as someone semi-qualified who seemed like a good fit to teach Intro to Linguistics this fall in addition to my regular ESL duties. The class will apparently rotate among 3 of the ESL faculty including me from now on.

Some of you may balk at a non-PhD teaching this course, but I think having ESL faculty do it actually has some advantages, mostly in their classroom experience and a degree in pedagogy itself (as opposed to Phatic Communication among Adolescents in Balinese Coffee Farming Communities) which makes up for any shortcomings they (meaning “I”) may have in Linguistics knowledge. I found that generally, I knew enough to talk freely about the topics in our OER (https://essentialsoflinguistics.pressbooks.com/front-matter/introduction/) with a few accommodations towards different jargon, and was never at a total loss when asked a question in class (except on Neurolinguistics, which I can only talk about in very general terms with token “Broca’s area”s thrown in to up my credibility). I did sometimes come up short when asked about specific languages that I don’t speak (how Greek gender corresponds to that of Romance languages, for example) or about which languages in the world have a particular feature that had come up (“which languages have infixes?” – besides Arabic and Tagalog, my only answer is “Google it”). I hope someday to have the gravitas to be able to pass on questions like these with my teacherly authority intact. I suppose “professorial gravitas” must only be part of PhD coursework, since it wasn’t treated at all in my MA.

I was forced to review quite a bit of material that I hadn’t used (either in the classroom or in posts) since my MA, though, and I definitely have a more well-rounded grasp of basic Linguistics now than I did at the start of the semester, especially in areas that aren’t directly of use for ESL like the aforementioned Neurolinguistics, case marking in Old English, Speech Acts, or languages that American undergrads are more likely to know than ESL students are (mainly German and Spanish). I wouldn’t put my knowledge up against a Professor of Linguistics at a university with an actual Linguistics department, but I figure the curse of knowledge might actually make it harder for PhDs to meet undergrads where they’re at, and after all, I have a degree with a pedagogical focus.

The rest of this post is divided into “stuff that worked” and “stuff that didn’t work”. Enjoy.

Stuff that worked

Discussions

Each week had one module that covered 1 topic – week 5 was Phonology, week 8 was Syntax, etc. – and in each module, there were 3 main recurring assignment types:

  1. Quizzes (on Canvas, can be taken twice, no Proctorio or anything like that)
  2. Inquiry Activities (worksheets that combined closed-ended comprehension checks with open-ended applications)
  3. Discussions (mostly of the open-ended scavenger hunt variety)

I put my grades from this class into my usual spreadsheet for analysis, and it turns out that of these three, Discussions have the highest correlations with final course grades on average, and differences in grades on Discussions produce a statistically significant difference in final grades. I find this interesting since the Discussions are the least direct assessment of knowledge of the three, and also include points for responding to classmates, which are perhaps more indicators of personal involvement in the course than knowledge per se. All three had roughly the same amount of points, so it wasn’t weighting that made Discussions more predictive than other assignments.

As an example of what they tended to be like, here are the directions for our Phonology discussion:

Go to https://accent.gmu.edu/index.php. (Links to an external site.) Click on “browse” from the top menu and choose “English” from the list. You will see a list of samples of native English speakers reading from the same prepared text.

In one post, give one or more links to samples from this site that show allophones used in complimentary distribution. Point out where they are and do your best to describe the rule of their distribution. You don’t need to use the notation for derivations from the OER for this. (6 points)

Leave a thoughtful reply to a classmate for 2 more points.

Example post:

I found this guy from Pittsburgh using [p] and [ph] in complimentary distribution. https://accent.gmu.edu/browse_language.php?function=detail&speakerid=61 (Links to an external site.) 

He starts by saying “Please”, and the /p/ is pronounced [ph]. Later, when he says “six spoons”, the “p” in “spoons” is [p] instead of [ph]. I think the rule is that when “p” is the first sound in a word (or maybe a syllable – there are no words where “p” starts a syllable other than the first syllable in the word), it is [ph], and otherwise, it is [p].

Discussions were held on Canvas (i.e. all text) and on Flipgrid. For the next go-round, I may increase the amount of points given to discussions and reduce the amount given to the solo inquiry assignments.

