Gender in job ads, and subconscious bias in language

You don't know what you are saying.

Sure, you think you know. You've chosen the content of what you want to say. You're thinking about what linguists call the "propositional content," or the "message" of what is being said. The fact is, though, that you're saying so much more. The complexity of language carries all kinds of information about social alignment, and the individual's stance in relation to social groups and in relation to larger discourses in society (other times the same topic has been referenced).

Yesterday I ran across this link through @geardrops on Twitter. I was fascinated, but not at all surprised, given my background in linguistic anthropology. The article talks about a study in which it was demonstrated that job postings contain language that is "gendered," or biased toward either a female or male expectation - and that potential applicants can feel this when they read the postings, and gravitate toward the ones that match their gender. Wow, right? Interesting. Within three minutes of having picked up this link, I ended up in an argument about whether we had left behind bias in the workplace, and whether a person could be considered sexist for writing an ad in this way, and whether or not "gendered" language was just something made up by academics that had no basis in reality. Geardrops' conclusion, which I loved, was this:

"Hey guess what language bias exists and is subtle."

That's right. The social messages in language are not just subtle, they are subconsicous. So let's look at what that means for a second, starting with the question of politeness.

If you ask someone, "You need to ask your professor for a pen. What do you say?" that person will give you an answer, but chances are it won't be exactly what that person would say if you recorded them in the situation. It will be - and this is important - what they think they ought to say. This is a trend backed up by all kinds of politeness research: live recordings will get you different results from what people say they do. In fact, I found precisely this in my own research. When I studied polite and casual forms in Japanese, I asked three teachers how much of each type they thought they were using in the classroom, and each one said they used formal forms most of the time. When I went in, took a videotape, transcribed and counted these forms, it turned out that they were using formal forms 35% of the time.

I'm sure you can see the difference.

The propositional content of what you say is consciously chosen. But for the most part, the manner in which you say it, and the social messages conveyed, are outside of your conscious control.

Can we say that there is "fault" involved, or accuse a person of being sexist, because of what they express in a subconscious manner?

Well, yes and no.

I often imagine language use within a culture as an enormous piece of fabric. Each person's contribution is a tiny thread within that fabric. The color of the fabric varies depending on which part you look at, but each thread will tend to be roughly the same color as the ones around it. We don't speak in a vacuum. The patterns of speech and expression that we use are learned from those around us, and the more closely our color matches the surrounding pattern, the more likely everything is to appear normal. Sexism, racism, and any other -ism that you might care to identify are mostly built into the fabric. If you're a member of an insider group, chances are pretty good that the language use patterns that show bias will be difficult to see. It's the whole "by saying mankind I mean all people" thing. The distinct color of the thread containing the word "man" will be more easily recognized by someone who doesn't match it - i.e. a woman.

I think it's more effective to talk about "gendered language" than "sexist language" because the latter implies intent. Intent is a tricky thing. In our society these days "bias" itself is seen as being wrong - which is I think a good thing - but what it means is that people can get attacked for subconsciously engaging in the fabric of the discourse around them. Should they be blamed? Probably not. Engaged with, probably yes. Every time we question biased language we're acting on the front lines of societal change. Increasing our consciousness of these mostly-invisible markers is the way to get people to notice biases they don't mean to convey, and act to change them.

Here's an important point, though. Social language is not a plus-or-minus proposition. You can go one direction or another on the road, but you're still on the road. We will always, always mark our social position, our posture, relative to others. We will always express our membership in social groups. We will strive to distinguish ourselves from groups that we don't want to be a part of by emphasizing our membership in other groups. I don't see "gendered language" ever going away; I do, however, see it changing.

Language is something that reinforces itself, and at the same time changes itself, every time that it is used.

So to be a positive force for cultural change in the world, think about increasing consciousness - your own as well as that of others. Go easy on yourself - be aware that you yourself are constantly enacting culture and social alignment without thinking, and you won't be able just to "stop." Don't blame yourself terribly if you find yourself doing something you don't want to do - just think through it and try to exert your conscious will to change it. Be willing to engage with others. Be willing to question yourself.

