Terminology

Lyric (or line): One line of a song. Some songs/videos have non-music lines, like spoken dialogue. These should ideally also be included in the lyrics.

Curated: Both the lyrics and words for this song have been carefully edited, and this song is ready for use by Roxxem users.

Song Description: Write 2-3 sentences describing what the song is about. This description is meant to excite our users about the song and make them want to learn it! You can also include key language tasks that users can learn in the song (in the general format “Learn how to… [task]”). This description should be 1 paragraph or less, ~2-3 sentences. We've found that around ~45-50 words feels good. No more than 60 words.

Example: “The breakout smash hit from superstar Mexican singer and actress Thalía. This sensual and vivacious Latin pop song was not only a huge crossover success, its black-and-white music video made Thalía into a 90's style icon. Sing along to one of the songs of the decade! Learn words for feelings, practice metaphors, and more.” (54 words)

Maturity: MPAA-aligned ratings, using this website as a guide to determine the maturity rating of the song (G, PG, PG-13, R); Roxxem will be used in American classrooms, and as such teachers need to be able to easily filter by maturity rating

Start Time: The time at which the line starts (maximum two decimal places)

Stop Time: The time at which the line stops (maximum two decimal places)

Difficulty: Roxxem rates songs and words on a scale of 1-10, equivalent to the Common European Framework of Reference for Languages (CEFR) proficiency levels as follows:

• 1: A1 - Words that are common and simple, usually understood by beginners. These are basic everyday vocabulary.
• 2: A2 - Words that are slightly more complex but still common, known by - elementary learners.
• 3: A2+ - Words that are slightly beyond elementary level, used in basic conversations and simple descriptions.
• 4: B1 - Intermediate level words that are used more frequently in everyday language and familiar to intermediate learners.
• 5: B1+ - Words that are more advanced than intermediate, used in more complex sentences and conversations.
• 6: B2 - Upper-intermediate level words that are common in academic and professional settings.
• 7: B2+ - Words that are more sophisticated and used in more advanced texts and conversations.
• 8: C1 - Advanced level words that are less common and more specialized, known by advanced learners.
• 9: C2 - Highly advanced words that are rarely used in everyday conversation, often found in academic or technical texts.
• 10: Above C2 - Words that are extremely rare, specialized, or technical, typically known by experts in a particular field.

Videos

A. Generally, songs should be OFFICIAL MUSIC VIDEOS, not audio, posted by the artist themselves. Audio only is not great for Roxxem (but exceptions can be considered). Official videos are also unlikely to be deleted!

B. NO DUPLICATE VERSIONS OF SONGS. (Remixes, lyric versions, audio, whatever, etc.). This is confusing. Exceptions should be carefully considered > i.e. there shouldn't be multiple “Recuérdame” versions -- should just pick one as the only curated version

C. Unfortunately, not all Youtube videos are available in all countries. “Video unavailable: The uploader has not made this video available in your country.” We need to make sure that the Youtube videos we use are VIEWABLE IN THE UNITED STATES AND CANADA (where most of our audience is)--and, ideally, available in all countries.

Here is a website to check the availability of Youtube videos in each country: https://watannetwork.com/tools/blocked/

Adding Lyrics:

Lyrics will be auto-generated where possible. Otherwise, you will need to transcribe them by hand. (Do not copy from other sources)

Lyric Line Lengths:

Roxxem is designed for LANGUAGE LEARNERS to learn the lyrics, not just sing them, so our purposes are different from songwriters.

Lyrics should generally make sense as a single lyric. Each Roxxem lyric is a lesson, so it should kind of make sense as a complete thought (sentence or phrase).

Generally, Roxxem lyrics should follow song lyrics as subtitled in the music video, if there are any–i.e. the same line should also appear in Roxxem. But not always!!! See these important rules below.

Often lyrics you find elsewhere are too short (like one word, or something). This is not good for Roxxem.

Squash lyrics together if they're too short and make more sense as a single lyric. This is doubly important because it makes more semantic sense and also because it makes our games easier to play.

Example: in the song “”, combine “Cette nuit” and “Intenable insomnie” into 1 lyric (new combined lyric: “Cette nuit intenable insomnie”)

Of course, lyrics also can't be too long… see the rules below. Have to use your own judgment here--there's a lot of wiggle room.

• For Spanish and French songs, we've found that around “64 characters” is a pretty good upper length for a lyric. Lyrics should not go much about this character count.
• Ideally, a lyric should be 3+ seconds long.
• Any kind of lyric that's like 1-2 seconds long is not good for Roxxem. This is too short.
• It's too short a length for a lesson, the video player doesn't respond well to this super short duration, and it makes it really hard to play the games.
• A lyric CANNOT GO above eight word/phrase segments (see note above).
• A lyric shouldn't be ONE segment, either, ideally. A lyric that is one word long makes the games pretty bad.
• Our experience is that lyrics which are ~4-6 segments long, feel good for the lesson and for the games. But, it all depends on the lyric. Some lyrics may only have 1 or 2 segments. Segment the lyric in a way that makes sense. For example, ““Me enamoré” = “I fell in love”

Editing Lyrics and Translations

Lyrics SHOULD NOT go over EIGHT separate words/phrase chunks in total. ( Probably a lyric that's so long should be split, or some of the words can be combined into phrases.)

