The OF Blog: March 2005

Friday, March 25, 2005

Scott Gall Interview

Editor note: Scott Gall is a translator who works for a Japanese company, currently working on localizing a MMORPG for American audiences. Unfortunately, due to a non-disclosure agreement, he was unable to tell what company, nor what game. The interview was carried out by our own lord-of-shadow on the Games Board, and it takes a look at the difficulties of translation from one culture to another, using games as the medium.

So, that being said, enjoy

Scott Gall is a localization project manager for a game publisher in Japan.

Max Nichols (lord-of-shadow): My first question was one that came up quite often while I was asking for input from the online gaming community. How much freedom do you have while translating? Are you given strict guidelines by your company and told to follow them carefully, or are you given some leeway in adopting things for different audiences and cultures? And a few were curious about whether you ever get to put in personal touches. Maybe you name an NPC after a friend, or put in an old family joke, or maybe even a reference that only you and some friends would understand?

Scott Gall: Different types of freedoms are awarded and withheld. The first one that comes to mind is the public relations freedom, which I don't have as a localization lead, to mention my project and company here. I'll do my best to avoid boring you to death with generalities though.

The poetic license type of freedom to liven up the dialog, stretch meanings, and ignore certain text elements is different from game project to game project. Some loc projects are set up to trust the translators, some are not. Two in-house translators working for the same company, but on different titles, could have different levels of freedom awarded. Even something like schedule problems, which often result in lighter or more haphazard editing, have sway over what the publisher intended to allow the translators to do.

I used to expect to see a more progressive or enlightened language lead or editor above a translator whose text takes liberties, but I haven't found that to be true. Schedule and budget and managerial oversight often end up playing a part in the amount of translation freedom. In business management terms, much of the game industry is still immature when it comes to localization.

Translation, however, will eventually offer the chance to leave one's personal touch. I suppose the difficulty of inserting a coded personal phrase depends on how big of a mark you try to leave. I don't know anyone who hasn't inserted a little footprint here or there, but these days it isn't as often or as big as players imagine. After a while I think as a translator you realize that there's no reason to sneakily insert esoteric meanings at every turn like a Free Mason. All you have to do is keep working month after month, year after year, and eventually you will have a hundred thousand words that are more or less an expression of you. A few "Look what I got away with!" stories are nice to recount at parties but it's a fleeting joy. I recommend shooting for volume of text as a footprint.

That covers footprints in text translation. Naming, on the other hand, is done separately from translation and is sometimes nothing but a collection of people's footprints. Many of the translators may be asked to participate in naming. They should. Naming is a hugely political task in both the original and overseas versions of every game, and it affects translation freedom indirectly. Before I got into the game industry I worked with marketing text and other entertainment industry writing. I saw very quickly that every human has a burning passion to name something. Especially in text-heavy games like fantasy RPGs, naming and its approval process take weeks. In-game content names are hotly contested within loc (localization) teams, and between dev (development) and loc teams. Usually the loc team is striving to keep the dev team from enforcing use of words with sounds that are too strange or lame in other languages. The dev team might see this as encroaching upon their intellectual property and simply refuse even good naming suggestions because they don't have the ability to differentiate between reasonable and unreasonable ideas for another language. Then you've got the UI command names, which might also need their own abbreviations for certain menus. Then the biggie: the game title. Naming might be the most accessible and visible part of every localization project but it can cripple a project's schedule from the beginning.

The preferred loc process is to create a glossary first. A glossary is what holds all the names decided for items, characters, locations, etc. It's a collection of the company's preferred terminology for that title. The style guide exists separately to help translators know how to notate the mechanics of their writing to achieve a look and feel of uniformity throughout the game. Style guidelines can be assembled quickly but glossary approval is like a first child: It just doesn't want to come out and labor is long and sometimes painful. A delayed glossary inhibits the freedom of translators when beginning work on the text. I mean, imagine you are asked to go through an entire day without using a name. You can use any word to describe a person or thing, but if it has a known proper name, you can't say it. Can't say Knife, can't say Robert, not even Sweater. Try to pass a couple hours like that. You'll know how uncomfortable a translator can feel when starting text without a finalized glossary. I call that lacking translation freedom because the power to envision the whole story in detail has been curtailed, at least for a while.