And next time you're looking for a job, be aware of gendered language. Be aware, too, that the person who wrote the ad probably didn't try to exclude anyone. Choose not to let that shut you down.

Change takes time, but it's worth the effort.

More thoughts on Language and Thought, and gender

I've been sharing multiple links recently about the relationship between language and thought. One is this one from the Wall Street Journal; another is this one from the New York Times. Here's a quote that I found particularly interesting:

Some 50 years ago, the renowned linguist Roman Jakobson pointed out a crucial fact about differences between languages in a pithy maxim: “Languages differ essentially in what they must convey and not in what they may convey.” This maxim offers us the key to unlocking the real force of the mother tongue: if different languages influence our minds in different ways, this is not because of what our language allows us to think but rather because of what it habitually obliges us to think about.

This brings us (and Guy Deutscher, the author of the NYTimes piece) back to issues I've discussed like grammatical gender and relative versus absolute direction, as well as others like the defining of time periods (past, present, future). Here's another terrific quote from the article:

When your language routinely obliges you to specify certain types of information, it forces you to be attentive to certain details in the world and to certain aspects of experience that speakers of other languages may not be required to think about all the time. And since such habits of speech are cultivated from the earliest age, it is only natural that they can settle into habits of mind that go beyond language itself, affecting your experiences, perceptions, associations, feelings, memories and orientation in the world.

This really spoke to me. As a speaker of French, I have always found it difficult to remember the grammatical gender of words. I know a large core number, but once I start getting to the periphery of that, I start having to guess. It's so frustrating! For the first time, during my visit to France this summer, I started having a glimpse of the worldview that lies behind knowing all these grammatical genders - the fact that when a French speaker (or Spanish speaker, etc.) looks at an object, it simply appears inherently masculine or feminine. We English speakers have an easy time differentiating between events in the past, the present, and the future, which our language requires us to specify.

While I'm not sure I could attempt it without living in France, I got a glimpse of what the world might look like if I thought in this way. It really stretched my mind into a place beyond where it had been before, and that's one of the most rewarding feelings I can imagine. Perhaps that's why I try to share it with other people through my stories.

The language you speak influences how you think!

I'd been looking for a good way to introduce the work of Stanford Psychology professor Lera Boroditsky to my readers, and today I found it in this fantastic article that appeared in the Wall Street Journal.

Yes, the language you speak does influence how you think.

I think the first people to put forward the idea that a link existed between culture and language were the anthropologists Franz Boas, Edward Sapir, and Benjamin Lee Whorf. The last two of those are associated with the Sapir-Whorf hypothesis, which holds that differences in the way language encodes categories etc. affect the way speakers of that language think. There's been a lot of argument about this idea since it came out, and certainly one can argue against a strong form that says language restricts thought. However, Professor Boroditsky has been exploring exactly what the consequences of certain languages are for thought - and in fact, she's finding there is a huge influence between the two. Russian speakers who have more words for blue than English speakers are in fact better able to distinguish between colors of blue. Speakers of different languages conceptualize the "direction" of time differently: left to right vs. right to left vs. below to above. And those who use absolute direction instead of relative direction (N/S/E/W instead of Left/Right) use those directions to organize things chronologically as well as to think about spatial positioning.

I encourage you not only to read this article, but to think about how this affects your writing. What kind of society have you created? What language do they speak, and how does it frame its categories? Once you've figured that out, ask yourself how that categorization system might be further generalized into the society, its beliefs and its behavior. Language does affect thought - so something that shows up in the language should be evident in basic concepts of reality like how many different colors of blue there are, or whether something hanging on a wall is actually "on" it or "up" it. What kinds of distinctions might your people draw that we might not be familiar with?

