Curation style guide
Last updated: February 1, 2021
The goal of Moments is to showcase the very best of what’s happening on Twitter — the amazing conversations and memes, the breaking news, and the diverse voices and perspectives.
Moments are intended to feature compelling, original, and diverse content. When selecting Tweets for a Moment, we consider the content of the Tweet, any media, and the avatar and username of the poster.
Impartiality: We will use data-driven decision making when choosing Tweets around controversial topics, and highlight the Tweets already receiving the most engagement on Twitter. On topics which reflect public debate, we will select Tweets that represent many sides of the conversation, where feasible. Twitter curators should not advance their own viewpoints, but rather reflect the discussion as it appears on our platform.
Accuracy: When dealing with news or newsworthy conversations, we want to highlight quality Tweets that represent accurate information.
Corrections: If we become aware that we have highlighted content that turns out to be inaccurate, we will update the Moment with a visible correction, and issue an updated Tweet. In rare cases, we may delete the Moment and post a retraction.
Standards: Profanity, violence, nudity, and other types of potentially sensitive content should be avoided, except where it is necessary to tell a newsworthy story. We will not include content that promotes or depicts illegal conduct. We will provide individuals with clear notice that a Moment may contain potentially sensitive content.
This style guide is designed to complement the existing curation policies that help our curation team create engaging, factual, and impartial content that can resonate globally. It is meant to be a guide only: We may change the guide or deviate from it, from time to time.
The golden rule
We speak in the language of Twitter.
For example, when we’re adding emoji to the end of our sentences or questioning whether per cent is one word or two (answer: it depends on your country, but people engaging with Moments are actually way more likely to use the % symbol), we strive to stay true to the conversation on the platform.
When shortening country and organization names, we do not use periods. This reflects the common usage on Twitter. So it’s US and UK, not U.S. and U.K.
Active shooter reports
The Department of Homeland Security defines an active shooter as “an individual actively engaged in killing or attempting to kill people in a confined and populated area; in most cases, active shooters use firearms(s) and there is no pattern or method to their selection of victims.” It notes that these incidents are unpredictable, evolve quickly and are normally over in 10 to 15 minutes (The DHS pamphlet is also available in Spanish).
This term is often used incorrectly in breaking news situations, most often to describe reports of shots fired. Misusing the term active shooter, Factal says, “needlessly legitimizes and escalates a situation,” including increasing calls to police and fuelling panic among those nearby and concerned who are following along.
We should avoid using the term “active shooter” unless police confirm an active incident. Instead, such reports should be described as “possible shootings” or “police responding to reports of a gunman.”++
Ad hominem attacks criticize a person, rather than their idea or position in an argument.
It’s the difference between choosing a commentary Tweet that says, “The prime minister is an unhelpful person,” and “The prime minister’s statement was unhelpful.”
We avoid including Tweets attacking an individual or their character unless they are integral to the conversation as it is occurring on Twitter and can be placed appropriately in context.
The term allegedly is used to convey a claim that something has happened but it has not been proven. This should be used when discussing criminal charges or allegations of wrongdoing that have not been established in court. Allegedly should not be used as a substitute for verification in other reporting. In those cases, we should attribute the source of the claim.
Annotations provide context, background, biographical notes, translations, or subheadings to Moments. They can also be used to explain the choice of Tweets, for example, by explaining someone’s expertise.
Anything written in an annotation should be easily verifiable or attributable. For example, by including the original-language Tweet alongside a translation in an annotation.
We do not want annotations to repurpose or repackage other people’s content; we always prefer to use Tweets. When using information that is not available in Tweets, paraphrase or quote as necessary, but try to keep it short.
Annotations used as headings should follow the same style as titles, while annotation body text should be written in sentences.
Everything stated in a Moment title, description, and Tweet should also appear within the Moment.
When a fact is supported by three or fewer sources, we attribute to the source, by name, in the title, or description. Once it has been confirmed by more than three sources, the “reports” and attribution references may be removed.
Instead of using “reportedly,” “according to sources” or “it is believed to be” in our Moments, it is preferable to attribute the information to the source.
We should not curate just to gross people out. Accordingly, we shouldn’t create Moments just focused on bodily functions, especially when it isn’t being widely discussed.
If we do make a Moment like this, please consider the context in choosing our words and tone, and do not feature the gross content in the cover image.