Discussion Circles

Just twice during the semester, we had Discussion Circles, once for extra credit and once as a requirement in place of an asynchronous discussion. I consider these a limited success since, although they didn’t correlate strongly with final grades, I got to hear my students actually talking about the course material with each other, which is a window into their approach to and comprehension of it that I wouldn’t have had otherwise. As I often find with this kind of assignment, I also hear from people who are much quieter during lectures or who hadn’t attended a lecture in some time. I think it’s worth doing these a few times per course, even if they don’t have much validity as graded assignments.

Final Project

The final project, a 1000-word research report with accompanying presentation on a linguistic phenomenon of their choosing, was the last 40% of their final grade not taken up by assignments like the above. This was certainly worth having, as I saw enthusiasm and depth of understanding in the reports that weren’t really part of the regular assignments. Quite a few of the final reports displayed research skills that were impressive for undergraduates and, although no one ended up going through the IRB for the permission to undertake a project like this, insightful ideas for surveys and human subjects research.

Stuff that didn’t work

Phonetics and Phonology

As I mentioned earlier, the topics each had 1 module (=1 week) dedicated to them. This was fine for most topics, but the Phonetics module and the Phonology module could have been stretched – if I were teaching the class next semester, the first change I’d make is to allot at least 3 weeks for these topics combined. As it turns out, people are really not used to thinking of speech sounds in a systematic way, and the week after they’ve learned all this bizarre-sounding metalanguage like “biliabial fricative” and seen all those upside-down and backwards letters with odd diacritics in the Phonetics module, they have another conceptually difficult and almost as jargon-heavy module in Phonology. It would make sense to give them a little more time to digest these two, especially since many of their projects will need some phonetic transcription. Without enough time spent on these two topics, you might trick yourself into thinking students have got a handle on them only to see them writing things like “the is pronounced /the/” a few weeks later.

Nuts and Bolts

In addition to the topics that typically populate an introductory Linguistics course, students need the basic vocabulary to talk about language beyond what Linguists might assume undergrads have just as a result of graduating from high school. For example, in listening to my students’ discussions, I noticed quite a bit of confusion over what exactly constituted a noun – and I don’t mean whether a gerund should be considered a verb acting like a noun or just a noun like any other – I mean “is dog a noun or an adverb?”. This “basic metalanguage” module could be near the beginning of the course, and could anticipate other modules a bit with word classes (AKA parts of speech), syllable structure (i.e. “what is a syllable”), and the difference between a word, an idiom, and a phrase – but they’ll be using basic terms like these all semester, not just in the modules focused on those topics.

Final Project

The Final Projects are definitely worth doing for their self-directedness, research, and opportunity to show deeper understanding. On the other hand, this understanding tended to be focused narrowly on whatever their report was about – aphasia, the history of a particular accent, sound changes in loanwords, etc., and therefore wasn’t a reliable sign of competence in most of the topics that make up the course. Because so many projects were focused on pronunciation in particular dialects, it might behoove the next instructor (or me, next fall) to give a mini-project that is just about pronunciation sometime during the first half of the course (after Phonetics, Phonology and Sociolinguistics, and perhaps aimed at how particular features aid in characterization in a movie or TV show) and reserve the self-directed research project on a topic other than phonetics or phonology for the second half. That second project could proceed like the ones we did in my class did, with a project proposal and many scaffolded steps before the final report is due, but should probably also include a revised project proposal – many projects really needed more attention at this stage rather than the revision stage. As you might expect, students in their first-ever Linguistics course (and at the time of that the proposal is due, less than 1/3 done with the course) have little idea of what a good research topic in Linguistics is, and many need whittling down from something like “How does language affect culture?” to “What factors predict the adoption of a piece of slang in a sociolect into a standard variety?” Students also need more support in choosing sources, including a chance to submit a revised Research Matrix – quite a few resources that I saw in them, many of which ended up in their final written reports, were more of the listicle than academic variety.