See section above for some notes on how long the lyric should be.

DELETE empty lyrics which have little to no content (i.e. lyrics not in the target language), this is distracting to users and not helpful.

• Example: “Ay!”, “D.Y!”, “Fonsi!”, things like that
• Exceptions: Waka Waka by Shakira

Tag if lyric should be skipped

We don't want users to waste their time learning lyrics that they've already learned in this song

These lyrics will be skipped when learning the song

A lyric that is either identical (or almost identical) to a previous lyric in the song is a DUPLICATE LYRIC

A lyric that has no word or phrases associated with it is an EMPTY LYRIC

A SKIP LYRIC is a lyric that, for whatever reason, should be skipped when learning

Generally, most Duplicate Lyrics and Empty Lyrics should also be tagged as Skip Lyrics

Use your judgment to skip other lyrics. For example, the lyric is only a bunch of random word fillers, names, etc.

Translations should be fairly literal, almost word by word

The main point here is to understand the meaning of the lyric in the target language after all

There should be a balance between meaning and literal translation, especially as it relates to word order (i.e. “漂亮的让我面红的可爱女人 = Adorable woman whose beauty makes me blush”), we do want learners to have a flavor of the target language – more guidance on this later

That being said, if possible, translation should sound more or less natural in the language it is being translated to (accuracy is still more important!)

Use parentheses for lyrics that you would not expect to see in a dictionary of the language (i.e. lyrics that are not in the target language)

• Example: “Say yeah, yeah, yeah” in a French song should be enclosed in parentheses
• Example: The artist calling out their name or the name of a collaborator should be enclosed in parentheses

Use commas (,) when there is a clear run on sentence or phrase

Do not use commas at the end of lines

No periods (.) anywhere

First letter of lyric should be capitalized

Should delete “sound effects words” like “shhh”, “uh”, “ayyy”, or other words that have no real meaning in any language and are not important to the meaning of the lyric/song

All numbers in the lyric should be spelled out (i.e. “diez” instead of “10)

• Exception: things like telephone numbers, addresses, dates and years, etc.
• Ex: for a lyric like “3615 Bonheur”: we wouldn't spell it out, it should be written as “3615” in the lyrics themselves.

Editing Words and Phrases (Segments)

Words should be chunked into PHRASES if they mean something together, or if they are individually unclear (in the context of the line)

Bias our mindset in favor of PHRASES (multiple words chunked together) instead of individual words, which are often meaningless outside of the grammatical and pragmatic context

It's fine for an entire lyric to only be composed of a few phrases, or even just one!

Example: “Me enamoré” = “I fell in love”

The Words and Phrases list should include ALL words in the lyric. Duplicate words and phrases should be included.

• Example: The lyric “Me enamoré, me ena-na-namoré” has two duplicate phrases in it. The word list should look like this: [me enamoré, me enamoré]
• Example: The lyric “Sube, sube, sube” should have the word list: [sube, sube, sube]

Exception: words in “( )” aka. parentheses

If some of the words in the lyrics are in “( )“, that means these words are NOT part of the main lyric, and SHOULD NOT be added to the word list.

• For example, line 24 of “PLUS FORT”: "es superhéros (plus fort, on est plus fort)" should have the following word list: [“des superhéros”]
• Yep, that's it :)

The reason is that the words in parenthesis are extraneous. This happens when there are overlapping or background singers, the line has random sounds like “oh-oh-oh” or “la-la-la”, or there is non-native language (people speaking English in a French song). All of these are not important to learners and not part of the main lyric, so we should put them in parenthesis. And we don't put them in the word list, because that works better with our lessons/games.

Remove non-standard, fake, weird words like “ena-na-namoré” – replace with the actual standard form of the word

Phrases should be translated when the combined words have a different meaning than the individual words

• Example: “lo” means “it” and “que” means “that” but “lo que” means “what”
• Example: “tengo” means “I have” and “hambre” means “hunger” but “tengo hambre” means “I'm hungry”
• Example: if “Cuando me levanto tengo hambre” is the entire lyric, “tengo hambre” should be translated as a phrase

base verb form infinitives should be translated with “to” (like respirar = “to breathe”)

Words and phrases should not be capitalized unless it's a proper noun (i.e. “la Seine”)

Translations should not be capitalized unless they include a capitalized English word like “I”

When a word has multiple common translations, the translations should be separated by a semi-colon, followed by a space

• Example: “to; that”

Subject drop languages (Spanish, Chinese, etc.): if the subject is dropped in the line, the subject should be included in the translation of the word

• Example: “Tengo hambre” - I'm hungry

Slang and abbreviations

Slang words and abbreviations SHOULD be accurately transcribed as they are spoken

• Ex. "pa'que" is an abbreviation of para que, used commonly in colloquial Spanish

Roxxem teachers real-world language. Therefore, "pa'que" should be transcribed as that word. Add an annotation that it is a colloquial abbreviation of "para que", and you can also link it to the base word (lemma) "para que

Words with multiple definitions

Each word should only have one definition. In fact, Roxxem "words" are actually more like "definitions" (a distinct unit of meaning). In order to add a new word definition, you should create a new word.