When I create my societies, I like to see how far I can push these categories and the judgments surrounding them. In doing so, I hope not only to tell a great story, but to show my readers a new way of imagining and conceptualizing the reality that we all share.

I hope you find this article as inspiring as I did.

How Language Reflects Good and Bad in the World

Here's a really interesting link that I ran across on the UC Berkeley Found in Translation blog. A study has shown that there's more good than bad in language, in much the same way that there's more good than bad in the world (trust me on this one; read the article if you're curious).

So maybe there is no universal grammar...?

Here's an interesting article I got from Marian at the Analog forum, talking about linguistic diversity and possible explanations for patterns of human language, including the weakness of so-called language universals. As a person who always maintained a degree of skepticism for Chomsky's theories, I find these ideas compelling:

http://www.newscientist.com/article/mg20627621.000-language-lessons-you-are-what-you-speak.html?full=true

How language learners speak

Just a short post for tonight, so please do ask me to follow up on this topic if you find it interesting. I was out on the Absolute Write forum today and the topic of language learning came up - one of the writers there was working with a character who spoke Arabic but was learning English, and trying to figure out how to portray her speech.

A number of people on AW gave advice about the sound structure or grammatical structure of Arabic - and it's true that the structure of a person's native language has an influence on the way that person speaks a new language. Others suggested that this writer go out and find a native Arabic speaker learning English to get a good language sample - certainly an excellent approach.

But there's more to the issue. It reminded me in some ways of the discussion of dialects we had here on this blog, because portraying speech errors in a sensitive way is equally difficult. The moment I think about altering spelling to reflect pronunciation errors, my internal alarm bells start going off. You can always go with word order, grammar and vocabulary usage, which is somewhat easier (and research on language learning sequences can help guide you on that). On the other hand, I had a character recently who gave me a lot of trouble, because he was supposed to be a native Chinese speaker who spoke perfect English but was putting on a strong Chinese accent. And every time I tried to contract his syntax, I felt I was losing the depth I wanted this character to have - that he was becoming to the reader what he was only pretending to be. In the end I went with greater language complexity, and a description of his accent.

The one thing that no one had yet addressed at AW (before I got there :) ) was discourse - the higher-level strategies that language learners use to get through a difficult interaction.

The first and foremost approach that a language learner will take in a tough spot is silence. I've been there - listening and listening but unable to respond. People often choose to stay silent in order to avoid mistakes. Another type of silence strategy is avoidance. Language learners will try to avoid grammar areas that cause them trouble, or vocabulary they don't know, by talking around the problem area. Avoidance also means that very often in conversation, language learners will make abrupt topic changes - when their resource pool for a particular topic runs out, they will switch to another (often without warning or explanation).

I would just encourage anyone dealing with language learners to remember that silence and avoidance are strong learner strategies - and best of all, they are among the easiest to portray in text without the risk of inadvertently ridiculing the learner.

More about Designing Languages

After my post on designing languages I had comments (here and on the Analog forum) from three people, which I thought I'd follow up on tonight.

Bill Gleason asked about translators.

A universal translator like the one in Star Trek is effectively a deus ex machina, an intervention on the part of the designer to allow us to bypass the real problems of different languages.

A mechanical translator can do a lot when it comes to predictable language structures - finding word boundaries by identifying repeating sets of sounds, tracking the structure and the sequence of those words. A really snazzy neural-network-style translator would probably also be able to do a good job of tracking phrases that repeat, and be able to do fuzzy categorization of words so it could capture exceptions to rules and things like that. It would also be able to do meaning to a certain extent, but I'm not sure how it would push past literal meaning to social meaning, for example. I suppose it would have to depend on how high you assumed the technology was.

I think it would be really interesting to have a translator machine that was really helpful but also limited in predictable ways. One issue a really "universal" translator would have to address would be how to incorporate information from the surrounding visual and social context - and that would be very tricky. Certainly it would require a device that could follow things like eye gaze and body position.