Language and Woman’s Place

I had a few weeks where instead of one module that everyone needed to do, there were two modules of which they had to choose one (the full list of modules is below). One that I will probably strike from future versions of the course is Lakoff’s “Language and Woman’s Place”, a very specialized module in which students just needed to read the classic article (not the book, although they say very similar things) and do the usual quiz, discussion, and inquiry activity on it. It was interesting enough, particularly when students came up with their own examples of words whose male and female versions are quite different in connotation if not denotation (“bartender” and “barmaid” sparked a memorable discussion thread). Unfortunately, I also heard a lot about the article being out of date (not one of my students thought “She’s a professional” implied what Lakoff thought it did), and I don’t think it earned its keep. I had hoped that after the fairly abstract Language Variation, Phonetics, Phonology, and Morphology modules, this one might convince the students that Linguistics was also the study of problems in contemporary culture, and generate a bit more buy-in and immediacy for the Syntax and Semantics modules to come. I think something by Tannen might serve this purpose better next time, and without as much 1970s cultural baggage.

Unevenness in Modules

A list of the modules is below with their correlations with final course grades and mean scores. As you can see, some were much more predictive of final grades than others. Again, each module was worth roughly the same amount in points (even the optional modules), so an especially high correlation with final grades means more than that module’s score predicted better how that a given student would do on every other module and the final project, not just that it was a larger part of the final grade itself.

The following modules were required:

  • Semantics: 0.84 (87.1%)
  • Language Acquisition: 0.76 (78.0%)
  • Syntax: 0.72 (86.4%)
  • Sociolinguistics: 0.66 (78.8%)
  • Phonology: 0.60 (80.4%)
  • History of English: 0.53 (86.4%)
  • Language Variation and Change: 0.50 (85.0%)
  • Morphology: 0.47 (88.8%)
  • Phonetics: 0.30 (92.5%)

And these modules were optional – presented in pairs of which the students needed to choose one:

  • Language Policy and Preservation: 0.96 (81.4%)
  • Neurolinguistics: 0.92 (72.5%)
  • Graphemics: 0.86 (86.6%)
  • Corpus Linguistics: 0.76 (88.0%)
  • Language and Woman’s Place: 0.64 (89.0%)
  • Speech Acts: 0.45 (81.1%)

I’m not exactly sure how to go about doing this, but I’d like to have all of the modules be roughly equally predictive of final grades. I don’t like the feeling of having a few modules that have such little connection in terms of assessed skills to the rest of the course that someone could do very well on one and not on the rest of the course, or vice versa. Incidentally, I wonder if courses like Biology have this problem – if the module on Speciation for some reason predicts final grades much less than the one on Mitochondria, for example.

Class meetings

Classes are 1 hour (remote synchronous, but not counted for points) and the course is 3 units, so almost all of the work that makes of grades takes place outside of class time. It is hard to make the class meetings relevant under these circumstances. I tended to make them supplemental, for review games and discussions or a deeper dive into the material than the OER and other pre-made materials gave, but it turns out that making your class meetings interesting but clearly extraneous isn’t a good way to drive up attendance. Spending time on things not covered in the official course materials or assessed by graded assignments only cements in some Ss’ minds that the class meetings are not necessary.

Because I didn’t give points for attendance, I don’t have an entry in the gradebook to correlate with final grades to see if students who showed up actually did better on average. I have an inkling that they did, but probably for the same reasons that extra credit and tutoring often have positive correlations with final grades (as opposed to what you might expect if students who were behind were using them to catch up) – students who are likely to have good grades are more likely to do every kind of classwork, even extra credit or work that isn’t graded at all. I.e., a positive correlation between attendance and final grades wouldn’t prove that attendance caused high grades – just that students who tended to do well in class tended to be “good” students in general.

Failure by disengagement

As in many other courses, low grades for the course are primarily correlated with numbers of missing assignments, not graded assignments that got low scores. There seemed to be roughly three tiers in the class grade-wise:

  • Students who did almost every assignment well and got As
  • Students who did almost every assignment but not as well and got Bs or Cs
  • Students who missed many assignments and got Ds or Fs

No one who completed >90% of their assignments failed (regardless of whether they did a good job), and no one who did <90% of their assignments passed, making just turning things in the biggest hinge on which passing or failing turned. This bothers me because I want my grades to have construct validity: to measure competence in Linguistics. When a student turns in something that shows no understanding of the material, I can say more or less for sure “this student doesn’t understand Linguistics”. When a student doesn’t turn in an assignment at all, it might be because they don’t have notifications on for Canvas, they lost their phone, they got COVID, or their Internet went out in a winter storm (I reached out to students who were missing a lot of assignments, but part of the frustration of remote teaching is that the technology you use to reach students who are falling behind is often the very technology they’re having difficulty with). In any case, I prefer that students do well or poorly precisely in proportion to their competence in the fields assessed for the course, not for some hidden X factor.