Max Nichols: Humor, especially humor containing cultural references, must be very hard to translate. When you come across a joke or something that, if translated normally, would simply not be understood by the new audience, what do you do?

Scott Gall: The first thing I ask myself is whether there's anything I can change in the preceding or subsequent text to better frame the material in a more humorous light. Then within the joke, I might shoot for a lesser laugh but at least stay within the flow of conversation or narration. The point of laughter is often context dependent and if you step too far outside the context flow to get a laugh at the same time as the source text, you also might end up introducing some needless confusion, all for what might not have been a very big gain anyway. Not to mention that translators are always on deadline and don't have the time to sit back, have a beer, and brainstorm over how to handle every single instance of humor.

Unless the joke is critical to depicting character or story at that precise moment, you might just sacrifice a difficult joke altogether and opt for plain phrasing, knowing that you can make up for lost ground in the laugh department in other spots where the source text is flat. That's easier to get away with if the acting on screen is not too specific.

If the timing in conversation or narration is truly critical and a laugh must be attained now, then I like to look back several lines for a place where I can introduce background explanation earlier. Try to get the imagination pointed toward where I know it needs to go for their joke, or replacement joke, to have some impact. The price paid is that for a few sentences before the joke, the translation might resemble the source text very little.

Then there are one-liners where there is no buildup at all in the source text. If there's no time to really fudge anything, you're stuck so you rack your brain for a while and do your best. On deadline, you might be able to skip to another file and come back when you have an idea, but humor in games is everywhere, almost in every batch of translation. Try explaining the situation to a friend or coworker and ask for a funny replacement punchline. I know a few hilarious but logical people who can sometimes imagine a replacement for me.

You'll notice a pattern of problems with humor as a translator. In Japanese-to-English work, a translator is repeatedly challenged by having to make a laugh out of content that is often too timid or subtle to amuse most North Americans, and I would guess Europeans too. Japanese translators, on the other hand, are tasked with trying to bring some semblance of meaning to English jokes, which often rely on knowledge or sensibility that is too spontaneous or ironic or violent or simply unimaginable to be even faintly realistic or relevant for a Japanese person in the situation at hand.

So you've got English game players who are hard to surprise and amuse with text that was interesting but not funny, and the translators are working hard to sharpen the jokes. And you've got Japanese game players who are easy to confuse or bore with text that is funny in extremity but not necessarily clever, and the translators work to boil it down to its core and just translate that, so that it is at least understandable. It's safer to bank on Japanese appreciating cleverness more than volume of humor, so if the joke has some layers to it that highlight the human condition, it will be easier for the translator to get some mileage out of it.

Here's a film example I would teach in translation class if I were a teacher. I saw Wayne's World in Japan when it was first released in theaters. The Grey Poupon TV commercial campaign of rich people interrupting each other to borrow mustard was known in the States, but it was totally unknown to Japanese people. Now, rather than fudge the lines leading up to "Might you have any Grey Poupon?" so that some attempt could be made at understanding the high society angle, the Japanese subtitles simply replaced "Grey Poupon" with "condom." I've asked over a dozen Japanese people through the years who saw the film if they thought that was funny and not one has answered yes. However, none of them were perplexed with the sudden abstractness of that sentence until I asked them to reexamine the scene. I used to think of that subtitle as a failure but years later I've decided it's genius. The translator used a word that does not make sense or generate laughs, and almost any single word there would fail to do so. But the word "condom" is apparently just within the scope of conceivable wackiness for the Japanese that they can subconsciously brush it off as "That silly guy, he's asking for that? What will he ask for next!" The moral is if you can't get a laugh, at least you can stay in character of the person speaking, and then you might be able to keep from drawing attention to the fact that it wasn't actually funny.

Max Nichols: Cultural differences are certainly not limited to humorous situations. What do you do about symbols such as a swastika, which I understand is considered a lucky symbol in some Asian countries, but would be instantly associated with Nazi Germany in America?

Scott Gall: What I would do is ask the dev team what they were smoking when they designed it in the first place!

There should be few surprises these days about what upsets people so much that sales are impeded and bad publicity results. At least, in the markets usually shipped to. Doing your homework during development can fish out a lot of problems. What you can't avoid is current events. What if an original symbol in a game halfway through development turned out to be similar to, I dunno, the signature of a murderer who gets arrested one month before the game's release? You're stuck and you say a prayer to the PR god of your choice and hope everything will be better in the morning.