We avoid including content from brand accounts in Moments unless we think it’s integral to the conversation.
Curation helps give people the context and narrative they need to understand the things that matter on Twitter. So we want to make sure that our titles and descriptions give people the context they need to decide if they want more information.
We still want our titles to be conversational, informative and creative, without veering into clickbait. Clickbait intentionally withholds information to mislead people, or poses questions that force them to click.
Names of businesses, political parties, and groups are singular and we treat them as such, e.g. Company X released its earnings Wednesday.
Bands and teams are treated differently between markets, so our style is to write around the singular/plural discrepancies. For example, instead of writing "The band released it's/their new album" choose "The new album is here." When it’s difficult to write around it, we default to the way people on Twitter typically speak (e.g. The band (with a plural name) are dropping a Canadian tour).
When covering conversations about crime, we should refrain from putting photos of the suspect or accused on the cover, particularly when the only photos available are candids, rather than police sketches or court appearances.
We should avoid using the names and images of alleged attackers prominently in our titles, descriptions, and covers. Our Moments should prioritize those affected by the event.
Curating about misinformation
Our mission is to provide quality content and context to help people judge the accuracy of the information that they’re seeing.
Misinformation comes in many forms, including, according to UNESCO:
- Satire and parody: no intent to cause harm but potential to fool
- False connection: titles, images and/or captions don't support the content
- Misleading content: misleading use of information to frame an issue or individual
- False context: when genuine content is re-shared with wrong information about the situation
- Imposter content: impersonating genuine sources
- Manipulated content: altering true content in a way that deceives
- Fabricated content: falsehoods
Before curating conversations about misinformation, we need to consider how visible it is on the platform and the communities that are sharing it. We want to avoid moving a niche topic into the mainstream conversation. Therefore, we should prioritize providing accurate information around something that is in a country’s top trends rather than seeking out conversations with limited reach.
When curating on these topics, we must provide multiple sources for verification and offer very clear attribution.
Do not repeat the false claim in the title and description. As First Draft explains, “It doesn’t matter if the 850-word article provides all the context and explanation to debunk or explain why a narrative or claim is false, if the 80 character version of that context is misleading, it’s all for nothing.”
Be cautious of repeating unfamiliar phrases used in the misinformation, as these keywords may be strategically created to drive people to discover more misinformation. You can learn more about that phenomenon, with the example of “crisis actors,” from the Data & Society Research Institute.
We should not use the terms yesterday, today, or tomorrow in titles or descriptions because it may be Thursday in the United States, but it’s already Friday in Australia. Or a late-night game on Tuesday night could still be featured on Wednesday morning. Use the day of the week instead.
For live events, include relevant time zones, for example: Tune in live from 2pm AEST, 12am ET, and 4am GMT. If there isn’t enough space, default to the time zone of the country of origin.
When writing a date, please style it as the month (not abbreviated) and the day. Example: May 12, September 12.
Please style decades as the numeral followed by an “s,” such as the 60s, 70s, 80s or 90s.
In breaking and developing news situations, we should be clear about the degree to which information has been confirmed, along with the sources of that information (see Attribution). We do not need to include “more to come,” or “stay here for updates” in our descriptions, but we should update the title and description as the situation develops.
Disabilities and illnesses
We avoid language that appears to define people by their condition or illness. For example, use “people with disabilities,” rather than “disabled people,” or “the disabled.”
Additionally, mention disabilities, or a disease, only if they are relevant to the conversation. We should avoid descriptions that make a disease or disability seem like an affliction, such as “suffers from Parkinson’s disease,” or “battling cancer,” opting instead for “has Parkinson’s disease,” or “has cancer.” We may use mental illness as a general term, but we prefer to use specific conditions when possible.
Please refer to people the way they self-identify. For example, some people with hearing impairments identify as Deaf, which encompasses identity and culture, instead of deaf, which means someone with limited hearing. (Read more about that distinction here).
For more, please visit the National Center on Disability and Journalism’s style guide, which has an alphabetized list of terms and considerations.
Our role is to help people see as much context as possible around information gaining popularity on Twitter. If popular, high-profile conversations include information that is factually disputed by a wide number of sources, we should clearly cite those sources in the context that we provide. To every extent possible, we should avoid repeating disputed information in the title.
