1 Posted by Philip Withnall on 24 July 2007 | Permalink

I'd say they should be blockquotes, since they're closer to quotations than pure prose. Even if the quotation is edited slightly, it's still the person's thoughts and ideas, and should still convey what they wanted to get across, using most of their words. (If this isn't the case then there's some shoddy journalism going on.)

2 Posted by Rob on 24 July 2007 | Permalink

I think the / more accurately describes the question/answer pair than anything using a blockquote, since it retains the pairing AND gives overall structure to the interview. What would the blockquote construction be: ? ?

With the exception of the blockquote that feels almost like arbitrary markup and retains no connection between the questions and the answers. A definition list not only retains the connection between the questions and answers but also retains the overall relationship between each question and answer pair.

3 Posted by Rob on 24 July 2007 | Permalink

damn, it looked fine in preview. here it is again without brackets.

I think the dt / dd pair more accurately describes the question/answer pair than anything using a blockquote, since it retains the pairing AND gives overall structure to the interview. What would the blockquote construction be: div - p - blockquote ? div - h* - blockquote ?

With the exception of the blockquote that feels like arbitrary markup and retains no connection between the questions and the answers. A definition list not only retains the connection between the questions and answers but also retains the overall relationship between each question and answer pair.

4 Posted by Tino Zijdel on 24 July 2007 | Permalink

WA1.0 (and thus the base document for HTML5) says about blockquote:

The blockquote element represents a section that is quoted from another source.
( http://www.whatwg.org/specs/web-apps/current-work/multipage/section-sections.html#the-blockquote )

and about dl:

Note: The dl element is inappropriate for marking up dialogue, since dialogue is ordered (each speaker/line pair comes after the next). For an example of how to mark up dialogue, see the dialog element.
( http://www.whatwg.org/specs/web-apps/current-work/multipage/section-lists0.html#the-dl )

However, dialog is also marked up with dt/dd pairs, and since dialog is not an existing element in HTML4/XHTML1 using dl with dt/dd would imo be the best compromise.

5 Posted by Quentin on 24 July 2007 | Permalink

I think this is up to the author.

By comparison : in software development, there is always more than one way to fit with the "good practice" (here the guidelines), and not always a "best" way to implement but several ones with their advantages and disadvantages : more efficient, better designed and evolutive, closer to the specification's way...

So i suppose both "blockquote" and "dt/dd" are compliant with the guidelines and may be acceptable.
In my opinion, Blockquote is more semantically correct and dt/dd is more structurally correct.

as long as there is no "interviewQuestion" tag and "interviewResponse" tag, there is always a part of subjectivity.

6 Posted by David on 24 July 2007 | Permalink

Agreed, dt/dd would be the most appropriate at this point imo. Since dialog in HTML 5 uses a dt/dd structure, dl is the closest you get in HTML 4.

7 Posted by Eric Meyer on 24 July 2007 | Permalink

Fascinating: we just had this discussion on the AEA team, related to our "Better Know A Speaker" series. (I know, I sound like I'm plugging. Sorry!)

The original BKAS posts' markup had the questions as paragraphs, and the answers in blockquotes (with paragraphs in them, of course), just as you illustrated. Some called that an abuse of blockquote, while others felt it was exactly correct. We didn't use the dt/dd construct to encapsulate questions and answers, though we considered it.

You observe that interview answers "are almost never direct, literal quotations of the interviewed person." I agree with you when it comes to in-person interviews. In our case, though, since the interviews were conducted by e-mail, we were directly quoting the answers with no alterations (save correction of spelling errors). Does that change things?

And I don't know if this complicates the picture any further, but we should also consider that in the case of interviews, it will be very rare to have something to 'cite'. Does the lack of an attribute value indicate the element is inapplicable?

In the end, we took out the blockquotes, and just ran paragraphs. It will be interesting to see if people think we were right to do so, or not.

8 Posted by ppk on 24 July 2007 | Permalink

Thanks for all the answers so far; this is clearly a tricky problem.