Bill also asked about body language. I talked about that yesterday so I won't go too deep, but if you assume that two groups of aliens are each able to identify discrete objects in their environment, then I think there would be a high likelihood that pointing would be used by both. Beyond that, a human trying to interpret alien gestures would need to be very sensitive to surrounding context and remember not to make assumptions about natural meaning for a given gesture. A shoulder-lift might not mean the same as a shrug, or possibly the alien might use an eye-gaze gesture similar to ours, but that is not accompanied by a shoulder motion. I don't think it's too far out to have two alien groups reach a basic level of mutual comprehension, but watch out for places where categories of objects are defined differently.


Tom Ligon brought up two very interesting ideas: a language based on bioluminescence, and a sound-based language with such vast range that human sounds seem undifferentiated.

Fireflies communicate by bioluminescence, but in a very simple way ("Where are you?" "Here I am!"). For a species with such a language to be able to grasp that humans are communicating, they would have to have an awareness of sound and its potential for carrying messages. They might even have unique alternate means to communicate via sound, as humans have used Morse code on Aldiss lamps, or semaphore (with flags).

As far as the language with vast range is concerned, that's a very interesting one. Human babies must gradually learn what not to pay attention to in their environment, so as to make communication easier without distraction, and so I think this is a completely solid concept. A human would have a tough time with this initially - it makes me think of H. Beam Piper's Fuzzies, who spoke with supersonic pitch. But given a good receiver, a lot of analytical work, and an appropriate sound generator, they might do a good job of speaking the range-language. Among range-language speakers, I think it would be a question of special individuals being able to grasp human language sound concepts. I've met many people who aren't able to hear sound distinctions that differ from their own native language - but on the opposite side, there are always exceptional individuals who are able to hear more subtle sound distinctions.


Greg Ellis peppered me with examples - what an exciting range of species! Look for them in his FOTS universe. The first place I'd start, given a large number of groups like this, would be sound design based on physiology, and believe it or not, the minute you name a group, you've already started language design. I'm going to comment a little on each group he gave me (except two he called the Swarm and the Machines), and I hope this will give a sense of how I get started when thinking through alien languages.

Saurians, Dwa'Kim
It might be good to ask whether dinosaur-like aliens have flexible lips, for example, because inflexible lips might restrict their ability to create consonants like p/b/w/m, or vowels like o/u. It looks already as though the Dwa'Kim have such flexibility - that could be used as a distinction between the two languages.

Jareen'qwar
My question about caterpillar aliens is whether they have a hard palate, and whether they have a nose. Both are useful to have, the palate being good for consonants like j (assuming it's pronounced "dzh") and the nose for sounds like n. The species might well have a different way of re-routing the air stream to produce a different resonance, the way we do with our noses. The name of the species as given here would actually be possible to pronounce without a palate if the j were pronounced like a y, and the q low down in the throat. But generally speaking the transcription of sounds would have to be considered approximate.

Rrowlk'aa
A doglike group. I've got the vibe. It looks like these guys are going to have some really cool growl-like consonants, or multiple qualities of "r." Does the apostrophe mean a glottal stop, a plosive consonant, or some kind of dropped sound? These are possibilities that might help the language expand.

Kahs
A more humanoid group, and their name is short, so lots of directions to go with this. I'd be looking to culture and social situations of language use to find out more about how the language is used.

Kon'ta Py'ron, Q'Tez, Grymphon
I'm grouping these together because they are relatively insectoid, with hard body parts, which suggests to me that they'll have a lot of means for clicking, creating hums and vibrations, and potentially whistling, all of which are difficult to use English letters for. But for clicks you can always turn to k/p/t/! , and for hums or whizzes you can go for s/z/f (or ph!) Whistles might give you vowels. I wonder a little about "n" and "g" for these groups, but this is all about getting creative - these might indicate sounds created by a combination of hum and click. Think through the type of mandible structure they might have, and whether they would have rubbing surfaces on their legs or other types of sound-producers they could use for language.