We use emoji as emphasis or to replace emotion in our titles and descriptions, not to replace words. Keep emoji at the end of titles and sentences, as a lot of people on Twitter do. We should not use emoji for serious or potentially sensitive topics, or for breaking news.
We consider esports (lowercase e, except when styled differently in an event’s proper name) a catchall term for competitive multiplayer video gaming. Competition normally occurs in leagues, or tournaments, where commentators are typically called casters. The two main variants of esports are first-person shooters (note the hyphen), also known as FPS, and MOBAs, which stands for Multiplayer Online Battle Arena. We should define both acronyms when referenced in our Moments. Simulations of conventional sports, including FIFA and Madden, are also popular variants, while the largest tournament is Evo, focusing on a range of fighting games.
We give video game titles the same treatment as movie and book titles, with title case but no italics or quotation marks.
When possible, our preference is to use the non-gendered term to describe a person’s job or role. For example, police officer instead of policeman.
For celebrities, when a preference isn’t stated, we default to “actor” or “singer” for both genders. To quote Whoopi Goldberg: “An actress can only play a woman. I’m an actor – I can play anything.”
Featuring and removing people from curation
As part of Twitter’s mission to serve the public conversation, we strive to respect the customer’s voice and avoid situations where someone may unexpectedly become the target of large volumes of unwanted interactions.
We should exercise caution when curating personal narrative content around issues such as identities and discrimination, including ethnicity, sexual orientation, gender identity, religious affiliation, disability or disease.
Be cautious when featuring debates including devoted fandoms, as this dedication can result in bullying behaviors.
We also need to consider the difference between public and private individuals. Public individuals are those who voluntarily choose to live in the public space, through activities such as running for public office, performing in the arts or competing in pro sports. This does not include people whose photos or videos go viral.
Users can delete their Tweets, set their accounts to private or block the relevant @TwitterMoments handle to remove their Tweets from Moments.
We should not include hashtags in Moment titles, as they’re not clickable for readers.
Words within hashtags may be capped, to improve readability within a description or Tweet (e.g. #StarWarsDay vs #starwarsday).
Health and science studies
In our view, a good health or science Moment:
- Presents the entire story.
- Provides the necessary context to understand the research.
- Doesn’t overstate the study results.
We start by looking at the study itself:
- Where is it published?
- Is it peer-reviewed? (That means other experts in the field have evaluated the methods and results).
- What was the sample size?
- Read the study abstract and analyze what the researchers are saying. Are the news headlines exaggerating or overstating the claims?
We look for experts in the field who are fact-checking or responding to the study. We hope that by presenting alternative viewpoints, from experts, we help our readers draw their own conclusions.
We aim to use people-first language in our curation to avoid labeling or inviting judgment on people based on their situation. In lieu of calling people homeless, we should use “people experiencing homelessness.”
When describing people and groups, we defer to the way they self-identify. We avoid generalizing or guessing people’s identities, including around gender, ethnicity, political affiliations, or faith-based beliefs.
We should only mention an individual’s race or ethnicity when a conversation is definitively about that topic, for example in historic, groundbreaking events or certain social justice coverage.
When it comes to an individual’s or group’s positions and beliefs, we should show, rather than tell. Instead of applying a political or ideological label, we should try to describe the specific position or behaviour of the person or organization in order to provide the context that is relevant to the conversation.
We also keep in mind that sharing an idea that appeals to a certain group does not necessarily mean an individual is part of that group or holds more of that group’s views.
Following the lead of the Associated Press and others, we should only use the term “illegal” to describe actions related to immigration, not to describe people. We avoid using the term “illegal immigrant” unless it is part of a direct quote, choosing instead to say someone entered or is living in the country illegally. We also avoid the term “undocumented.”
We should attribute any details about immigration status, for example as disclosed by a lawyer or the relevant government agency. When possible, we should also share the person’s individual circumstances, such as whether they overstayed a visa or crossed a border illegally.
The term “migrant” is generally considered a neutral term to describe a person voluntarily moving from one region or country to another without implying anything about their legal status.
A refugee usually refers to a person who is forced to leave home to escape persecution, war or natural disaster.
People who were brought to the US as children should not be described as having immigrated illegally. When such individuals are listed as having “temporary resident status,” they are typically called “DREAMers,” referring to the Development, Relief, and Education for Alien Minors Act.