Eric, I considered e-mail interviews only after writing this entry, but I'd say that in theory the interview mode should not matter.

As to citing, maybe we should define a quotation as a bit of text that has a (more or less) public source we can refer to. In that case an e-mail interview would not consist of quotations, since the original (the mail) is not public.

In fact, I'd say that the interview (e-mail or not) becomes the source that others can quote from, but does not itself have a source.

Does that make sense?

9 Posted by Alejandro Moreno on 24 July 2007 | Permalink

A good interview is typically a dialog, and HTML 5 has a good solution for them, IMO; a -dialog- tag that includes zero or more pairs of -dt-, -dd- tags. The Working Draft has this to say about them:

"Text in a -dt- element in a -dialog- element is implicitly the source of the text given in the following -dd- element, and the contents of the -dd- element are implicitly a quote from that speaker. There is thus no need to include -cite-, -q-, or -blockquote- elements in this markup. Indeed, a -q- element inside a -dd- element in a conversation would actually imply the person talking were themselves quoting someone else."
http://www.whatwg.org/specs/web-apps/current-work/multipage/section-prose.html#dialog

10 Posted by ppk on 24 July 2007 | Permalink

One more remark: for my report HTML 5 is irrelevant. The Web Guidelines are based on HTML 4, and my eventual answer will have to comply with the Web Guidelines.

11 Posted by quentin on 24 July 2007 | Permalink

What if the video of the interview can be found on youtube ?
is it a source ?

will you change the markup because the video exists ?

12 Posted by Richard Walter on 24 July 2007 | Permalink

The observation that an interview is a dialog between two people is the relevant idea. So I think that both the and dd/dt structures should be used!

The blockquote should wrap the dd/dt structure!! The blockquote should wrap the entire conversation since you are quoting both people, not just the one who answers the questions. The dt/dd structure identfies (in the current absence of ) each speaker.

13 Posted by Borgar on 24 July 2007 | Permalink

PPK, your #8 comment is on the money in my opinion.

Would you put quotation marks around the answers? If yes then the blockquote might be justified.

If you want cater for screen readers, then a p>em / p pattern will get the readers to use different voices for different people.

Better still, headings for questions and paragraphs for answers will allow nice document outline generation.

In the end, I don't think there is a fixed rule for this, and I doubt we want one.

14 Posted by Hristo Drumev on 24 July 2007 | Permalink

I think that dt/dd are not suitable, because they should be used for a specific terms/data, wich are not always present in an interview. For instance we dont have always a concrete response.
The combination paragraph/blockquote it seems to me good for an interview transmission, because the question for instance has not a concrete attitude for a quote- we use paragraph, but the responses has- using a blockquote is acceptable for them.

15 Posted by AlastairC on 24 July 2007 | Permalink

Is it reasonable to say that the intent for using a quote (block or inline) is to identify another speaker/source?

Assuming that the page/site has a 'voice', even if it's an organisation, a quote is then useful to programmatically identify sections that are not from the same source.

I can't help feeling that even if the definition isn't very precise, that it was intended to identify external sources (thus the optional cite attribute).

In that case I would say that the report was right, however, I'm assuming it is a quote, rather than a conversation.

Mr Clark's previous example in this area might help?
http://interviews.slashdot.org/article.pl?sid=02/12/09/1446221

Thinking of various devices, I think a quote is (potentially) more useful than just paragraphs. Even if you use the dt/dl, I would still include quotes for those bits that are not 'you', whether you can cite them or not.

16 Posted by AlastairC on 24 July 2007 | Permalink

Oh, sorry, I forgot Tantek covered this as well:
http://tantek.com/presentations/2005/09/elements-of-xhtml/#slide9

17 Posted by Scott on 24 July 2007 | Permalink

Interesting topic. While the definition list does seem like a viable solution, I am curious why no one has brought up the quote element. The inline q tag is part of the spec and could be used in conjunction with a p tag like so:
-p- Interviewer: -q-Is this mark up semantic? -/q- -/p-
-p- Interviewee: -q-Yes, I do believe it is. -/q- -/p-

18 Posted by Nicholas C. Zakas on 24 July 2007 | Permalink

It may be helpful to cite writing style guides in this case. In America, the APA style guide says that information taken from any source outside of the author's immediate knowledge, and is a word-for-word reproduction, must be quoted. In HTML, I'd envision this as being a blockquote.