Anyway, a lot to think about for next year, and I’m pleased with what my students and I accomplished together.

Remote teaching causes failure from missing assignments

Or, to put it in a less clickbaity way, “remote teaching in my classes over the past year is correlated with an increase in grades of 0 for assignments, which is highly correlated with a low grade in the course”. But too late — you’ve already clicked.

The past year has roughly divided my classes into half face-to-face and half remote, 101 students in the 3 terms ending with Spring 1 2020 and 76 in the 5 terms since (yes, enrolment is dropping here as everywhere). I mostly taught level 5, 6 being the last before ESL students in our program go on to their undergraduate or graduate programs, and mostly 4-skills classes except for the most recent term.

This situation has given me an opportunity to use my usual plug-and-play spreadsheets (updated for me but not for thee) for a bit of meta-analysis of the grades my students have earned over the past year. The tl;dr is just this:

Students in remote instruction were more likely to get 0s

Blue is f2f; red is remote. The two populations differ at statistically significant levels at p=0.05

Obviously, this is the headline and one of the main things teachers were afraid of before the switch to remote teaching: that students would simply stop engaging with the class and completing assignments. In my classes, that seems to have happened to a degree, with the average student missing 5.7% of assignments before the switch and 9.6% after.

If you’re wondering why I fixate on 0s, it’s because I almost never give a 0 except for a missing or very late assignment (plagiarism usually gets a chance to resubmit at a deduction). That means that students getting high numbers of 0s are simply not doing their assignments.

These 0s are enough to cause students to fail classes. The correlation of % of assignments with 0s to final grades is quite strong, -0.89 for only f2f students but -0.93 for remote students. That means that more 0s on assignments make everyone likely to get a low grade in the course, but especially remote students. I tried doing a t-test for % of 0s on the populations that passed vs. those that failed, and I couldn’t scroll far enough past all the 0.000s to get to a number.

Just for fun, here are average final grades for students who missed >5% of assignments, >10%, >15%, and >20%. First, only f2f students:

And remote students:

Which is just a “colorful” way to say that, in addition to remote students missing more assignments, the missing assignments were more likely to lead them to a failing grade.

Taking the Floor on Zoom

I’ve been remembering a JALT event I went to where the speaker, Thomas Farrell, showed us that I later learned was called a sociogram:

His were meant to illustrate the tendency of teachers to call on certain students repeatedly and ask many display questions (questions to which the teacher already knows the answer, like “Billie, how do you compute the circumfrence of a circle?”) between long periods of lecture. This is not what teachers recall the classes being like when asked: teachers usually recall themselves as having less teacher talking time (TTT) and asking questions to a wider and more equal distribution of students. But I’m straining to remember some of these details. I remember the graphs quite clearly.

It struck me that we need a new kind of graph to describe the relationships between the population of a class when it is conducted over Zoom. No longer is anyone “farther” from the teacher than a classmate or has some classmates “nearer” to them than others. Students are no longer part of a group that is sitting and facing one person who is standing and facing the opposite way. Tables (of any shape) or rows of chairs no longer create natural cohorts among the students who might whisper to each other, inaudible to students seated 2 rows away. On Zoom, any speech at all (indeed, any sound) instantly moves you to the “center” of the room and directs everyone’s attention to you in a very literal way – your face suddenly takes up their entire field of view (assuming you aren’t always in gallery mode).

Screen Shot 2020-08-17 at 10.36.16.png
This is a traditional classroom, accurate except in that everyone appears to be having a good time.
Screen Shot 2020-08-17 at 10.36.19
This is the closest I could come to graphically representing “space” in a Zoom classroom, although it fails because every participant is not equidistant. Everyone is the same distance from the speaker, who is not necessarily the teacher but whoever last made any kind of sound. It’s a panopticon in reverse – every prisoner can see the guard, and any prisoner becomes the guard automatically by saying anything.

