اصول پیاده سازی
دوشنبه, ۱۱ شهریور ۱۳۹۸، ۱۲:۳۸ ق.ظ
به دلیل حجم زیاد مطلب، متن ترجمه نشده است.
Style Requirements
Style requirements vary from Basic to Verbatim Transcriptions. Refer to one of the two sections below depending on which kind of task you are working on.
General Rules
The following rules apply to both Basic and Verbatim Transcriptions.
Sound Events that Interrupt Dialogue
When background noise occurs, use brackets around a short description of explanation of the sound. This can include silence or a specific tone. Try to be as descriptive as possible in 1-3 words. For example:
-
Background noise - When there is background noise that is not speech, indicate it with a note.
- [laughter], [applause], [phone ringing] and [music] are all acceptable tags that indicate what is happening in the speaker's environment.
-
Silence - In the case of a jarring or prolonged silence, leave a [silence] note.
- If a speaker cuts off or stops and is notable, use [cuts off] or an ellipsis with no spaces. For example:
"I really think... She is mad at me.”
- If a speaker cuts off or stops and is notable, use [cuts off] or an ellipsis with no spaces. For example:
- Tone indicators - If a speaker is angry or is using a joking tone, include an [angry] or [joking] note before the word/phrase.
If the speaker must stop because of a sound event, then include the bracketed note in its own paragraph. Then, continue with the speaker in a new paragraph.
Examples:
Correct
Patient:
The pain is in this area...
[gasp]
Patient:
...below the knee.
Doctor:
How strong?
Patient:
I would rate it around a six.
Incorrect
Patient:
The pain is in this area [gasp] below the knee.
Doctor:
How strong?
Patient:
I would rate it around a six.
Showing Uncertainty
Indicate when you are uncertain of a spoken word or a phrase by placing brackets ,"[ ]", around one of the following indicators before each word/phrase in question:
[crosstalk] - Use this note when two speakers are talking over one another. Try to get as much from each speaker as possible.
- For example, if a patient is talking and the doctor interrupts with a question that is answered, do the following:
Patient:The pain is in this area [crosstalk] below the knee.
Doctor:How strong?
Patient:I would rate it around a six.The doctor interjected before the patient finished, but we do not want to split the sentence into two paragraphs. Therefore, the patient's sentence is completed, and then the doctor's question is inserted into a new paragraph.
[inaudible][00:00] - Use this when you are not able to guess what was said.
- It is very important that you use this instead of leaving words out without letting us know.
- Additionally, please make sure to mark the time (or time range) of the inaudible language.
[foreign] - Use this note when a language other than English is being spoken.
[phonetic] - Use this note when you are unsure of the spelling of a word, title or name, but can make out the sounds. After using this note, you will spell the word out phonetically, to the best of your ability.
- Use this note instead of [inaudible] if you can make out parts of a word but not the whole thing.
- Use this note for unfamiliar words you’re not sure how to spell, such as scientific or medical terms.
Timestamps
Each transcription requiring timestamps must include timestamps at least once every 30 seconds and at every change in speaker..
Place timestamps in brackets - [00:30]. The timestamp should be placed after the speaker's name and before the transcribed language is shown. If there is only one speaker, the timestamps should be inserted every 30 seconds within the transcript.
Correct
Dave:
[00:30] How are you?
Frank:
[00:40] Good. How are you?
Dave:
[01:05] Great!
Incorrect
Dave: How are you? [00:30]
Frank: Good. How are you?
[01:00]
Dave: Great!
Always use only minutes and seconds when inserting a timestamp. Thirty seconds is represented as [00:30], and one hour and 10 minutes is represented as [70:00].
Speaker Labels
Speaker labels must always be used with a colon, include a hanging indent and be bolded. The goal is to make all of the speakers' names stand out so that the transcription is more digestible to the reader.
Correct
Dave:
This was a nice meeting.
Steve:
Yes! I agree! When should we meet next?
Dave:
Let's meet on Tuesday for lunch.
Incorrect
Dave- This was a nice meeting.
Steve- Yes! I agree! When should we meet next?
Dave- Let's meet on Tuesday for lunch.
Use a speaker label whenever speakers change or if there is a long pause.
Only use names, titles and gender as speaker labels. Use labels in this order:
-
Name
- Use the full name (if available) on first reference.
- Use the speaker's first name for the remainder of the transcription.
- If the speaker's title is known, include the title with each label. Use with the last name (Dr. Mathews).
-
Title
- Use if the title is known but the speaker's name is not.
- Include gender to add more description to the speaker.
- Doctor, Manager, Male Patient, Female Professor
-
Gender (Woman or Man only)
- Use the speaker's gender if his/her name and title is not available.
- Man, Woman
In the event that there is more than one undefined speaker of the same gender, title or role, use numbers to separate them. “Man 1” “Man 2”
Make each label as descriptive as possible. For example, "Female Reviewer" is much more helpful than "Reviewer" would be.