If the answers in the report are paraphrases or summaries, then they're not considered quotes though still should cite the author.

19 Posted by Tino Zijdel on 24 July 2007 | Permalink

PPK: I'm a bit surprised by your remark that HTML5 would be irrelevant. Surely you know that HTML5 aims to more closely match the real-world usage of HTML (call it progressing insight), and that HTML4 is largely outdated as a guideline (and should by no means be considered a set of rules carved into stone).

But if you want some advice from the HTML4.01 specification, it's right there:

"Another application of DL, for example, is for marking up dialogues, with each DT naming a speaker, and each DD containing his or her words."
( http://www.w3.org/TR/html4/struct/lists.html#h-10.3 )

20 Posted by Marty Alchin on 24 July 2007 | Permalink

I've always preferred <dt> and <dd> for Q&A segments, whether it be in a FAQ, product feature description, or whatever. In my mind, an interview is simply that, with the journalist asking, and the subject responding. However, there's the added fact that the subject's responses are in fact quotes in the journalistic sense.

Personally, I'd expect to see <dt> and <dd> for the overall structure, with <blockquote> used with <dd> to indicate the response as a quotation.

As for citing, it's unfortunate that RFC2396 disallows fragment identifiers as part of a URI. If it did, something like <blockquote cite="#source-info"> would be best, in my opinion. The page would then be required to have an element with an id of "source-info" and that element would be responsible for describing the source. This could be just a paragraph describing the date, time, place and subject of the interview, or it could be external media (audio or video) recording during the interview. That way, there are no external sources, but each <blockquote> can still cite a reference. But, alas, this is all moot, since those don't seem to be allowed.

21 Posted by Lasse Rintakumpu on 24 July 2007 | Permalink

In his book "Transcending CSS" Andy Clarke suggests marking a conversation as an ordered list. Interview is in my opinion close enough to a conversation that the same logic could be applied.

I know that it might feel somewhat weird to use an ordered list, as it is so closely associated with preceding numbers (easily styled out). But when you come to think of it, ordered list makes sense and a sound option.

And yes, there are blockquotes involved. I would even use them in the context of an interview. Even as the quotation might be edited, it still is a quotation.

Andy's example goes like this:

-ol-

-li-
-cite- Speaker 1 -/cite-
-blockquote-
-p- Lorem ipsum -/p-
-/blockquote-
-/li

-li-
-cite- Speaker 2 -/cite-

...and so on.

22 Posted by Harmen Janssen on 24 July 2007 | Permalink

I have to agree with Philip Withnall in Comment #1. Blockquote seems closest to what you mean to mark up.
In addition, I would structure it using a definition list, creating a bit of HTML similar to this:
DL
DT/DT
DD
BLOCKQUOTE
/DD
[...]
/DL

23 Posted by Tino Zijdel on 24 July 2007 | Permalink

Marty Alchin: "As for citing, it's unfortunate that RFC2396 disallows fragment identifiers as part of a URI."

The href attribute of an anchor also needs to be an URI, so you're saying fragment identifiers aren't allowed there too? RFC2396 discusses fragments in chapter 4 where it is recognized that most usage of URI is within URI-references. I'd say in HTML actually URI doesn't denote an absoluteURI or a relativeURI but a URI-reference, so a fragment identifier (or even a same-document reference) would/should be perfectly acceptable for the cite-attribute (maybe the HTML5 spec should make this more clear).

24 Posted by Tanny O'Haley on 25 July 2007 | Permalink

I'm with Tino Zijdel here. I prefer to use a definition list for an interview. However I think that Borgar has a good idea in using headings for questions and paragraphs for answers. I tend to use blockquotes for text from an external source not an interview.