I struggle to describe the effects on participants’ spatial (and everything that goes along with that) relationships to one another if only because Zoom meetings are not held in actual space. There are metaphorical positions that participants can take which are analogous to ones that used to exist in the physical space of the classroom, but one needs to do a Matrix-like bending and breaking of the normal rules of physics to continue to see Zoom as analogous to the classroom for very long. For instance, the teacher is usually given “the floor” by default in a traditional classroom, and the attention that he or she is given facilitated by his or her position at the front of the room, facing the students, who are seated with little space for adjustment facing the teacher. In a Zoom meeting, as long as the teacher is the only one to talk, this arrangement is re-created by the teacher’s appearing to face each student by looking at them through their computer screens. The students’ comparative lack of floor-taking ability can be enforced by muting of mics. But if any student speaks, rather than staying safely ensconsed in a crowd of sitting listeners, he or she is metaphorically transported to the front of the room and given a mic while the teacher disappears. As you can imagine, the Zoom meeting only retains the metaphorically spatial properties of a face-to-face class if we allow everyone to move at light speed to the podium every time they speak (or sneeze).

There are many implications for a non-lecture class, as most ESL/EFL teachers have, many of which are corollaries of the implications of Zoom for conversations generally. A non-lecture class (for example a student-centered seminar or discussion, or just a class where the teacher is less content to suck up all the oxygen in the room) is less likely to feature one participant (the teacher) as the focal point. Instead, many people might speak in turn, either in one large group or many smaller ones. Students and teachers will need to cede “the floor” to each other, as one does normally in conversation, dozens of times without a set sequence of speakers or a predetermined idea of how long they are to speak or sometimes what they are to speak about. All of this sets a non-lecture class apart from a lecture, in which one person speaks for a pre-determined amount of time and often with pre-determined content.

I happened to have speech acts and conversational analysis on the mind as the whole world turned to this new medium of instruction, and put some of my initial thoughts about Zoom for remote instruction in a video you can see here:

To be honest, I started writing this blog post soon after that, and then promptly got lost in other work and forgot about it (and the blog entirely) for about 3 months. When I remember that I was writing a blog post about this, I set out to find out a bit more about speech acts and conversational analysis and was recommended an article which I’ve quoted extensively below. Basically, I was looking for ways in which “taking the floor” in order to speak was different on Zoom than in a real-life classroom, and I found what I was looking for.

The recommended article in question is an attempt to systematize turn-taking in conversation, but the assumptions that it makes about normal conditions under which humans converse bring to light nicely the strangeness of trying to conduct a student-centered class on Zoom. Below, I’ll list some quotes with commentary.

(a) If the turn-so-far is so constructed as to involve the use of a ‘current speaker selects next’ technique, then the party so selected has the right and is obliged to take next turn to speak; no others have such rights or obligations, and transfer occurs at that place.
(b) If the turn-so-far is so constructed as not to involve the use of a ‘current speaker selects next’ technique, then self-selection for next speakership may, but need not, be instituted; first starter acquires rights to a turn, and transfer occurs at that place. (Sacks et al., 704)

The above quote sets out some basic rules of turn-taking in conversation, namely that the speaker can choose (through a variety of methods, for example asking a question to a specific person) the next person to speak, and if not, anyone can “take the floor” by filling the conversational void. One aspect of the flatness that Zoom brings to conversations is the elimination of several nonverbal ways for the current speaker to nominate the next speaker – by physically approaching, gesturing, making eye contact, or turning towards, to name a few. The process of nominating the next speaker therefore becomes more deliberate — few options are available besides explicitly naming the person who is to come next. It is possible for a next speaker to be implied by an ongoing back-and-forth, but at any point the ball can be dropped with a plausible ambiguity as to who was meant to speak next.