We may curate Tweets that do not follow our preferred terms, but we should exercise caution and keep these guidelines in mind.
“Impartiality is not the same as objectivity or balance or neutrality, although it contains elements of all three. Nor is it the same as simply being fair – although it is unlikely you will be impartial without being fair-minded. At its simplest, it means not taking sides. Impartiality is about providing a breadth of view.” (BBC).
We try to help our readers access the information they need to make an informed choice about their own opinion. We believe that impartial curation means not taking sides and taking care to provide the relevant arguments, facts, and views on any issue without judgment or bias, and we aim to do so at all times.
To take a page from the CBC’s Journalistic Standards and Practices, “on issues of controversy, we ensure that divergent views are reflected respectfully, taking into account their relevance to the debate and how widely held these views are.”
Indigenous populations in Canada, the US, and Australia
Indigenous is the preferred term in Canada and Australia for those whose roots precede European colonialism. In the US, use Native American, or in the case of Inuit in Alaska, Alaskan Native. We avoid terms such as American Indian, Amerindian, and Eskimo. Use the term Native Hawaiians for the descendants of the original people of Hawaii, who make up about 6% of the island’s population. Use Hawaii resident for anyone living in the state, per AP.
If an individual or a group identifies as a member of a specific nation, we may refer to them using that term.
There are three distinct groups of Indigenous peoples in Canada: First Nations, Métis, and Inuit. First Nations is generally used to describe membership. We avoid describing people as Métis, unless they explicitly identify as such. The term Inuit refers to any Indigenous peoples inhabiting the Arctic regions of Greenland, Canada, and Alaska. The singular version of Inuit is Inuk. For example, “An Inuk man lives in an Inuit community in northern Canada.”
In Australia, Aboriginal refers to clans of first peoples that originate from mainland Australia (including Tasmania). Torres Strait Islander refers to the clans that originate from the Torres Strait Islands. We should refer to both Aboriginal and Torres Strait Islanders as Indigenous. We avoid “Aborigine,” “Aboriginal” or the abbreviation “TSI.”
In many Aboriginal and Torres Strait Islander communities, there are cultural protocols around using the name, voice, and image of a deceased person. The family or community will release a statement indicating whether the deceased can be identified. If the deceased cannot be identified, we should explain why in the description or an annotation.
This group goes by many names, including IS, ISIS, ISIL, and Daesh. We follow the BBC’s convention of referring to it as “the Islamic State group” or “self-styled Islamic State” in the first instance and IS for subsequent mentions and titles.
LGBTQ, which stands for lesbian, gay, bisexual, transgender, queer or questioning, is our preferred term. We avoid using less common abbreviations, such as GLTB, but if we must, we should define it. Queer can be considered a pejorative term, so please use it cautiously.
We should only reference an individual’s sexual orientation when it is essential to understanding the conversation. Nothing in a Moment should question or devalue a person’s gender expression, identity, or sex.
When referring to an individual, use their stated identifiers. If we don’t know their correct pronoun, please use the singular they or them. We do not use quotation marks around names, or pronouns, used for transgender or non-binary people.
Gay is the preferred term to describe men and women attracted to the same sex, though lesbian is more commonly used for women. We do not use gay as a noun.
As the term homosexual is considered pejorative in some markets, we try not to use it. By the same token, sexual orientation should be used in place of sexual preference, and LGBTQ people should be described as “out” rather than “openly gay.”
We avoid defamatory terms about the LGBTQ community in any display copy or selected Tweets unless they are part of a newsworthy direct quote.
The transgender community
Transgender persons have a gender identity that differs from the sex they were assigned at birth. It’s an umbrella term that can include anyone who feels they don’t fit into their assigned sex or gender. Drag queens/kings are not necessarily transgender, though some trans people are also drag performers.
We should not use “trans” or “transgender” as a noun. “Trans” is most often used as an abbreviation for transgender but it can also mean transsexual, transvestite or other terms with potentially negative connotations, so it should not be used without a signifier.
We should respect people’s gender identities. Thus, we refer to transgender women as women, transgender men as men, people who identify as non-binary (sometimes “NB” or “enbie”) as neither a man nor a woman. We avoid using “trans man” or “trans woman” (never “transman”/”transwoman”) unless someone self-identifies as such.
We do not use quotation marks around names or pronouns. And we should avoid referring to a person by their “dead” (pre-transition) name.
People who aren’t transgender may be termed as “cis” or “cisgendered.” Some LGBTQ people object to “cis” as an oversimplification, while some non-transgender people dislike being given a label. We should avoid “cis” in titles and descriptions, and be mindful of using it carefully in curated Tweets.
Phrases like “tranny,” “she-male,” “he-she,” “gender-bender”, or insults like “freak”, “unnatural” or “fake” are considered offensive, although advocates may use them to prove a point. We avoid these terms in titles and descriptions (use “offensive term” or “transphobic term”), but these terms may be used in curated Tweets if it is essential to the conversation and the context of their use is clear. We should not include Tweets that use these terms as direct insults.
We avoid using terms like “transexual”, “pre/post-op”, “born in the wrong body” and “sex-change.” We generally use “crossdresser”/ “cross-dresses” in place of “transvestite” and “intersex” rather than “hermaphrodite.” We avoid terms that imply someone is “changing their sex.”
Transgender people can have any sexual orientation. We should not assume a link between their gender and their sexuality.
Memes are images, video or text that rapidly take on a life of their own, repeating and trending on the platform. Some become icons (Salt Bae, tea lizard), while others fade quickly into obscurity.
Not everything that is being shared or remixed is a meme. We should be seeing a critical mass of repeated use, with a common thread between the posts, before we label something a meme.
When flagging monetary values, we should lead with the currency of the originating country, for example, British Pounds for Moments occurring in the United Kingdom. When targeted to other countries, please also convert other currencies into USD, which should be indicated as $X (US).
Names and titles
Books, films, shows, and albums
We capitalize the titles of books, films, albums, songs, and TV shows and do not put them in quotation marks. We reserve double quotes for lyrics, dialogue, and actual quotes.
We capitalize references to TV seasons, for example, “There were many dragons in the Season 7 finale.” We write the number of the season as a numeral.
When including a Twitter handle in a description, we should include the person’s real name if it does not match their username.
When describing the people we feature, we avoid calling them "users." Instead, share their name or the role or detail from their bio that led to their Tweets being curated. If there isn't any biographical information available, we should use their @username alone.
Try not to start a Tweet with a @username. When using a @username in a Tweet or sentence, consider the @ symbol silent. For example, instead of Player X signs a new contract @team it should be Player X signs a new contract with the @team.
We respect what people, teams, and organizations call themselves. We do not remove hyphens or accents. This also applies to names translated from other languages.
We do not capitalize on political titles unless they precede a name. The Queen is always capped, as is the Pope.
We write out one through nine, while 10 and above are written as numerals. Please avoid starting a sentence with a number, except in titles.
Sports scores are always written in numerals, as are temperatures.
For ages, use the numeral, such as "the girl is 15," and use hyphens for ages used as adjectives, e.g. "the 5-year-old boy."
Police shooting terminology
We should always strive for clear, specific language in our titles, descriptions and Tweets. We should avoid ambiguous and jargon-filled terms such as “police-involved shooting” or “officer-involved shooting”. These phrases make it unclear whether a police officer has been shot or whether the officer shot someone. Instead, we should specifically outline whatever details we know to be facts, with relevant attribution.
Wrong approach: A Black man died after a police-involved shooting in Wisconsin on Thursday.
Correct approach: A Wisconsin police officer shot and killed a Black man on Thursday.
Wrong approach: A man armed with knives died following an officer-involved shooting.
Right approach: A man armed with knives was shot and killed by police.
It’s OK to use Tweets including phrases such as “police-involved shooting” if they provide the necessary context.
All usual Curation policies surrounding allegations of wrongdoing apply.
Moments based on opinion polls must include enough information and context for people to evaluate the importance of the results. At a minimum, we should always cite the source of the poll, when it was fielded and what the sample size was. Titles and descriptions must also avoid exaggerating the meaning of the findings. Remember that polls can be wrong, are only a sample of public opinion, and people can change their minds between responding to one and casting their vote.
We should be clear about who conducted a poll and who funded it. The best polls are produced by non-partisan, independent polling groups and media groups. Any curation of polls which are paid for by candidates or interest groups must highlight the sponsor within the Moment so that readers can be aware of the potential bias.