A dev team should take some time in the beginning to decide what it will create for each market. If they determine that 90% - but only 90% - of their target audience can accept a middle finger or swastika or nudity, and therefore only one version of the game will be sold everywhere, then they really have no right to complain if their sales or rating in 10% of markets is bad. For any reason.

If a game was never intended for loc but somehow was picked up for international sales later, the publisher should review the game carefully before deciding where to sell it. In the case where something must be deleted or added to the game after it has been completed, I suppose a contingent of people from the dev team, quality assurance, and localization would come together and crank out whatever changes are needed. When a game asset unexpectedly upsets an overseas culture or ratings board, everyone in the industry learns from it and life goes on. Game content adjustments may be expensive but in the long run they pay off. If the swastika-looking symbol absolutely must be kept by the dev team for the notion of artistic integrity, then expect a higher age rating for the game or lots of little print from the legal department and even a ban here or there. That's how it works in just about every business I can think of that depends on exports.

Max Nichols: Censoring has always been a big deal among many people I know. Do you find yourself having to constantly alter things to make them more suitable for their new target audience? In my experience, things that might be considered taboo or unspeakable here in America (incest in the Fire Emblem games, for example) are less so in other cultures¡­ and vice versa.

Scott Gall: I've never worked on a project that pushed the envelope on sensibility. What I notice is that you constantly have to change the NPC dialog to make it interesting.

If I felt I ever wanted to recommend a game element, I would go to the dev team and producer and state my case.

North American culture has low tolerance for child exploitation, for example, but has high tolerance for violence and themes of lawlessness as self-defense. I think this goes back to your question on culturally sensitive symbols. Most of the time, a developer sees the hot water ahead before stepping into it. You can either try to change a society and its laws to accept your art, or you can change your game. I can't say which is better, but I think you can see which way is more profitable in the short term.

Max Nichols: I have known fans that are angry at the slightest change to the original meaning, people who would prefer to have the Japanese originals translated as literally as possible - cultural references, taboo subjects, and everything else. And that makes sense, to some extent - the original intent and meanings should never be lost in a translation. Yet it would not be in the best interests of the companies involved to make a game that is inaccessible to the general American public, due to a translation that is too literal. Obviously, a balance must be reached here. How do you go about achieving this balance? And what is your opinion on the subject of original intent and meaning versus localized intent and meaning?

Scott Gall: First a translation must be understandable and enjoyable or the publisher won't stay in business. Enjoyability is achieved with the highest success rate when the translator knows the culture of the people he is translating for and adjusts to their sensibilities while maintaining as much of the flavor as is permitted within the localized sensibility. This issue has been blown out of proportion past what is really happening in real-world translation. Have you ever seen a translation of second-generation Korean-Japanese gang members calling each other "Nigga"? I haven't. But I should be seeing that and other similar text, according to this argument that dialogue is being overly localized. There are words that Japanese street punks call each other that are slightly derogatory but which also show the closeness of brotherhood and which cannot be used by anyone outside that circle without inciting a fight. People who don't know any foreign languages well are more likely to mistake current translations as a sell-out, but in fact that's nowhere near the case.

Japan megafans who want the Japanese sound effects or extremely literal translation should go study Japanese and then buy the original game. The localized English version exists for people who have no intention of studying Japanese. Translating some of the English but keeping Japanese-isms sprinkled within is a whole different version and would require a separate localization project. I admire the energy of those fans who are willing to accept harder gameplay in exchange for cultural adventure, but look, just because a comic or game or animation is Japanese doesn't mean it's intrinsically more hardcore and accurate and cool than the same thing produced in Seattle.

Max Nichols: Many niche titles are already bought almost exclusively by fans of Japanese culture and language, or at least by those who have some appreciation of it. Seeing that this is the case, why don't we see more subtitled games, or at least games with the option to keep original Japanese voice-acting? I myself have set some games to Japanese voice acting with subtitles, rather than subject myself to the torture of poorly done English voices.