I have had many interactions on Zoom where I seem to be having a back-and-forth interaction with one student only to have them suddenly stop responding. It is not that they turned off their camera or went to the bathroom — turn-taking in Zoom is just such that fading into the background and letting questions dissolve in midair is a possible conversational “move” in Zoom that isn’t as possible in 3D space. From each student’s perspective, after all, I am never “facing” any of them in particular, and if I haven’t said their names in a few sentences, it’s easy to revert back from me addressing one of them to me addressing all of them. I have never had a similar back-and-forth in a physical classroom that simply ends with the student looking at me and seemingly not recognizing that the baton has been passed to them. Zoom always allows for plausible deniability as to whether nomination as “next” was made.

Overwhelmingly, one party talks at a time. This fact is provided for by two features of the system: First, the system allocates single turns to single speakers; any speaker gets, with the turn, exclusive rights to talk … Second, all turn-transfer is coördinated around transition-relevant places, which are themselves determined by possible completion points … (Sacks et al., 706)

Zoom both facilitates breaking this one-at-a-time rule and strengthens the penalty for violation of it by allowing the floor to be seized by any interruption, not matter how trivial or tertiary to the conversation the interrupter, or forcing any unintentional interrupter onto the floor, giving them the full spotlight and forcing the microphone into their hands. A lecture in physical space can proceed with only slight difficulty if someone in the nosebleed seats whispers to someone else, if someone coughs, or if one student in the crowd keeps raising his hand and repeating “Professor? Professor?” In a Zoom meeting, on the other hand, a student’s allergies force every other student and the professor to watch and listen to every one of her sneezes with monopolizing focus. The “floor” is liable to be seized rather than ceded, and not necessarily with the intention of the person seizing it (usually not, in my experience).

Who speaks how much to whom in the group is a ‘brute fact’ characterizing the actual present situation. Speaking takes up time. When one member speaks, it takes time and attention from all other members of the group, some of whom may want to speak themselves. To take up time speaking in a small group is to exercise power over the other members for at least the duration of the time taken, regardless of the content. It is an exercise of power that may not coincide at all with the status position of the individual based on outside criteria, or even on special criteria developed within the group …Within the small group the time taken by a given member in a given session is practically a direct index of the amount of power he has attempted to exercise in that period. (Bales; 1970:62, 76-7)

This was the closest I could find to a definition of “taking the floor”, which when you look at it analytically, seems pretty selfish, although it happens anytime one of us speaks to another. I just want to point out that the “exercise of power” is even more literal and imposing on Zoom than it is in real life. A person talking to you doesn’t necessarily demand that you look at them or prevent you from seeing other members of the conversation, as happens in active speaker mode on Zoom. As pointed out before, this happens even when the floor is seized unintentionally.

Further, in being compatible with differing numbers of participants, [the turn-taking system] is compatible with varying numbers of participants within any single conversation, since there are mechanisms for entry of new participants and exit for current participants (though we will not describe them here).

Though the turn-taking system does not restrict the number of parties to a conversation it organizes, still the system favors, by virtue of its design, smaller numbers of participants. (Sacks et al., 712)

The number of participants greatly encourages the “fizzling out” of conversation that I mentioned above, as strategies for nominating the next speaker are constrained by the uselessness or unreadability of physical gestures. Unlike in physical space, a Zoom conversation allows for no space-facilitated mini-conversations to emerge, except explicitly by forming breakout rooms. It is impossible for a student to take the local floor of just his or her table or group without taking the entire floor of the classroom.

Zoom facilitates much larger numbers of co-conversationalists, but keeps flat the ability and necessity of taking the floor for the entire group, allowing for “exercise of power” by each member over larger numbers of people as the number of co-conversationalists increases. It is as if, in a fight, the firepower that each participant carried increased with the size of the battle – a fight between 2 people has knives, a battle of 10 gives everyone guns, and a war of 100,000 gives each participant a nuclear bomb. This phenomenon, in addition to explaining why students might be less reticent to ask questions in a 20-member classroom than a 100-member Zoom lecture, might also explain some of the appeal of “Zoom bombing” – an exercise of power that affects 100 people with exactly the same force, as opposed to walking into physical classrooms and just shouting expletives at people who happen to be standing near the door.

Thanks for reading this far. My takeaway from looking at conversational analysis on Zoom is that “taking the floor” is more explicit and more powerful (which may make initiating a conversational turn more appealing or less appealing to our students, who are doing this in a second language as well), and conversation moves that can be made subtly in real life need to be made explicitly through a computer screen.