Any reliable poll should include a margin of error. This should be taken into account when curating on the subject. For example, if a poll says the difference between two candidates is within the sampling error but not exactly the same, then the candidates are “about even.” If the margin of error is plus or minus three percentage points, and one candidate is ahead by around double that, the poll is clearly saying they are ahead. This distinction should be reflected in the title, description and annotations as appropriate. Please note, some polling agencies differentiate by percentage and some by percentage points. These are not the same thing. For example, moving up from 40% to 44% is a four percentage point increase, but it’s still a 10% increase.
In order to publish a Moment on a poll, you must make sure you understand the math and the comparisons or key findings.
We avoid profanity in our titles, descriptions, annotations, and Tweets from our @Moments accounts. Tweets used in Moments may contain profanity, but we prefer to feature Tweets without profanity whenever possible. Please remember there are cultural differences in regards to profanity. For example, a common phrase in Australia may be considered offensive in the US, or vice versa, and therefore should not be included in curation across markets.
Brackets and quotation marks
If there is a full sentence inside the parentheses or quote, the period/full stop should be inside. If only a sentence fragment or portion of the sentence is inside the brackets, the period should be outside. For example, "we have spelling differences between markets. (The UK and Canada spell travelled with two Ls.)" versus "There are punctuation differences between markets (the UK and Canada spell travelled with two Ls)."
We avoid using ellipses. If we do need to use one in a description, for example when abbreviating a quote, use a space on either side of the ellipsis.
Hyphens (-) are used to connect words such as ages, the 10-year-old boy, or to modify a term, a 15-point game.
We don’t use the Oxford comma. When writing a list, do not include a comma before the “and” or final list item unless required for clarity.
We default to ‘s for all singular possessive nouns (Chris’s), but follow Buzzfeed’s style on some exceptions:
- Proper nouns ending in “s” that make a “z” sound
- Corporations, brand names and proper names that are pluralized
- Words ending with an “s” sound before a word that begins with an “s” take an apostrophe and an “s” (for appearance's sake, but for goodness’ sake)
- Don’t use an apostrophe when a word is descriptive rather than possessive (teachers college, writers room)
- Personal pronouns never take apostrophes (hers, his)
Race and ethnicity
In most conversations we curate, we identify people by nationality, rather than race or ethnicity. We should not use these identifications interchangeably. We should only mention an individual’s race or ethnicity when a conversation is definitively about race.
When we do identify people’s race, we should provide a clear explanation of why it was relevant to do so. When specifying race, we prioritize how people self-identify and should not generalize or guess about anyone’s race or ethnicity.
We should be wary of coverage that results in a narrow, singular narrative about diverse groups of people, which can be seen in broad social issues. This could include the coverage of Black and Hispanic people in the US, or certain immigrant populations in Europe, or generalizing reactions by racial or regional classifications, e.g. Asian women love this new TV series.
We strive to make our Moments as globally accessible as possible. We try to avoid words in titles and descriptions that have different spellings or meanings in different markets, e.g. rumor vs. rumour, favorite vs. favourite. When there isn’t an elegant replacement word that can work across all markets, please default to the style of the country where the news is happening, e.g. a shopping mall for the US but shopping centre for the UK.
Tweets from our @Moments accounts should reflect the spelling for the associated region. For example, a Tweet that says “favorite” for the US should be edited to say “favourite” for the UK and Canada.
When referring to regions, please be specific. American can mean either North American or from the United States, while Britain does not apply to all of the United Kingdom (which comprises England, Scotland, Wales and Northern Ireland). By the same token, England and the United Kingdom cannot be used interchangeably.
Northern Ireland is on the same island as the sovereign state of Ireland, but is part of the United Kingdom. Ireland is a sovereign state and not part of the UK. Generally, if someone refers to Ireland they mean the sovereign state, but in cases where both countries are referenced and there may be confusion, including Brexit coverage, we can refer to the independent nation as the Republic of Ireland.
Sexual assault and domestic violence
When dealing with allegations of sexual assault and domestic violence, we prefer to use “survivor” or “alleged victim” rather than “victim” or “accuser” (unless in direct quotes) to describe people who were allegedly assaulted. We should not publish the name of the survivor unless they have come forward publicly. Use “alleged” to describe circumstances where claims have not been proven in court.