If there is a large group talking, then refer to them as a whole body, such as "Audience." Additionally, if a single man from the audience talks, then refer to him as "Audience Member" rather than "Man."
Basic & Verbatim
Basic Transcriptions Rules
Basic transcripts should be grammatically correct throughout. Refer to the CrowdSource Style Guide for all requirements regarding punctuation, grammar, spelling and syntax.
There are also a number of additional style rules you must follow listed below. Check this document often, as these lists may grow.
Persistent Sound Events that Do Not Interrupt Dialogue
When a sound event occurs that does not interrupt dialogue (typing sounds, the AC kicking in), note it in the transcription once at the point it occurred. For example:
Correct
[typing sounds throughout]
Susie:
Let’s get burritos for dinner tonight.Randy:
That would work.
Filler Words and Statements
Remove filler statements such as “ugh”, “um”, “yeah” and so on from your finished transcription. For example:
Correct
Randy:
That would work.
Incorrect
Randy:
Yeah, that would work.
Conjunctions that Start a Sentence
Remove conjunctions (And, but, etc.) that start a speaker’s sentence. For example:
Correct
Flo:
That’s how I bought a cockroach.
Incorrect
Flo:
And that’s how I bought a cockroach.
False Sentence Starts
Remove false starts (unnecessary repetition or instantaneous revision of words) in a speaker’s sentence. For example:
Correct
Flo:
I would like to think that I’d have bought it again, if I could do it all over.
Incorrect
Flo:
I would like to think, I would like to think that I’d have bought it again, if I could do it all over.
Special Punctuation Rules and Exceptions
Check this section often, as the list of exceptions to Basic style guide punctuation rules and special, additional rules may grow with time.
Colons and Semicolons
Generally, colons and semicolons should be used rarely, if ever, in a transcription. Stick to the most basic forms of punctuation whenever possible.
Verbatim Transcriptions
Because it’s necessary to capture nonverbal and verbal nuances in verbatim transcriptions, the style requirements are somewhat more complex than those of Basic Transcriptions.
False starts, filler, repetitive phrasing and slang should be included in the transcripts. If the speaker clearly says "cuz," then write "cuz" and not "cause."
Punctuation should be accurate, but do not make any grammatical changes to the transcript.
Sound Events that Do Not Interrupt Dialogue
In verbatim transcriptions, indicate every time an unobtrusive sound event (typing noises) occurs, even if it does not interrupt dialogue. For example:
Correct
[typing sounds]
Johnnie:
That was the day I knew I would become a clown.
[typing sounds]
Barbera:
That’s fascinating.
[typing sounds]
Filler Words and Statements
Retain filler statements such as “ugh”, “um”, “yeah” and so on in your transcription. For example:
Correct
Randy:
Yeah, that would work.
Conjunctions that Start a Sentence
Retain conjunctions that start a sentence. For example:
Correct
Johnson:
And he just flew away.
False Sentence Starts
Write out false sentence starts verbatim. For example:
Correct
Johnson:
And I think, I think I would like the steak tonight.
Special Punctuation Rules and Exceptions
You will need to use specialized punctuation and apply more common punctuation in unique ways to capture some of the nuances in verbatim speech. Please refer to this list in lieu of the CrowdSource Style Guide in most cases, with one exception listed below.
Basic Punctuation (periods, capital letters, etc.)
At the most basic level, you need to use appropriate punctuation, even in verbatim transcription. For example:
- A period ends a sentence or sentence fragment or phrase, unless the person is cut off. You’ll read more about that in the Abrupt Speaker Shifts section.
- The first letter in each new sentence or in a new fragment is capitalized.
- Commas are applied where they normally would be under Basic Transcription circumstances unless something about the speech calls for one to be omitted or included.
Stutters and Repetition
Indicate a stutter mid word or the unnecessary repetition of a word with a hyphen. For example:
Correct
Joey:
So I-I'm clairvoyant, and I'll be watching what I call a TV screen that goes on in-in my mind throughout the whole reading.
Speaker Pauses
To indicate a mid-statement pause, use an em dash (—) with a space before and after. For example:
Correct
Christine:
Wait a second — did you hear that?
Abrupt Speaker Shifts and Ensuing Sentence Fragments
As with Basic Transcriptions, use an em dash to indicate a sudden change of speaker (as would happen if he or she were interrupted) or to set off a sentence fragment from a subsequent complete sentence. For example:
Correct
Tim:
I love it. Okay, so let me just explain to you again, um, you know, how I'm — just get my information, and then we'll just jump right into it, okay?
Speaker Trailing Off
If a speaker trails off in the middle of a thought, use ellipses (...) to indicate that the speaker has left the thought unfinished. This is different from an abrupt speaker shift because the speaker does not finish the thought and does not change thoughts. For example:
Correct
Stumpkin:
Oh it’s beautiful. And uh learning about the Medicis was just great. It was just...
رفرنس: