Sorry if I be too bold to start a project on my very first non-anonymous edit, but, I believe this project can get a substantial community interest.
I should be submitting some sample entries within next week. For now, let us see how it fares.--Luhmann br16:40, 16 August 2009 (UTC)[reply]
I think the headword "Picture Dictionary" is well chosen for the project page, uniquely identifying the concept. (Apart from that it should be capitalized "Picture dictionary" per Wiktionary conventions.) Even if it turns out that no picture dictionary will be developed on Wiktionary, the page documents all the activities around the possible project, including lack of activities.
I have been creating something like pages of a visual dictionary at Commons, using gallery pages. The results can be seen in Commons:Architecture, Commons:Drinkware, or Commons:Pointing device. Commons seems to work fine for the job, as long as I am creating an English visual dictionary. I have been using Wikisaurus and Wikipedia's category structure to estimate what broader headwords are suitable for collecting images under them. Image titles or captions can be used to describe images; a subdivision into subclasses of the class of the headword is possible using headings.
The page User:Luhmann_br/Pictures:en:Home_appliances has originally been created at "Pictures:en:Home appliances". The original title suggests a design scheme for the headwords of a visual dictionary: (a) it is located at "Picture:" namespace or pseudo-namespace; (b) it is multilingual, (c) the language is indicated using its ISO-code (":en"), and the rest of the headword is modeled on a topic category. An alternative is "Appendix:Pictures of home appliances", which reminds of "Appendix:Glossary of ...". --Dan Polansky18:23, 18 August 2009 (UTC)[reply]
I found a picture on Wikicommons that illustrates my ideal for the images for a visual dictionary. To me an important feature is an image that shows the relationship among components. I only wish we could add more lines and numbers to the image conveniently. I don't know how many of such images there are. DCDuringTALK16:23, 22 August 2009 (UTC)[reply]
I agree. But all these words are not directly related to the word eighteen-wheeler, they are more related to the thing. This is why I would put this image in Wikisaurus. On fr.wikt, we just created a page showing what could be the contents of a good Wikisaurus entry. I've added such a commented image, thanks to newly created templates : see fr:Thésaurus:renard (français). Lmaltier15:21, 23 August 2009 (UTC)[reply]
The other words are in semantic relationship with eighteen-wheeler. In this case they are hyponymsmeronyms. I expect that would be typical. We should have images for our terms anyway. Selecting images that give use more wikilink opportunities seems just a better realization of the opportunities that hyperlinks offer. Putting it in Wikisaurus would be putting it where it will be much less seen. When I hear thesaurus I do not expect images.
Adding images does run the risk that we would get more users who are not multilingual and may not even be very verbal. Next thing you know they'll be pointing out shortcomings, contributing, and thinking they have the right to vote.
It has occurred to me that a visual dictionary might be best as a shared resource among all Wiktionaries. The idea would be that one would click on the image and be taken to a page that had a large image with many more links connected to components in the picture. The language of the legend could depend on the language of the wiktionary from which you came. The user could select from any of the other languages offered and instantly see the legend in another language, best of all more than one at a time (at least two at once). DCDuringTALK16:18, 23 August 2009 (UTC)[reply]
To clarify: The words on the image are meronyms, as they denote parts of an eighteen-wheeler. The term "eighteen-wheeler" is a hyponym of "tractor-trailer", as each eighteen-wheeler is also a tractor-trailer. --Dan Polansky06:47, 24 August 2009 (UTC)[reply]
Yes, words mentioned would be hyponyms, typically. Wikisaurus is a very good place for hyponyms, hypernyms and the like, and this would save space on crowded mainspace pages. Another reason is that including these images in the main space might make pages look like an encyclopedic dictionary (while we are a language dictionary). I don't think that they would be seen less in Wikisaurus: if users understand that they can find here a very good thesaurus which is also a very good visual dictionary, it will be consulted much more (and this would be a very good thing, in its current state, users tend not to consult it). Another reason is that the same images might be shared by many pages, because they are not related to the word, but to the thing. It makes more sense to have them in a single place, the appropriate Wikisaurus page. Lmaltier17:02, 23 August 2009 (UTC)[reply]
Well. You made my case.
"In its current state users tend not to consult it" Why should any new project be condemned to join a moribund project in its grave?
If being related to a thing makes the content encyclopedic than we should dispense with definitions and limit ourselves to phonology, translations, etymology and grammar and dispense with any connection with the world outside of language arts altogether.
Wikisaurus is moribund because of its current contents. Most common English dictionaries are language dictionaries, not encyclopedic dictionaries, but they include a definition. Actually, the definition is the common part between a language dictionary and an encyclopedia. In French, the distinction between language dictionaries (e.g. Petit Robert) and encyclopedic dictionaries (e.g. Petit Larousse) is clear, and including such images is typical of encyclopedic dictionaries. Including them in the mainspace would blur this distinction. Lmaltier17:37, 23 August 2009 (UTC)[reply]
The distinction is mostly one of economics of printing and perhaps of on-line performance in the Anglophone world. Almost every English monolingual print dictionary has illustrations, even Longmans DCE, aimed at language learners. If you would like en.wikt's visual content removed, make a suggestion on WT:BP to lay the groundwork for a vote. DCDuringTALK09:09, 24 August 2009 (UTC)[reply]
No, not at all, I would like much more visual content. In many cases, pictures make the meaning very clear, without having to read the definition. But images used only for showing the meronyms would be best placed elsewhere. Lmaltier19:41, 24 August 2009 (UTC)[reply]
Why "best placed elsewhere"? I can provide rationales for best placed at entry. The rationale for placement in the entry is that it would be a plausible place for users to search for information about parts of what they know the name for. It is at the very least on click closer that Wikisaurus. In addition we would have to train users to expect visual information from a thesaurus, which is against their prior experience. DCDuringTALK20:23, 24 August 2009 (UTC)[reply]
I already explained my reasons. About you last remark, you are right. Combining a thesaurus and a visual dictionary would be very original. But it makes sense, if you think at it. We might look for a name combining visual dictionary and thesaurus (it's not easy...). Lmaltier20:41, 24 August 2009 (UTC)[reply]
I don't know which of the refuted arguments you are still sticking to.
As to your first statement that the meronyms are not in semantic relationship to eighteen-wheeler. That would seem to be in direct contradiction to meaning of meronym and WT:ELE.
Crowded namespace entries. Some namespace entries are long. But they are long with or without pictures. Many of the longest entries are for verbs which will almost certainly elude being captured by still images. I am not averse to images being in thumb size on long entries, with a legend inviting the user to click.
Language dictionary vs. encyclopedic dictionary. The sharpness of this distinction is something backward Anglophone users are not familiar with. Nor do I grasp how an image legend with wikilinks to meronyms is different from a meronyms header with such links.
Wikisuarus needs help. That may be. But, if images with legends on the headword page give users more intutive and more direct access to the information they seek, why does that matter to the proposal at hand?
Originality of combining thesaurus and visual dictionary. I love originality, but not when so much of the originality will be required of users in guessing what lies behind the wikisaurus link, especially when all available information suggests that users are unlikely to make any click that does not give obvious benefit.
Too many users tend to confuse Wiktionary and Wikipedia. Everything likely to increase this confusion should be avoided. You are right, I would transfer all hyponyms, hypernyms, meronyms... (and most synonyms) to Wikisaurus. Wikisaurus already includes them in many cases. What do you think about fr:Thésaurus:renard (français). Don't you think this kind of organization could be successful? If something is really very useful, it becomes popular sooner or later. I would like to know what other editors think. Lmaltier06:17, 25 August 2009 (UTC)[reply]
It is a very interesting idea. I would love to have that kind of information available in English for whatever I was interested in. I wonder what project is the right home for it. It actually would seem that the conceptual work of creating an entry structure would best be shared among many languages, though I wonder how the work could be coordinated among many monolingual contributors. Thesaurus doesn't seem the right label. I also wonder about how much of the context is specifically about a fox vs. a canid, a quadruped, a mammal, a living thing. That makes me wonder about how inheritance might work in a wiki framework (both the software and the relationship to contributors). That brings me to technical and sociological issues beyond my pay grade. It seems like a large project. I may not live that long.
Given the dependence of wiki projects on volunteer labor, a lot of contributors have to buy off on a given concept for it to work. And it needs to fit user expectations. To me that means that major innovations that don't correspond with user expectations are not likely to go far. It seems difficult to simply get pictures from commons into our entries. My ambitions are more limited. At the moment, I am looking at What's What: A Visual Glossary of the Physical World (1981). It has about 500 pages of pictures and tables with a table of contents and an index. I would be thrilled if we could find 1-2,000 such images over the next few years. DCDuringTALK10:51, 25 August 2009 (UTC)[reply]
(Unindent) I don't see any problem with the diagram placed in eighteen-wheeler. Meronymy is a valid part of mainspace, as of current WT:ELE. A diagram documenting meronymy is no worse than an unlabeled photo of an eighteen-wheeler.
Also, I don't see any problem with the dictionary-encyclopedia distinction; semantic relations including hyponymy and meronymy are part of Wiktionary; they are relations between senses (semantic objects), which are no less important than words (syntactic objects). The main reason why printed dictionaries avoid pictures is print space. Even within space constraints, extensive dictionaries contain pictures. For instace, see “mango”, in The Century Dictionary[…], New York, N.Y.: The Century Co., 1911, →OCLC. and its illustration of various parts of mango tree including flower, pistil, fruit and seed. --Dan Polansky07:51, 25 August 2009 (UTC)[reply]
Is there a statistics on how much people consult Wikisaurus? I do not know of one and have for some time wanted to know the numbers. --Dan Polansky06:55, 24 August 2009 (UTC)[reply]
I'd love more statistics of any kind. (I already have linked to the ones on Dan's page, thanks.) It seems extra difficult to do it by type of page. DCDuringTALK09:37, 24 August 2009 (UTC)[reply]
I don't think that we are yet at a point where we know what to do, let alone how. Perhaps we could start by trying out some different approaches using materials already at WikiCommons. Or someone could try to produce some drawings. I would suggest few constraints on the process until we have several different kinds of examples or one approach that seems so good as to make further experimentation unnecessary.
We may find lots of different ideas about this and even objections to the not-so-linguistic nature of this kind of effort, relegating it to somewhere off the beaten path. DCDuringTALK16:29, 23 August 2009 (UTC)[reply]
From the above discussion, it's clear that there is a disagreement about what thesaurus means. And I must say that I disagree with the current Wiktionary definition (in thesaurus): a thesaurus does not include only synonyms and antonyms, but also all words strongly related to the title of the thesaurus entry, because of their meaning, or for cultural reasons, all words that somebody writing about the subject might be likely to look for. This includes hyponyms, meronyms, etc and many other unnamed -nyms. A few examples (taken from Roget's thesaurus) : at train, you can find wagon-lit or rails. At refrigerator, you can find ice-cubes. At pungency, you can find piquancy, etc. but also tobacco, or spicy. At dressing, you can find millinery or tailoring. And it's the general rule in the whole book. This is exactly what we tried to do in fr:Thésaurus:renard (français).
If you consider a thesaurus this way, I think that, clearly, the visual dictionary project could be integrated to Wikisaurus (but I also want many more pictures in the main space). Lmaltier17:17, 25 August 2009 (UTC)[reply]
It could be that Wikisaurus needs a fresh look. There is no reason for a good picture not to be in more than one place. One thing that Wikisaurus certainly could use is more contributors. DCDuringTALK17:52, 25 August 2009 (UTC)[reply]
I have started working on diagrams of constituent parts, you can see the preliminary results here. Those images are in SVG with embedded pictures it took me just a few minutes to make them. They can be translated either in vector graphics editor (such as Inkscape), or even in a text editor. SVG can use hyperlinks, which I should probably use on next versions --Luhmann br00:39, 6 September 2009 (UTC)[reply]
Looks very promising. But we need to make some decisions. Images aimed at our principal namespace have not been large:
The way of incorporating text seems to make that text not separately zoomable in HTML so that the text and picture size decision is not entirely user adjustable, which means that we have not have more information on the user than we have at the moment or risk missing those with vision that does not meet some standard. An example of such a user is me. The text size means that I cannot read any of the text. The poor contrast means I can't tell how many text captions there are. DCDuringTALK01:38, 6 September 2009 (UTC)[reply]
Parts of the body labeled on a Botticelli picture
Features of human body
A truck, labeled with numbers instead of with descriptions; see eighteen-wheeler
Feather numbered
Labeled in bubbles or rectangles with round corners and white background
Bicycle
I am posting another example of an image of a body with labeled parts, to a gallery above that can be expanded in the future. Unfortunately, the labels are unreadable when the image is scaled to the thumb size. For the purpose of putting to a thumb in Wiktionary, portrait form images seem to work better than landscape ones, still depending on the font size of the labels in the image. To the same gallery, I am moving the parts of the body on a Botticelli picture, and adding other images linked to the the article page of this discussion. --Dan Polansky16:43, 30 November 2009 (UTC)[reply]
I'm thinking about starting to put together a multilingual picture dictionary mostly for tools, and maybe also bike parts. So it would be detailed, illustrated, but the focus would be to have also the picture(s) and different languages of the same thing. I'm not sure if all on one page and where to host it. I'm also planning to make a poster out of it (picture and the different words under it in many languages, like 3-4 maybe?) It would be useful for many of us who work together with people on D.I.Y. projects (as well as industrial and craft workers) who don't have the same mother tongue. and other thematic "pictionaries" can be created in a similar way. Do you think Wiktionary would be a good place for this? how can we connect the different languages and the pictures at the same time? this picture dictionary idea was only for one language as I understood. Should it be hosted somewhere else? any ideas? (I think for us it's important to have it in a noncommercial site - non-profit, open source, copyleft etc)--Ovis.bp20:28, 15 December 2009 (UTC)[reply]
Wiktionary would be the perfect place for you, though we can be a little tough to understand right at the start. The way things work here is simple, the English page bicycle has a picture of a bicycle and links to all the translations of bicycle under the Translations boxes. Then all the translations have their own pages (which you might not need even) that contain all the pronunciation and information (again including pictures) that you need, e.g. vélo (though no image there yet). Let me (or anyone else) know if you need more help. Conrad.Irwin20:57, 15 December 2009 (UTC)[reply]
Multi-lingual picture dictionaries are an excellent idea. They allow new vocabulary to be found very quickly and I'm sure will prove useful to many. I agree with Conrad's suggestion about implementing each language on a separate page with links in the Translations box. Pgr9408:37, 25 July 2010 (UTC)[reply]
I forgot that I had a page about the solar system laying around (now at Wiktionary:Picture dictionary/en:Solar System). It reminded me that we should be using mw:Extension:ImageMap so that the use can click on certain parts of the picture rather than having to go through the legend. Hopefully we get more of these.
Quote: "4 - For more abstract entries (such as verbs), we should probably draw a sketch specifically created for this project, uploading it to Commons."
Hello HenkvD, you've done some great work with those templates and as a result I'd say picture dictionaries are closer to realisation. Well done!
Out of the four alternatives, I prefer "Picture dictionary with legend and wikilinks": it is very clear and understandable. The templates become too complicated as soon as there are clickable regions on the image, so I'd suggest avoiding this direction, at least for now. Does anyone else agree/disagree?
Some small suggestions:
for simplicity, the template parameter "detail2" could be renamed "legend"
the box title "Legend" could be made part of the template (avoids duplication and ensures consistency)
it's not clear to me why there is a second image with a heavenly body. The entry is for "solar system", right? I'd suggest a separate picture dictionary entry for "heavenly body" with an image showing a star, a planet, a moon etc. Then we can perhaps drop parameter "detail1"?
Not every entry is suitable for (numbered) legends. Therefor additional clickable images can be used. Labels is my personal preference. Personally I don't like Imagemaps, as you don't have a clue which parts are clickable or not.
Regarding parameter "legend": this could be useful, but I want the result to be language independend.
The image for heavely body is a link to a hypernym (a broader term, like Earth to planet to Solar System). As stated I am not not 100% pleased with the way to add these.
Hm, there seems to be a technical problem with the "label" kind: the labels have large empty spaces on both sides that can overlap other labels at the same height, making them unclickable. Is this fixable? Also, the template uses "anchor=" for linking to language sections, using the full language name in the parameter, rather than the standard Wiktionary method of having |lang= followed by the ISO code of the language. I think it would make sense to replace the {{{anchor|#English}}} in {{picdiclabel}} with #{{language|{{{lang|en}}}}}, as is standard. --Yair rand (talk) 02:08, 28 July 2010 (UTC)[reply]
I tested 2 texts on the same line (see the example at the right)
I can click on both links, and each links to the correct page. Does this not work for you? What browser are you using? What skin are you using?
The size of the labels is 300px. The text is aligned in this box. If the size is too small the text will not be aligned properly. Does anybody know how to calculate the text length in pixels?
As for the #{{language|{{{lang|en}}}}}: a very good idea. I will implement that later.
The link "one" is unclickable for me except at the top and bottom rows of pixels. I'm using Chrome 5, and monobook and vector have the same result. --Yair rand (talk) 19:00, 28 July 2010 (UTC)[reply]
Links three and four should react the same as one and two. I added background colors yellow and white for better recongition of the areas. Feel free to test with link five and six. For me it still works if I change height:12px to height:1px.
The problem is most likely to occur in the align=center mode.
Height 1px works still leaves a 1px line covering other links. Would there be any problems with just changing the default height to 0px? --Yair rand (talk) 19:05, 29 July 2010 (UTC)[reply]
The Web Colors are not the same as everyday colors. See w:Violet (color) for some explanation. For example, the web color is not the same as the spectral color used when discussing the visible spectrum or when discussing the color wheel in art. Nor does it match the shade normally used by crayon manufacturers. --EncycloPetey23:39, 31 July 2010 (UTC)[reply]
I just saw here that the ImageAnnotator gadget has been installed on Wikipedia allowing people to annotate sections inside a picture. You can see the effect by turning on the gadget via the wiki's Special:Preferences page and then looking at an annotated image (eg w:File:Solvay conference 1927.jpg). Does this look like something we'd want here? We could presumably bring over the gadget as it should just be JS. --Bequw→τ02:09, 10 August 2010 (UTC)[reply]
I don't know the relationship this has to what User:HenkvD has been doing. I would expect or, at least, hope that something being tested/used on WP would end up being more reliable and possibly better. Only if our distinct needs can't be met using what WP has would there be any point in using something else. At the very least we should have the gadget installed for now to compare. Once the annotation is complete, does anything have to remain installed server-side or elsewhere for users to benefit? DCDuringTALK14:07, 10 August 2010 (UTC)[reply]
What you've done is much more in line with what we need. What other wikis use your approach? It seems much more desirable for all wikis with links. Am I correct in assuming it needs handwork to lay out the link rectangles? Is there a prospect for making that process graphical? Do we have a single depository of the rectangle layouts and the link to the image on Commons? Even if some languages don't parse a physical object the same way in every detail, most of the work would carry over, I think. The commons format seems less likely to allow a clumsy contributor (eg, me) to mess up the table. DCDuringTALK18:38, 10 August 2010 (UTC)\[reply]
What advantage would a separate image namespace be for this project? Would it allow you to simplify code and/or make it a little easier for users to add zones and links/labels without messing things up? DCDuringTALK18:49, 10 August 2010 (UTC)[reply]
I've just used your templates to create a picture of an eye with labels linked to the appropriate pages (not all already created). I quite like the idea of picture dictionaries, and I hope it grows and bears fruit; but I notice nobody has been giving this attention since 2010... Anyway, please add the eye picture from acs to your list of pictures! And again thanks for the effort! --Pereru (talk) 19:21, 25 June 2012 (UTC)[reply]
I would think each picture would be it's own Appendix subpage, with see also links from the entries. Perhaps templates (with ifexist) could be used to populate a column with terms from a language chosen by the user. DCDuringTALK13:28, 20 February 2013 (UTC)[reply]
Yes, subpages could make sense. I took these from a Swedish-German dictionary, so I had the German and Swedish columns for free, and kept them. But perhaps there should only be picture and English words, and then the German Wiktionary could have the same appendix with German words, and so on, all interwiki-linked? I'd like to find a solution that scales to a larger system than just one page in three languages. --LA2 (talk) 17:09, 20 February 2013 (UTC)[reply]
If you go here, a Flash animation will start that offers a picture dictionary. In the selector "Välj språk", pick the first language (Albanian), then switch to English in the upper right corner. As an example, pick No. 27 "Farm and agriculture", then "tool 2" for a picture of hand tools. Click on one of them, to hear their name. As this is a Flash animation, there is way to link to that page. However, is the tool on the far right really a "skewer"? --LA2 (talk) 17:44, 20 February 2013 (UTC)[reply]
It might be a skewer, though I don't think of such a thing as being a farm tool, rather than a cooking tool. My mindset might be because of some cultural difference, relating to the prevalence of specialized outdoor grills in the US, rather than open-pit cooking, say, of a pig. My skewers are more dainty implements, for, say, kebab, rather than something that could support the weight of a pig. DCDuringTALK00:48, 21 February 2013 (UTC)[reply]
It is an obvious mistranslation of the Swedish spett, which is cognate with spit, and is used both for a cooking tool (large spit, small skewer) and for the crowbar in this illustration. I love it that I stumbled on a mistranslation, because it gives me a chance to study how these folks handle error reports. :-) --LA2 (talk) 23:51, 21 February 2013 (UTC)[reply]
I'd forgotten the word spit, which is clearly what was intended or should have been intended. I think that it is an example of something not well illustrated by an image of the thing itself rather than the thing-in-use. DCDuringTALK00:16, 22 February 2013 (UTC)[reply]
I split the page into ten subpages, with a gallery as the start. I think this was a huge improvement. I categorized each subpage, and the top gallery leaves room for many more such entries. --LA2 (talk) 00:55, 22 February 2013 (UTC)[reply]
Multilingual Picture dictionary on Wikidata?[edit]
I have been experimenting a bit, but I think this ENGLISH wikitionary is not mutilingual, in the sense that it does not use the Universal Language Selector. HenkvD (talk) 19:49, 14 November 2017 (UTC)[reply]
Could this not be overridden with another parameter? So you could embed the picture in any language section, and show the appropriate labels, depending on the section it is located in (and not depend on the user settings). The data would always come from Wikidata, but we could automatically link to local pages. So for French, d:Q405 would link to lune. This assumes that there's a corresponding entry for the label, of course. – Jberkel (talk) 22:37, 14 November 2017 (UTC)[reply]
@Jberkel I don't understand what you want to achieve. My examples are for Solar System. On that page it should link to Earth, Moon etcetera. For système solaire it should link to Terre, lune etcetera. I don't see a possibility to link the language to the section (French). A parameter for language could maybe be programmed in LUA, but Wikidata should and will use the Universal Language Selector. I won't waste time on that programming. HenkvD (talk) 19:53, 17 November 2017 (UTC)[reply]
@HenkvD The idea was to embed the picture dictionary inside a Wiktionary language section, with Wikidata pre-filling the labels. Don't worry about programming anything, I can take your work as a starting point and see if it's possible to adapt it for what I have in mind. – Jberkel (talk) 09:45, 18 November 2017 (UTC)[reply]
Parts of the body labeled on a Botticelli picture
Features of human body
A truck, labeled with numbers instead of with descriptions; see eighteen-wheeler
Feather numbered
Labeled in bubbles or rectangles with round corners and white background
Bicycle