Please avoid sensationalizing language, such as graphic or invasive details. For example, name the crime as domestic violence instead of calling it a “tragedy” or “horror.”
We avoid words that can convey judgment, for example, choose “said” instead of “claimed,” “admitted” or “confessed.”
We also avoid words that shift blame to the victim and language that suggests pleasure or consent when curating conversations related to assault.
For detailed and potentially distressing coverage, use the sensitive tag, when appropriate.
For further reading, please see Level Up’s guidance on covering domestic violence and the Dart Center’s guide to reporting on sexual violence.
Sexual imagery and language
We avoid the use of sexual language, including sexual anatomy, and gratuitous imagery in our titles, descriptions, and covers.
We should only curate our own Moment based on a single piece of reporting or media (such as an article, video or podcast) if it is generating significant conversation on the platform, and is reflected in the top trends in the country. This includes profiles of celebrities, politicians, athletes, and other public figures.
For these Moments, include attribution in the title.
While people on Twitter react in real time, we don’t want to spoil anyone’s experience watching a show or movie. We avoid giving away major plot twists in the title, description, cover image or Tweet and should also include a warning for fans, for example, 🚨THIS MOMENT CONTAINS SPOILERS 🚨.
Please check release dates in other markets before targeting Moments that could contain spoilers.
We speak to the fans in the language they use. For major sports events we should provide context, but for other sports coverage, we may use team nicknames and insider terms.
The playoffs do not take a capital, but the Stanley Cup Playoffs as an event does. The same rule applies to series and games, so we would capitalize Game 3 and Round 2. We do not capitalize finals unless it’s part of a proper name such as the Western Conference Finals.
Postseason is one word.
Soccer in the US and Canada is football in the UK. We try to write around this problem by using the name of the league. If it’s about MLS it can default to soccer, whereas if it’s occurring in the UK it should be called football.
Suicide and self-harm
Mental health advocates recommend referring to deaths by suicide as "died by suicide," rather than using terms like "committed suicide" or "victim of suicide." The American Foundation for Suicide Prevention offers further guidance on reporting on suicide.
We should not label anything as a terror attack or terrorism until the country’s responsible investigative unit, such as the police or government, officially declares it as such. In the event of an unclear, developing breaking news event, please refrain from using “terror” or “attack” and instead use “incident” or “event.”
Titles and descriptions
Titles and descriptions should explain the nature and tone of the conversation, and provide context to make it relevant to a wide audience. Titles and descriptions should be neutral, and if they do convey a viewpoint, we should state the source of that opinion. Anything stated in the title and description should also be illustrated in a Tweet in the Moment.
Questions in titles should be used exceedingly sparingly, and we should not pose questions that are not fully answered in the Moment.
We write Moments titles in sentence case, not title case, so we only capitalize the first word of the title and any proper nouns. For example: 9 best looks from the red carpet, not This Baby Seal is Adorable. Note that while we avoid starting a sentence with a number in descriptions, we may do so in titles. Titles should be less than 10 words or 280 characters, while descriptions should generally be about two sentences.
We do not use periods at the end of titles. We try not to start a title with “Report,” instead use it in a sentence or attribute the report.
We should not use a comma in a title to replace the word “and.” This is a common newspaper convention, but it detracts from the conversational nature of our titles and is not commonly used in many markets, including the UK.
If we use a name or phrase in a title, avoid repeating it in the description. We should not repeat Tweets, or other Tweet content, in the title and description.
When translating text, keep the meaning as close as possible to the original, while also making the translated text easy to read and understand in the second language. This means it’s more important to keep the sense of the translated text as close as possible to the original than to have a word-for-word translation.
When sharing translated text in a title, description or annotation, indicate the original language and that the text has been translated, for example, The police officer, speaking in French, said the road closures would continue through the weekend.
We may quote translations from trusted government officials, news agencies and journalists.
We should not include words we do not recognize from foreign languages in our Moments.
Tweet is the word
We always capitalize the mark Tweet, regardless of how it is used. Retweet is also always capitalized.
We avoid using “Tweetstorm” or “thread” in Moments display copy.
The Curation team can curate Twitter Lists, choosing accounts related to a subject or event for people on Twitter to follow. Accounts added to a list are selected by the Curation team based on the account’s focus on a subject, and not filtered by keywords for specific Tweets. Accounts are included that:
Demonstrate experience or insight on the chosen subject
Have a track record of sharing engaging and informative content
Frequently post about the subject
Represent a wide range of perspectives
For political subjects, we take extra care to seek non-partisan accounts
Lists are monitored to ensure there are enough Tweets to reflect the live conversation, there is a variety of voices and perspectives being shown, and that most Tweets are on topic. Our goal is to highlight quality Tweets that represent accurate information. Our team receives regular training on social media verification practices and most lists undergo a peer review process before they are featured.
Nouns and verbs need to agree and use the correct tense. We use informal but correct language.
We must be consistent in the verb tense used throughout the Moment’s title, description, and Tweets from Moments accounts. Some Moments make sense to be positioned in the present, e.g. "beauty bloggers are painting their cheeks yellow," while things that are a done deal belong in the past tense, (e.g. "the team moved on to the conference finals.")
Moments content should be conversational, so we avoid perfect and progressive verb tenses, e.g. "The prime minister has announced a trip to China," when we could say, "the prime minister announced a trip to China."
A weather watch generally alerts the public to the possibility of extreme weather, while a warning means it is likely to occur. Please default to the language used by the national weather service in the affected region.
Climate is usually defined as the average weather. Climate change, meanwhile, is the variance in that climate “which is attributed directly or indirectly to human activity that alters the composition of the global atmosphere and which is in addition to natural climate variability observed over comparable time periods.” (Source: United Nations Intergovernmental Panel on Climate Change (IPCC).)
In simpler terms: there’s a consensus among leading scientific and international bodies, including the United Nations, that rising greenhouse gas levels, caused largely by burning fossil fuels, has been linked to increases in the average global temperature on Earth.
It is appropriate to acknowledge in our curation that global warming is a scientifically-proven phenomenon, citing the UN IPCC or another appropriate official source.
While our aim is to present a breadth of viewpoints on all topics we curate, to achieve impartiality we do not need to include outright denials of the rising temperatures. We should include different viewpoints on the rates of change, predictions and suggested responses.
Scientific experts are beginning to prefer the term “climate emergency” as a more accurate representation of the situation. As The Guardian’s Style Guide explains, the term climate change “is no longer considered to accurately reflect the seriousness of the situation.” This term is also used by Fundeu in our Spanish markets. Meteorologists, including those at the UK’s MET office, prefer “global heating” to “global warming” (which is the gradual increase in global surface temperature). When appropriate, and when we believe these terms will be clear to the people reading our curation (for example, if they are used in curated Tweets), please feel free to use these updated terms.
Temperatures get higher or lower, not warmer or cooler. Temperature figures are written as numerals. When targeting a Moment globally, we include both Fahrenheit and Celsius.
Storms, hurricanes, and typhoons
Storms are assigned a type based on the speed of winds and the location where they form. Hurricanes form in the North Atlantic Ocean and Northeast Pacific, while cyclones form over the South Pacific and the Indian Ocean. Typhoons form over the Northwest Pacific Ocean. Hurricane categories are based on wind speeds. Those ranked as category three and above are considered major storms. When calling out wind speeds in a global Moment, we should provide both kilometres per hour and miles per hour. For storms in the US and UK, lead with miles, like 160 mph (260 kph), while global storms should be formatted with kilometres per hour first, so 260 kph (160 mph).
We capitalize hurricane when it is part of the name forecasters have given the storm, e.g. Hurricane Irma. We refer to hurricanes as “it” rather than he or she.
Guidance from the National Weather Service suggests avoiding the term “weakened” in storm coverage, as it can downplay the perceived threat from high winds and rain. Once a storm has been downgraded to a tropical storm or depression, it can be referred to by name or qualified as remnants, e.g. Residents clean up from Harvey flooding or Cities brace for remnants of Hurricane Maria.
Earthquakes are registered in magnitude and, according to AP style, should not include a hyphen, so it’s a magnitude 5.9 earthquake or a 5.9 magnitude earthquake. Earthquakes generally need to have a magnitude of 2.5 to 3 to be felt by people, while they begin to cause moderate to severe damage above 4 or 5. It can sometimes take days before seismologists have calculated the final magnitude, so we try to hedge early figures with “preliminary” or “estimated” and cite the reporting agency, for example: “The USGS estimates the earthquake had a magnitude of 4.3.”