And that's it for my thoughts! Thanks for sending me the questions, guys - and I hope these examples have been thought-provoking for other people facing similar challenges.

Want to design a language?

I'm back after a great trip. Great adventures seeing sea lions and seals and kelp (in nature and in the aquarium) and playing with my parents and my kids and my brother's family too. Then I came home to the news of one rejection and one honorable mention from Writers of the Future - so quite a mixed surprise!

While I was away, I had this list floating through my head of the things I think about when I want to create a language. I always like to find reasons for the structures and words I create - because even though language is arbitrary in theory, in practice most of the things we say interconnect and make sense. Plus it's really hard to make up random strings of sounds off the top of my head. So here are some things to think about:

1. Language evolution. I've heard a number of people, particularly at the Analog forum, talking about the way they've thought through the physiological evolution of their aliens. So why not think about how their language evolved? What was it first used for? Distress calls across long distances? Cooperative activities of some kind? A language used for hunting might come out differently from one used while building tree homes, or one used to find other members of a family in a dark den.

2. Mouth shapes (sounds). The most obvious - but not the only! - application here is to aliens in science fiction. The sounds of the language have to be produced somehow, so if your alien has a lot of teeth, or a very long neck, that might influence the type of sounds in the language. Mouth shapes are also relevant to fantasy human languages, and to real-world languages. The sounds of English just don't require us to move our lips as much as French speakers do, but more than the sounds of Japanese do. Those general mouth patterns contribute to the feel of languages in real life, so why not in our stories?

3. Word structure (morphology). This means breaking down words into their component parts, such as re+cite, walk+ing, dog+s etc. And actually, it's one of the things most likely to be noticeable about your language in a story. Why? Because the language elements that tend to be included in a story are words for things, for languages and peoples, and possibly for activities. I've seen a lot of stories where the sound "i" is stuck on the end to make a plural. But why not use something else? Or, as the Japanese do, just (mostly) forget about plural nouns? The word for a town might have a prefix or suffix meaning "place." There are lots of options, and they can really add a sense of depth to your world. If you look at Tolkien's writings, you'll notice that all the elf names can in fact be broken into parts, and translated into literal meanings. I always thought that was amazing!

4. Sentence structure and above... Unless you're writing full sentences in your created language, you usually don't need to work out how sentence structure works. This is where it's good to turn around and consider how you want to capture the feel of your language in English. If you're not working with real world languages, then choosing an existing language dialect becomes problematic, because very often it will be recognizable. The last thing I want is for someone to ask, "I wonder why Wade's aliens are speaking Cockney?" So I go for alterations in rhythm and feel. English tends to have a natural speaking pattern of alternating high and low stress,

"x X x X x X x X" (iambic)

so you can reverse that, or alter it in funny places. In addition, you can have someone speak repetitively, or in very short sentences, or very long drawn-out ones.

5. Cultural concepts. Your people's belief system, and what kind of things are important or repugnant to them, can have a huge influence on language. This includes political ideology or religion. The Gariniki in "Let the Word Take Me" had an extreme version of this, where their view of the sacredness of their own language shrunk their public usage down to phrases without any sustained rhythm. The flip side of this was that when they spoke fluently, I tried to have them speak in a way that emulated the tone and language of sacred stories in English. I can hardly think of a world in which nobody ever swears (though maybe I should make one, hmm...), and the content of swearing has a lot to do with belief systems.

6. Language learning. Children in a society have to learn their own language, and they don't always need to learn it the same way. The Gariniki children learned to speak in the holy place by listening to stories and discussing them. In our society children generally learn to speak from their mother or primary caretaker. In some societies, children learn more language from their own siblings than from their parents. And generally, television- or computer-generated language is not responsive enough to play a significant role in child language learning.

Foreign language learning also comes to mind, but I need to end this post so I'll leave it for another time. I will say though that it differs in some key ways from child language learning.

I'd love to go into more depth on any of these points, so if you're curious about them just ask. Or it might be fun to do an example language. I'll think about it for upcoming posts.