Scott Gall: I would go back to the reason of money again. If you, Max, establish a Paypal account and collect $50 from tens of thousands of people and then contact a game exec, I think you'd get a game with English UI and text, with the original Japanese voice acting. Unless you go that far, you're asking a company to take a gamble that the money they make from Japanophiles will cover the customers lost who only wanted English. I agree that's a short-sighted way of looking at it, because some of the titles people seem to want in Japanese are the same games that never sold well to begin with, so why not take a risk and fill that obvious niche market? And depending on how heavy the data is, maybe you could just toggle it on or off the way you do on a movie DVD. I don't have a definitive answer for you, but I'd remind you to take a look at the human mind. No one wants to be the first to fail at something new anyway, and Japanese companies are particularly terrified of this.

Another thing, a game company must adopt different licensing and payment for the original voice actors if that plan comes to pass. If an expensive voice actor was used in the country of origin, that person might actually ask for so much money for international release that it becomes much more attractive to use those "poorly done English voices."

Max Nichols: Connotations - I imagine they can be very, very annoying. Heck, even within a single language, these can cause communication issues. How carefully do you translate connotations? Do you just stick to the straight definition, or do you make an effort to incorporate other meanings and implications?

Scott Gall: I think you have to translate text with important connotations just as much as they need to be translated. No more, no less, best you can. With games and subtitling you're always working with text length restrictions, so you can't afford to get fancy at every turn. Reasons why you might find subtle meanings untranslated are space, time, translator ability, and editor ability. Or you're reading too much into the original text! Stop imagining connotations! No, seriously though, you can't judge the quality of a translation based on single lines here and there. In the end, most games are created to be lighter and faster than an essay or poem. If a player or a translator is obsessed with text containing layers of subtlety line after line, then I'd recommend literature over electronic entertainment. Not every game has a lot of subtlety to the text, and not every translation is able to capture it, for one reason or another.

Max Nichols: What's your approach to translating honorifics, such as chan, san, and sama? I've talked with non-professional translators before, and these usually present problems. It's hard to effectively communicate the same levels of respect, restraint, relationship, and everything else in English, without getting to wordy.

Scott Gall: It's hard to do without getting wordy, but not impossible by any means. Failure to do it repeatedly in proper brevity doesn't mean Japanese is a hard language. It means the translator was a mediocre English writer before he or she ever got into translation.

Not everyone realizes that, in daily translation business, translation is often done by people who are not bilingual. They might be hotshots but aren't necessarily bilingual. And that's ok. The "weakness" of a non-bilingual translator is covered by having him or her translate *to* his or her native language. Johnny translates J to E, Taro translates E to J. So nowadays you've got native English speakers, for the most part, responsible for translation, or at least for the editing. If the text is too long or clunky or just plain boring, blame Johnny for not honing his English writing craft. Taro is probably innocent.

Japanese honorifics are expressible in the five registers available to the average native English speaker. A linguist told me there were many more than five registers, but whatever, five is plenty for those without a graduate degree in linguistics. You, as a native English speaker, have the word choice and voice tone that you use for immediate family and best friends, and another for a professor or police officer, and yet another when you give a speech. Those are three easily recognizable registers. By mixing registers, one can achieve a mix of high and mid-level politeness found in many languages, including daily Japanese.

In Japanese, -chan and -kun and other name add-ons can be chosen to show affection and social proximity. Just because it's not -san doesn't mean the person is being spit on. I know at least two cases where a female is called -kun, though it's fairly uncommon, and then there's the sarcasm, or sometimes just the innocent teasing, of attaching -sama. So you can't bet on those words to be the best road signs for how much respect is present, and carrying them over into English, they're just dead weight. Unless you're trying to stress a Japanese presence in the text for some odd reason, just drop those words and call characters by their names in English. Everything will be ok. Trust me.

If someone wants to worry about names, I advise them to spend time on a practical issue, such as whether [Japanese text cut due to formatting problems] should be translated as Taro or Taroh or Tarou.

Max Nichols: How would you go about translating something with a standardized form, like haiku or certain types of poetry? Would you prefer to try to maintain the original form, or the original meaning?

Gall Scott: I'd choose to aim for the original meaning. I wouldn't take the job if someone told me to match the syllables. Or I'd take it but quadruple my price. Or I'd attach the condition that kanji could not be used, which would hopefully kill the deal right there. If you exclude abnormally long scientific words, the average word length in English is about six letters. That represents fewer syllables than the average Japanese word. Poetry, I'll pass. Unless it's my girlfriend's poetry.

With music lyrics translation you're up against a similar challenge. However, the canvas tends to be bigger and this can help you get by as a translator. With a bigger canvas, you're more likely to be able to get the meaning across, and even have some fun doing it.

Last year a Japanese guy at the pub I went to a lot at the time in Tokyo asked me to translate his song. He's an amateur musician. I showed my translation to an American friend of mine and he said "Oh, I didn't know you write songs." So I thought it was gonna be good enough. Then I handed the translation over to the Japanese guy and he said, "This is not what I imagined. You made it stick to a time structure. Don't worry about that. Just be cool." Yes, sir!

Max Nichols: In English, forms of onomatopoeia, such as "bang" or "pow," or just attempting to emulate a real sound, are usually regarded as childish or unsophisticated. That is not the case in Japanese, where it is considered a formal part of the language, and has its own nuances and the aforementioned connotations. How do you go about translating these?

Scott Gall: When X-men the movie was released, I taught my Japanese friend the famous SNIKT for the sound of Wolverine's claws. At the time she wasn't an accomplished English speaker, but she managed to remember that word and when we were at a pet store months later she was holding a cat's paw and repeating, "Snikt! Snikt!" I'm like, "No, um, that's not quite right, but nice try!" The point being, to a Japanese person, the sound of one set of cat claws could be the same as the sound for another set of cat claws. The biggest difference between Japanese giongo and English onomatopoeia is that where you and I will argue whether it should be SWAPT or SWAAP or SWIFF, Japanese people more or less agree which spelling of which word best represents a given sound. Japanese comedians get a lot of mileage out of throwing a slightly modified giongo word into a story. Everyone knows that the comedian knows that it's not the usual choice for that sound and so it becomes funny.

To me, translating giongo gets tricky only when it's part of the dialogue. The same character might use it several times, or several characters might use it once each to describe the same item. If it's in dialogue, I do believe it should be translated. Try to pick something that a reader wouldn't get annoyed by after three times. I think you have to accept as a writer that others will have their own way of envisioning a sound in English. All you can do is try to be neutral in spelling it. If you do decide not to translate it, that's fine too, but watch out if it's within dialogue. It probably should be translated. If it's in narration or straight sound effects, the service you do a native English reader by translating every instance is debatable.

Max Nichols: Have you ever found a passage or section of writing that is almost completely untranslatable? If so, what would you do to tackle this problem?

Scott Gall: I would send it back and ask the client, or the requestor, to reassess what kind of translation would be best. That's the voice of experience speaking. I was once lucky enough to get a freelance translation assignment for Japanese children's language public TV programming. They wanted one episode to be subtitled in English so the show could compete for an educational programming award in Europe. I struggled with the translation severely but after two days I'd finished no more than five lines. Feeling totally defeated, I went to see a Japanese friend who was an accredited teacher of Japanese as a foreign language. I said to her, "Take me out back and shoot me, I can't do this." The show featured classical Japanese, giongo, syllable-based songs, children's whiney pronunciation, and it was all to be subtitled. A subtitling nightmare.

My mistake was in trying to translate in a way that explained why certain key words and lyrics were being introduced. Kids who watch that show don't understand the Why of the phrases in the skits, they just learn the What. But I was determined to get the meaning across for the sake of the programming judges and I failed to see that the meaning was not the point. The point was to have fun with words, and it wasn't fair to the content to be boiled down to just an explanation. The client tried to edit my translation and I think then they finally realized what I'd been up against. In the end they decided to let a lot of things go untranslated with just a short explanatory title at the beginning of each skit. Unfortunately, the visuals made little or no sense that way. You needed a slight understanding of the words to appreciate the interplay with the visuals. Looking back, I think the best way would have been to provide playful subtitling and then provide a separate explanatory script translation. That would have been unfair to other children's programming entries in the competition but I don't know how else to do it. I imagined the German competition judge watching this and wondering whether the entire show was thought up by hippies on acid who'd come through a wormhole from Woodstock. Visually, the show was all over the map.

The moral is, the monolingual client isn't always right, and not every kind of text is "ready" for translation. When you are told to "Translate this!" it can pay to sit down and think about how much to translate, and how.

Max Nichols: How much story is lost in translation, would you say? Obviously this would depend on how literally you're translating it, but it must still be something you need to worry about. I certainly know some fans that worry about it, and even go so far as playing Japanese versions that they don't understand to avoid translations they consider poor (I don't get their logic at all, but hey¡­.

Scott Gall: Plot? Very little is lost. It's all there usually. Emotion and reason and motivation? It can vary. What a Japanese person considers ample motivation for an action is not always going to make sense to someone from another culture, and vice-versa. In those cases you have to find a way to make the emotions more believable. You have to jettison the source text. You have to cut the anchor loose or else the ship will just sit there and get hammered by the storm of confusion. Let the boat of words float away and go where it must to find safety. If that's not acceptable to a player, then again I say, with all due respect, that you can buy three language textbooks for the price of one new game.

Max Nichols: Have you ever found a translation that you would consider superior to the original? If so, what was it? Or is that even possible, in your mind? After all, for a translation to be better than an original, it would have to deviate from it a bit, and that may not be acceptable.

Scott Gall: Lots of things can be said more clearly in translation than in the original language. But that doesn't necessarily make for a superior text. I wouldn't want Myst to have a translation that leaves no margin for doubt, wonder or imagination.

In cases where I think a translation feels crisper and more inspiring, I expect to see a very finely crafted source text. Translators recognize inspired writing and can motivate themselves to do a better job when they're working from something that started on a higher level. When I spot those rare cases, naturally I applaud the translator, but I also have come to applaud the loc project manager for being able to get the right translators enough time to create such careful work.

Max Nichols: Living in Japan and working for a Japanese company is something very few of us will ever experience, in all likelihood. Naturally, we're very curious. Do you enjoy it?

Scott Gall: My emotional outlook on working for a Japanese employer is similar to what you hear parents say after many years of raising children. Parents say, "I wouldn't trade the experience for the world. But if I knew it was going to be this hard then I wouldn't have had a second one." That sounds about right.

Max Nichols: How long did it take you to learn the language? Was it a difficult language to learn?

Scott Gall: I am still learning but the answer you are looking for is "four years." I studied for two academic years in the States, one semester in Japan, then busted ass in self-study for a while after moving to Japan. In some areas I have an affinity for the language that others don't. I believe that comes from genuinely liking the sound and, more or less, the structure, and having respect for the people I meet. Reading and writing are a hassle. Speaking is a reasonable challenge. The U.S. military still ranks Japanese as one of the most difficult to train for but I disagree. How can you compare it to all the verb tenses in Spanish, the consonant-heavy pronunciation of Eastern European languages, or the pronunciation of Dutch for that matter, or the gender issues in Italian, or all the different rules for referring to people's names in Russian, or the four-thousand more kanji in play at any given time in any of the Chinese languages. There are languages easier to fail at than Japanese.

Where students of Japanese really suffer is the learning process. The variety of educational materials is weak and many teachers seem tragically incapable of coming up with effective approaches to explain the age-old sticky points of the grammar and usage.

Max Nichols: Do you see yourself as an outsider at your job and in the area you work in, or is it an easy place to adapt to, in a business setting?

Scott Gall: Localization teams are an outsider in software development, unfortunately. You're either making the program or selling it, and everyone else in the process seems to be thought of, at times, as expendable. Within localization I'm not even a native Japanese person. I'm the minority of the minority in my company, and it can feel that way sometimes. But you have to be careful in assessing whether you feel isolated because you're a foreigner or because you're in localization. I used to work at a loc vendor, which means that all we do is localize software 24 hours a day. Literally. There were times when our clients simply used us a punching bag and tried to blame us for all the bugs in their original programs. I felt isolated, powerless and wondered if I would live through it. If you let your respect for yourself slip away or if you let your health slip, things can get dark fast, and then it doesn't matter who's picking on you or why. As for working under most Japanese managers, it's not an environment that a lot of people will thrive in. There's little or no encouragement to tackle problems from new angles. Get used to being your own cheerleader and guru and critic. Your foreign coworkers are critical for staying sane.

Max Nichols: Finally, a question asked by someone who said they're actually interested in a career of this sort: How do you get a job like this? What sort of things did you need to do and learn, what do employers look for

Scott Gall: Foreign language ability, strong writing and speaking skills in one's native language, a positive attitude, an international outlook, extended time living abroad, and a history of consistently accepting and completing cross-cultural challenges. All of that together is 50% and the rest is timing and luck.
Add to Technorati Favorites