Please look through the existing word definitions to see if the definition you are adding already exists. If it does, you should use the existing word.

Words that are not in the target language

Some lyrics may have words that are in the target language. For example, a Spanish lyric with some words in English, or a French song with some words in Arabic.

Roxxem supports adding words in multiple languages (not all languages are supported). When adding words, change the selected language to search and/or generate words in that language.

Lyrics that are completely in another language (for example, the lyric is 100% in English) should simply be marked as “Skip” and “Empty.” The word list should be left blank.

If the lyric is a mix of target and non-target words, follow these rules: -You should include all semantically meaningful words in the word list, including words in other languages. -These words should still be translated into English. English words should have a translation that is identical to the word itself.

Timestamping and syncing

When timestamping lyrics, it's important to make sure that all the words in the lyric are included between the lyric start time and the lyric end time. No words should be cut off when playing back the video of the lyric. All words should be fully audible. It's actually good for there to be some extra time “padding” (for example, ~1 second of non-vocal music) at the beginning and the end of the lyric. This is because the video playback is not exact and it's easy for words to get cut off. So you should bias towards adding some extra time at the start and the end of the lyric.. It's especially important to add some extra time padding at the beginning of the lyric. At the same time, there shouldn't be too much extra time padding. Over 2-3 seconds of extra time padding (in one direction) is too much. Our lyrics are constantly being rewinded by learners, so we don't want to waste their time listening to a lot of dead space. Even after timestamping with the “sync lyric” function, you should still individually check the playback of every single lyric. Make sure that no words are being cut off..

Lyric Start and Stop Times

-Lyrics CANNOT overlap in time. The stop time of a lyric should also never be the same as the start time of the following lyric. Each lyric must start at least 0.01 seconds after the previous lyric ends.

Overlapping Lyrics

In some cases, the words of lyrics actually overlap. This could be because there are multiple singers, or because lines have been recorded on top of each other.

Regardless, we still need to follow our rules outlined above. What to do?

Simply, you need to cut off one of the lyrics. (Usually, the lyric earlier in time. The overlapping words in the previous lyric are usually the least important.)

First, identify the words that are overlapping with another lyric. Second, mark these words in parentheses. Third, delete those words from the word list.

Example: (In Lyric 36, the third “lass uns geh'n” phrase overlaps in time with Line 37)

General Rules of Thumb:

-verb infinitives should be written as “to [verb]”

Spanish: Rules of Thumb:

• include subject with verbs (i.e. [yo soy] = phrase)
• Chunk pronouns with nouns: “mi pelo”, “tu mirada”, “mi suerte”, etc. [OR MAYBE NOT??? should probably change this ]
• articles like “la”, “el”, “un”, “una” (definite and indefinite articles) should be chunked together with the word that follows [OR MAYBE NOT??? should probably change this ]
• chunk “no” and a verb (like “no llores”)
• Translate phrases with articles as follows:
  • un perro -> a dog
  • una perra -> a dog
  • El perro -> the dog
  • La perra -> the dog
  • Los perros -> the dogs
  • Las perras -> the dogs
• Transcribe abbreviations like pa' accurately (i.e. keep the shortened abbreviation, but attach an annotation that this is an abbreviation of a longer word)
  • For example: “pa' que” [annotation: this is an abbreviation of “para que”, used commonly in colloquial Spanish]

French: Rules of Thumb:

• articles like “la”, “el”, “un”, “una” should be chunked together with the word that follows
• pronouns should also be chunked (i.e. “je suis”, “j'ai ta voiture”)
• Translate phrases with articles as follows:
  • un chien -> a dog
  • une chienne -> a dog
  • des chiens -> some dogs
  • des chiennes -> some dogs
  • le chien -> the dog
  • la chienne -> the dog
  • les chiens -> the dogs
  • les chiennes -> the dogs
• For contractions with “que” (i.e. “qu'un choix”), please write it WITH the contraction.
  • Example: “qu'un choix”
  • Annotation: "contraction of 'que un choix'

German: Rules of Thumb:

• Conjugations should be chunked together in a segment
• Articles with gender should be chunked together in a segment
• Chunk prepositional phrases like “für mich” and “für dich” into a segment

SEPARABLE VERBS: Unfortunately, Roxxem does NOT support separable verbs at this time.

They will just have to be separated into individual words. Or, alternatively, you can turn the entire phrase bookended by the separable verb into a segment.

Example: “Hört der Regen auf Straßen zu füll'n” Segments (Version 1): Hört, der, Regen, auf, Straßen, zu, füll'n Segments (Version 2): Hört der Regen auf, Straßen, zu, füll'n

Turning the entire separable verb word string into a segment may be the ideal solution; however, this segment cannot also be too long. (It would not make sense to turn an entire sentence into a segment). You would need to use your judgment here as to what is best.

Chinese: Rules of Thumb:

TBD

English: Rules of Thumb:

TBD