In the end I don't think there is a hard/fast rule.

25 Posted by Kit Grose on 25 July 2007 | Permalink

I'd have to say blockquotes are eminently appropriate here for responses, but seem thoroughly *inappropriate* for the questions they're answering.

I do question the semantic validity of using definition lists for questions. A good interview question would be itself already well-defined. An answer doesn't provide a definition.

My gut tells me blockquote is the solution here. I don't think paragraph tags will provide any semantic clue as to the content.

26 Posted by Brian LePore on 25 July 2007 | Permalink

Since I am of the opinion that a dialog is semantically not a series of definitions, I disagree with the thought of using a definition list for this purpose. That said, I can see where you are coming from by being weary of using blockquotes for quoting answers that have been edited, if even for grammatic purposes. So that raises the question, why not use blockquotes on the interviewers questions? Surely they will not have been edited and will be exact quotes, and it delimits it from the answers.

You would obviously have to override a browsers default styles for blockquotes, but there is no harm in a little CSS. :)

That said, this method isn't perfect because a blockquote is intended for long quotes. A question isn't typically long so that does seem odd to use it there. One could then argue that an answer may not be long so that would disqualify its use for certain answers but appropriate for others. You could argue that you should be able to get away with it because you want to stick to the same type of method for all answers.

27 Posted by Brian LePore on 25 July 2007 | Permalink

Alternatively, and I feel dirty writing this, you could use a table. I think one could reasonably argue that the table cells is tabular data. You could even use table headers with the scope property to identify who was speaking. I would envision a row with the names at the top with a column scope, and the interviewer's questions could also be a header with a row scoping.

Heck, you could even expand it to allow for multiple interviewers / interviewees that way. It doesn't make sense to me to try that with a definition list or a series of blockquotes. It might mess with the scoping or ordering of dialog, so I'd use it with caution.

28 Posted by Paul on 25 July 2007 | Permalink

What about -CITE- ?
Cite contains a citation or a reference to other sources. In this case the source would be the being interviewed.
-cite- Interviewer -/cite-
-p- question? -/p-
-cite- Person being interviewed -/cite-
-p- Answer -/p-

29 Posted by Paul on 25 July 2007 | Permalink

Or maybe better:
-p- -cite- interviewer -/cite- : Question -/p-
-p- -cite- person being interviewed -/cite- : Answer -/p-

This way the text referenced to the person who said it using -cite-, but by using p instead of blockqoute you don't reference it as a literal quotation.

30 Posted by Jake Archibald on 25 July 2007 | Permalink

A combination of blockquote, cite and paragraph elements are semantically closest to the data. I don't think there's a problem using an ordered list either.

But using a definition list? I really can't see a good reason for this. I mean, it's a definition list, how does that make it suitable for dialogue? Just because the structure is convenient (pairs of elements) doesn't make it right.

The term "John" is defined as "Doing anything tonight?"

The term "Jane" is defined as "No"

The term "John" is defined as "Want to go see Transformers?"

The term "Jane" is defined as "Yes"

Pfft.

31 Posted by P Boere on 25 July 2007 | Permalink

Standards developers like to use definition lists for everything that has nothing to do with defining data terms and data definitions.

Granted the paired structure is useful for organizing and layout, but then so are tables which the same standardistas claim are evil, because of being semantically irrelevant. duh??

Current xhtml specification has very limited semantic markup. So the correct way to markup, if in doubt, would be using the generic ul,p.

As for the particular question of using blockquotes. I would not as if the whole page is q and a then it doesn't seem to me as quotes because quotes are exceptions to the flow of a document. A document doesn't quote itself.

I would put each q/a pair in a li, and class two divs with semantic names, or nest an ordered list to create the pairing

32 Posted by Tino Zijdel on 25 July 2007 | Permalink

It really saddens me that a lot of people still think that semantics of an element are purely defined by the English meaning of a tag's name...

33 Posted by bruce on 25 July 2007 | Permalink

I wrote about this last year http://www.brucelawson.co.uk/index.php/2006/breaking-news-w3c-specs-are-not-word-of-god/

I agree with Tantek; a dialogue is for the names of speakers, and s for the text.

Removal of "errm" and tidying up grammatical mistakes that are common in speech doesn't stop the text being a quotation. It's a blockquote rather than a because it doesn't go "inline" .

I disagree with Tantek that it should be wrapped in an ordered list, however; all linguistic semantics in English depend on word order. For example, "The dog bit the man" is a completely different concept from "The man bit the dog", but no-one would mark up the sentence with each word being a list item in an ordered list, and I don't believe a dialogue needs it either.

34 Posted by Rob Kerr on 25 July 2007 | Permalink

Don't get hung up on the word 'definition' in the description of 'dl's semantics. IMO, 'dl' is absolutely
appropriate for the question/answer format, as it's the only HTML 4 tag that gives anything close to a
key/value pairing. The fact that multiple instances of 'dt' and 'dd' are possible is actually a benefit
in this case, as multiple questions may be asked, multiple responses may be given to any question, and
there is an inherent ordering within the tag. 'blockquote' is, for me, the intrusion of an external
document in to the current document (and 'quote', by extension, is an inline version of the same
phenomenon), which is, in the case of an interview, inappropriate.

35 Posted by Sander on 25 July 2007 | Permalink

As the ordering matters in a conversation/interview (as HTML 5 also remarks), <dl>/<dt>/<dd> are definitely not the right elements to use here (even though the way it _works_ is the 'closest' you'll get in HTML 4 to <dialog>, its _meaning_ isn't close enough).

<blockquote> would be okay, although I personally wouldn't use it either, exactly for the reasons you cite, chief amongst them that it's not a literal quote.

With no semantic elements currently available that accurately describe the content, you fall back to the most semantic element that _does_ cover the content, which is <p>.

36 Posted by Ben Gerrissen on 25 July 2007 | Permalink

Remember for whom semantic markup is meant for. A blind person might find it annoying and confusing to hear questions defined as definition terms and awnsers as definition descriptions. Especially if DT/DD's are used more frequently in a lay-out for other purposes.

It might be annoying to overlook pairing of data, but the goal justifies the means here, not the other way around.

imo:

[em]Do you like peanuts?[/em]
[blockquote]yes i like peanuts[/blockquote]

When reading that aloud with the semantic purposes of the tags in mind, it would sound appropriate and SOUND is what we're going for.

37 Posted by ben gerrissen on 25 July 2007 | Permalink

Actually... it would be nice if W3C pushed some "question" and "awnser" listtags through. The semantic demand for this is huge...

38 Posted by Chris Hester on 25 July 2007 | Permalink

I'd use a definition list. To me a blockquote is a QUOTE - ie: an excerpt from a longer text. But the answer shown might be the whole answer, eg: "Yes I do like bananas".

39 Posted by Ben Gerrissen on 25 July 2007 | Permalink

"To me a blockquote is a QUOTE - ie: an excerpt from a longer text."

Longer text like an "interview"? =P In fact, your text which I've put in between "quotes" can be marked up as a blockquote. Or I can put an entire post from someone inbetween blockquotes as an excerpt from a thread.

40 Posted by Joost Diepenmaat on 25 July 2007 | Permalink

I'm mostly with Ben Gerrissen here.

Since the specs do not provide for anything like a dialog (at the moment) we should probably not abuse semantically very different tags (i.e. dl) to implement them just because they happen to look somewhat like we want.

I'm not too sure about using blockquotes, though I feel they're semantically close enough to overlook that answers (and questions!) aren't strictly quotations.

Personally, I would probably prefer to use p's for both questions and answers, though possibly headers for the questions will work, provided they are short enough (i.e. no more than a single sentence).

41 Posted by Frank Boës on 26 July 2007 | Permalink

In my opinion a definition list (dl) is the best way. The relation between dt and dd is like a relation between a question and an answer.

And you will have now problem styling the different paragraphs via CSS. ;)

But as long as there are no real tags for an interview, maybe we should keep an eye on microformats ( http://microformats.org/ )? Even though there is no microformat for interviews (or any kind of conversation) at this point…

42 Posted by Sebastiaan Stok on 26 July 2007 | Permalink

Men reading al the reply's I cant seem to agree or disagree with anyone... DT looks good but the real meaning is not suitable for what you want to use it.

The only thing I can think of is:

em question
blockquote: answer

But could be better. The Web Guidelines. even forbid the use of the q tag so that is not an option.

I wonder hows got the final answer.

43 Posted by Martini on 26 July 2007 | Permalink

To me a blockquote would either be a direct quote attributed to the specific person, or a portion lifted from another source - verbatim. Meaning, this is exactly what they said, and you could take it to the bank (or court!)

Therefore, in the case of an interview, I don't think it's necessary - unless the interview is transcribed word for word exact.

44 Posted by Chris Hester on 26 July 2007 | Permalink

@Ben Gerrissen: I see where you're coming from, but I am doubting the validity of using blockquote to repeat the same amount of text - to me this is no longer a quote, but the actual source itself. Remember too that the interview is split up by answers, not continuous text.

To use blockquote surely some text must exist that is much longer than what is quoted. Dictionary.com define "quote" as "to repeat (a passage, phrase, etc.) from a book, speech, or the like, as by way of authority, illustration, etc.". Ie: not the whole text (which in this case is the answer we wish to mark up).

An example might be the first paragraph of this comment. To mark all of it up we should use the P tag. But to mark just part of it up we might choose the BLOCKQUOTE tag. At least that's how I see it.

45 Posted by thacker on 26 July 2007 | Permalink

Have always used dt/dd in conjunction with q. Whatever method is used in more in terms of design and how dialogue is presented to the Web visitor? As far as speech rendering, this is another good case of implementation of effective aural style sheet standards.

46 Posted by Ben Gerrissen on 27 July 2007 | Permalink

@Chris, you're perfectly correct and I concur. Though with the lack of "question" and "awnser" tags, I doubt there are any other tags that come closer symantically and contextual.

P tags could also work, but an interview is a dialog and not an article.

Every proposed tag concerning this dilemma will have an pro's and cons. To me the ruling factor is: "How will a visually impaired person hear the text".

47 Posted by Tino Zijdel on 27 July 2007 | Permalink

"How will a visually impaired person hear the text"

I'd say that depends on the assumptions made by the particular UA and/or the presence of an aural stylesheet and the UA's support for it, so I'm not so sure if that is a failsafe factor.

Althoug it is a good angle to look at semantics I wouldn't say it is necessarily the only or the best angle. Minority groups may have very specific needs that don't fully relate to other groups.

Imo this would in HTML4/XHTML1 still be the best option:

dl class="dialog"
dt - Interviewer
dd - Question
dt - Interviewee
dd - Answer

That can be enriched with maybe some other sensible classnames and styling (aural as well).

48 Posted by ppk on 27 July 2007 | Permalink

Right, it's clear that the answer to my question is anything but clear. The result of the vote:

- dl: 10 votes
- blockquote: 7 votes
- p: 6 votes
- author's decision: 2 votes

So there's no clear majority for any of the answers, although dl leads with 40% of the vote.

To me, Bruce's and Tantek's arguments against dls are very persuasive. I just can't see interview question and answers as any sort of definition. Therefore I'm not going to advise its use.

Unfortunately the Web Guidelines do not allow the use of the q element (incorrectly in my opinion, but my report must adhere to the Guidelines, so I have no choice here).

Since I still feel that blockquotes are improper because the interview answers aren't actually quotations, I'm going to advise the use of p elements. Feel free to disagree, by the way, but I have to come to some sort of conclusion today, and this is it.

Thanks for your feedback!