References

Bales, Robert F. (1950) Interaction process analysis; a method for the study of small groups. Cambridge, MA: Addison Wesley.

Sacks, H., Schegloff, E. A., & Jefferson, G. (1978). A simplest systematics for the organization of turn taking for conversation. In Studies in the organization of conversational interaction (pp. 7-55). Academic Press.

The Most Native Speaker

1. Lingua Prima VR

Early versions of what would become Lingua Prima VR simply applied existing paradigms of commercially available language learning software to a VR system. That is, a combination of flash cards, example sentences, game-like drills, and live conversation partners were projected onto the screens of a virtual reality headset. These were financially successful enough to convince software developers that this was a niche worth investing in.

Of the handful of companies investing in the next generation of VR language learning software, an Italian firm, Spedizione Software, was the first to bring on a team of academics, led by language learning theorist Geo F. Jensen from the University of Nottingham, from planning stage. As had been demonstrated by previous generations of Rosetta Stone and other applications, the visible involvement of linguists (although which particular linguists mattered less) was highly marketable, and the resulting higher retail price sometimes justified itself as a Veblen good among high-status customers, customers who Spedizione correctly predicted were also more likely to own VR headsets.

The only official direction given by the company to Jensen and his staff was that the product would be promoted as letting users “understand the language just like a native”. Obviously in retrospect, the directive, which may have had the hyperbolic, winking intention of something like “World’s Best Coffee”, had unique gravity (perhaps amplified by irritation at being treated as mere catch copy) to a team of academic linguists. The first day working on their own, they decided unanimously to completely scrap any and all nods toward “gamification” of the language learning process: levels, puzzle-solving, and indeed discrete language items of any kind, from single words to isolated grammar points, absent as these are in the learning process of almost all “native” speakers of any language worldwide. They then set about two main tasks: One, planning a route of exposure to authentic learning situations; and two, defining the “native” whose speech would serve as a target for users.

One task requiring much more focus than Spedizione intended was defining the “native” – in real rather than abstract terms. After four all-day meetings establishing criteria for nativeness, and (on Spedizione’s insistence) some accommodation towards marketability, the team settled on the idea of using a speaker from a major city in the American Midwest. Two of the American team members flew to Lincoln, Nebraska with the intention of finding candidates, and were fortunate enough to find an appropriate selection on the first day of interviews. That person, and the English speaker who eventually served as a model for about 95.4 million later users of Lingua Prima VR, was Nhi Nguyen, a student at the University of Nebraska, Lincoln (who would appear in promotional materials as “Tina Nielson”).

The necessity of a VR reenactment of her life product came about from a frustrating finding from neurolinguistics, that instantiation in the brain of language varies to such a degree from person to person that no Matrix-like shortcut of getting linguistic knowledge into every human brain was ever going to be feasible. Thus, the only route available for technology to reliably facilitate language ability in a varying group of people was to go the relatively old-fashioned route of through the eyes and ears.

As for the reasons that a real English speaker to serve as a model was required, the development team had determined that the only realistic means by which users might achieve the native proficiency level demanded by Spedizione was a virtual reality experience of a native speaker’s life, from early childhood to adolescence. That is, users would experience language input sufficient to produce native-like understanding exactly as Nguyen had experienced in producing her native understanding. Nguyen’s life would serve as the guide for the kind and variety of input necessary to produce a native speaker. To that end, a virtual reality experience was constructed based on a combination of interviews, memory recall through hypnosis, and thorough reconstruction of her childhood home, school, and the Gateway Mall in Lincoln. Efforts at anonymizing Nhi’s memories focused on changing her family’s appearance in the virtual world to match the image that consumers across the world would likely have of the American Midwest, as well as changes in verbalizations of her name to a user-supplied pseudonym.

2. Nhi

Nhi Nguyen didn’t have regular contact with consumers of the product her life had been a model for until she took her junior year to study abroad in Guangzhou, China. There, she was confronted with the full spectrum of exposure to her own life – people who were curiously knowledgeable about Nebraska’s weather, people who remembered the feel of the wood grain of her desk next the class cockatiel in Mrs. Wong’s 2nd grade class (and what had been etched into it by its occupant the previous year), and people whose memories of a particular softball practice in July drove them into a 2-hour sullen silence. She learned to moderate her normally jaunty personality to brace for these impacts of unexpected familiarity, which were somehow far more alienating than the culture shock she was also experiencing.

So when Nhi joined her History of Communication seminar the first day of the new semester at the University of Guangzhou, she tried to see in her classmates’ faces how much of her life they had experienced firsthand. From just walking around town and getting to know her neighbors, she knew the signs of a VR returnee (someone who lived their entire adolescence in a reconstruction of her childhood) – ostentatiously Chinese fashion choices coupled with eyes that always seemed to be scanning and verifying their surroundings. This was because Guangzhou was almost as foreign to them as it was to her (all of them effectively having been raised in Lincoln), and they unlike her felt the pressure to hide this fact. It marked them as not authentically Chinese and much wealthier than their peers – an extreme version of the kids who didn’t go on the school trip with their friends because their parents took them to Disney World. Those who had only visited her life for an hour or two a day, after the hours of homework that Chinese kids all had to do, usually seemed more relaxed and comfortable in their skin, and she knew they looked down on the returnees. Those who had spent no time at all in VR and had instead tried to learn English the old fashioned way usually looked nervous too, but more the nerves that come with being 18 in a room full of strangers (i.e. the experience of every university freshman) than the stress of trying to hide your eerie otherness.

Nhi thought a few of the dozen students present were probably VR returnees on first glance. For the moment, since they didn’t know her background, she was fine with letting them feel uncomfortable while she pretended to be a run-of-the-mill exchange student from no place in particular, USA. She unassumingly took a seat around the conference table and waited for the professor to finish connecting her tablet to the projector, confident in her ability to make a good first impression and start making friends.

The professor sat down at the head of the table and began to speak.

“Yeah, so, it’s nice to see everyone here today, in our seminar, so… if you’ve got the wrong class, this is History of Communication seminar, so, you can go down the hall if you meant to go to Communication Technology lab. It’s just down thataway.”

She pointed out the door, but Nhi had lost focus on her words. All Nhi heard after the first, aggressively articulated “nice” were a series of hard Midwestern Rs.

“Ok, so, I guess you’re all here because it’s where you’re supposed to [she pronounced it ‘spowsta’] be. Ok, so, let’s just go around and introduce ourselves.”

The first boy to her left, who Nhi had thought was probably an old-fashioned English learner, stood up and addressed the class in exactly the same accent.

“Yeah, so, I’m Jim Chen from Foshan. I don’t have a major yet. That’s all.”

Nhi tried hard to hide the tension now gathering around her ears as the next student stood up to speak.

“Yeah, ok, I’m Sarah Chen, I’m from Foshan too. I guess I think I wanna study Poli Sci. Nice to meetcha.”

Sarah Chen waved at everyone with her elbow tucked into her hip, just the hand flapping around, and sat down. Nhi had gone through a phase in 5th grade when she tried to re-feminize all of her gestures and mannerisms after becoming extremely self-conscious about tomboyishness. This wave was an expert, unselfconscious recreation of her wave from that period, ages 12-14. Even worse, Nhi had met enough people in Guangzhou to know how they usually pronounced the last name “Chen”. It wasn’t supposed to remind you of Cheese Frenchies. The student to Nhi’s left stood up.

“So, yeah, I’m Nh…  Nancy Li. Nice to meet everyone.”

Same wave. Nhi was next, and she felt like a spy about to be exposed and quickly gunned down. Only instead of guns, it was a roomful of strangers that she knew nothing about knowing everything about her life from her near death in discovering her walnut allergy to the fact that she tells people she peed herself on stage at a piano recital once when it was actually three times, and she had to wipe down the bench the third time.

Nhi stood up and faced everyone solemnly.

“Dear classmate. It is good, nice to meet you today. Please let’s be friends each other.”

People looked confused, but Nhi figured she had dodged that bullet at least for today. She was fine with everyone thinking she was just learning English, especially if the alternative were everyone knowing that their adolescence had been modeled on hers.

After the seminar ended, Nhi rushed home to watch English learners on YouTube and practice her cover for the next 9 months.

Oculus Quest 64GB VR Headset