Do you want to take on dialects?

I was thinking about writing a little on dialects, and then lo and behold, Bill Moonroe over at Asimov's board asked me to write about it! So that was perfect timing.

Dialects can be really fun to work with, and they can also be very challenging to work with, depending on how they're approached (especially the real-world ones). It's awesome to find a story where dialect is done really well - gives the whole thing a lot of flavor. But if it's done badly it can be insulting, as I'm sure many of you know.

Here is one useful principle of dialects: since language is constantly changing as it's being used, the longer a population has lived in a given area, the more that pockets of language use in the area will diverge. Some real life populations that come to mind are China, where the dialects are very often mutually unintelligible, Japan, which has a vast variety of dialects, and England. All of these are populations that have been in place for a very long time. In the US we have many dialects that originate from other parts of the world, but it's interesting to observe how many fewer accents there are on the more recently populated West Coast than in the East.

Intercommunication will not eradicate dialects, but it can slow their divergence. This is a useful thing to keep in mind if you're designing a fantasy world, or a SF planet, where the population has been around for a long time.

Dialects also reflect social status. (Don't get me started on social status! Today I have to stick with dialects...) There once was a study done in New York by Labov (earlier I said England, but my visitor Byron kindly reminded me of the real source) where different department stores had slightly different English usage (in pronunciation of "r"), and this was seen as an indicator of status (how classy the store was). So in your world/universe a particular profession, say, mariners or spacefarers, could easily develop their own dialect because they're relatively tight-knit, often isolated, and have a lot of pride in their own community. These regional dialects can simultaneously conflict, and coexist, with "standard" dialect (i.e. without disappearing), because social value is placed on being able to speak to the public, as it were, but a different kind of social value is placed on being able to mark yourself as one of the insiders for any particular language group.

If you want to work with a real-world dialect, there are different ways to approach it. Using alternative spellings of English can be clunky, both to write and to read, if it's not done just right (and for me, sometimes even if it is done right). Fortunately there's a lot more to dialect than just accent (phonetics and phonology, for the linguistics buffs). You can also work with vocabulary, or usages of words that are particular to the dialect. And you can also work with sentence structure, or with rhythm (prosody). Sometimes the easiest approach is to develop a feel for word-flow with a dialect speaker, and then use the surrounding text to imply the accent rather than trying to render it in spelling. In any case, for using real-world dialects the best bet is not to make any guesses. Find yourself a "native speaker" of the dialect and use them to help you grasp its patterns, and if you can, get them to proofread at the end. This is the best kind of insurance against inadvertently making an error that seems to belittle the dialect.

If you're not working with a real-world dialect, you can have a lot more free fun - though I will note that if the dialect appears to resemble a real-world dialect, or has "developed in the future" from one, then you might want to "hang a light on it," or bring attention to the fact that this is a dialect that has progressed on one hundred years since the cockney dialect, etc. so that people won't make wrong guesses.

My last thought would be that it's important to consider the ease with which the reader can grasp the dialect in question. My critique friends will laugh at me for saying this, after all the odd things I've asked them to try to read, but it's true. I just recently designed a language for a short story, but my main concern was less what the language itself sounded like, and more how that language could be rendered comprehensibly in English while still retaining the feel of an alien language. I think my friends would kill me if I actually ever tried to alter spellings, so I always try to go for using sentence structure and rhythmic patterns to indicate the difference. This is because I'm asking them for a real commitment, i.e. to read a whole story in this style. In "Let the Word Take Me" I did a similar thing with the gecko-girl Allayo, which is to say didn't worry much about what her language sounded like, but gave her a different rhythm of speech to contrast with the voice of David Linden.

If you've got one guy who talks funny, that's one thing, but the more extensive the use of dialect is, then the more reader effort it requires. It's worth putting in some effort as a writer to make sure the audience's comfort is being considered.

Because believe me, I don't want to make anyone suffer.