Why are descriptive terms sometimes missing in object names?

It’s weird that road classification in the bridge. I hope that both communities try to focus not only in a “local OSM”, but also and simultaneously in a “global OSM” and fix that in a coherent road network.

People refer to it that way when the feature is globally recognized. There is also in some ways the reverse behavior, referring to a feature only by its type (the context of the communication between the parties involved makes it unnecessary to make the name explicit: e.g. “meet me at the school”)

But just as the use of abbreviations is discouraged in OSM names, naming things based on natural language (which takes advantage of Economia lingusitica; a form of abbreviation) can result in confusion.

1 Like

This analogy is useful but doesn’t tell the full story. When it comes to geography, CJKV has a much more ingrained habit of affixing a descriptive word than many Indo-European languages. In English, you can informally refer to the River Thames as just “the Thames” or indeed any river as “the [Base Name] River”, as long as your audience can tolerate some ambiguity. Similarly, you can usually clip the name of a street without sounding stilted: “Take the M3 motorway to the A303 road and come off at Solstice Park Avenue.”

By contrast, in CJK, the notion of a “base name” is less useful, at least in prose or speech. Dropping the analogous words or characters from a Chinese sentence would be not only informal but probably also ungrammatical (correct me if I’m wrong). If you drop the 州 from 肯德基, it refers to Kentucky Fried Chicken rather than the State of Kentucky. The components of an address are smooshed right up next to each other, as in “福建福州连江丹阳新洋团结10”, so if you drop any of those type words, a human may be unable to comprehend the rest of the address or the phrase it’s in.

Even so, these affixes don’t necessarily make their way onto maps and signs, which must economize on space. Across many cultures, maps have traditionally omitted “boring” words from name labels. A map may omit “road” or “street”, “railroad”, “province”, “river”, and so on – and so might a sign, without affecting how people refer to the feature in speech.

Vietnamese is not quite as strict as CJK. Due to a different grammar and writing system, the tỉnh in Tỉnh Quảng Ngãi is optional in more contexts, but when you want to refer to the commune of Bình Mỹ in the district of Bình Sơn in the province of Quảng Ngãi, as opposed to some other Bình Mỹ, you’d still tend to say “ Bình Mỹ, huyện Bình Sơn, tỉnh Quảng Ngãi”. So that information does need to be stored somewhere, if not in name=* then perhaps in border_type=*. We could remove Tỉnh from the names of these administrative boundary relations, but Thành phố Hồ Chí Minh can’t be shortened to just Hồ Chí Minh without becoming very awkward.

Similarly, mappers with Kaart and HOT have disagreed about whether to include or omit đường in street names, since only boulevards and expressways would begin with a different word. Đường always appears on the sign, but it’s more like a classifier or title than a part of the name proper. Vietnamese often likes to introduce a proper noun or loanword by some more general word, so “pizza” gets translated as bánh pizza (literally “pizza cake”) and “OSM” might become trang OSM (“the OSM site”). The boring words in these features’ names are basically the same thing.

In English-speaking regions, we consistently include boring words in name=* tags, especially for roads and streets. This is partly because the boring words aren’t especially predictable: the U.S. has hundreds of “common” street name suffixes, and other regions have their own. A tagging scheme for these “types” would be tantamount to splitting name=* into more structured fields. Long ago, one mapper in my area experimented with shortest_name=* for the base name, but unfortunately this idea never took off.

10 Likes

Thank you. I am impressed by your insight.
Although I do not have the ability to verify your logic, many parts have become understandable and clear.
As I participate in the global work of editing OSM, I often think about cultural differences and cultural perception differences, and this work has made me more interested in the diversity of cultural perception.
This work has made me realize that global work such as editing OSM is a work that helps to understand concepts according to cultural perception beyond simply distinguishing by country.
Thank you again for your clear logic.

4 Likes

In addition, sometimes a street in an English-speaking country may be missing a boring word entirely. There is no default boring word that gets appended to these street names—they just are what they are. If I were to add my own boring word to such a street, there’s a chance my mail might not get delivered to the right place.

For example, Sciennes is the full name of a street in Edinburgh, Scotland. There is no Sciennes Street. Be careful not to confuse the address 7 Sciennes with nearby 7 Sciennes Road, 7 Sciennes Gardens or 7 Sciennes House Place.

5 Likes

For the record, and for archival purposes, I’ll add links to some related discussions.
Please note that the articles below are primarily for South Korea, so they may not apply to other regions.

I know I’m late to the party, but since I’m Norwegian I’ll just share my take on this.

In everyday dialogue it is almost always clear from context which Otta is meant. If you enter a train station and say “give me a ticket to Otta”, it is clear that you want a ticket to Otta station. If you say “I’ll be eating pizza in Otta”, it is clear that you’re talking about the town. If you say “I intend to swim across the Otta”, it has to be the river.

Some times the context is not enough, in which cases we add the necessary descriptors: “Otta river”, “Otta town”, “Otta station” in order to clarify intent. You might think that a map is such a situation, since the names all appear together. But in reality, the difference in visual appearance is more than enough to remove any ambiguity. A train symbol with the name “Otta” underneath is clearly referring to the station, etc. Perhaps the descriptors would be useful in a very simple black-and-white map.

Adding descriptors is also what we would do when searching in a map for these places. If you type in “Otta” and you get a list with too many options. The
instinct of a Norwegian speaker is to then clarify with descriptors. The expectation is that the search engine would find “Otta” the train station based on searches like “Otta station” and “Otta train station”, even if it is only called “Otta”.

I guess the point is that we don’t ever have to actually distinguish between “Otta”, “Otta” and “Otta”, because we clarify if cotext is not enough.

7 Likes

I’m not sure if Nominatim does that.

1 Like

“Otta train station” doesn’t work, but “Otta railway station” does (or just “Otta station”)

3 Likes

This is more a weakness of Nominatim than a problem with the classification system per se IMO.

2 Likes

I’m grateful for the many and varied comments from many of you.
At this point, I would like to clarify the point of discussion a bit more.
The essence of my question is, what is a good reason for not labeling points with descriptive terms in OSM?
Not whether it is searchable or not, or whether it is inconvenient, but whether it is common in everyday conversation, and if not, what are the reasons for doing so only on maps.
Let me clarify this point and add some thoughts.

The general consensus seems to be that it’s a matter of local custom whether or not to add ‘-station’ or ‘-river’ to station or river names.
And from the various comments I’ve heard from many people, it makes sense why points of interest often don’t have descriptive terms.
However, I don’t know what it’s like in real life, but it’s still not clear to me why we omit “-station” or “-river” in Google Maps or OpenStreetMap.
I can understand that in some languages, in some cultures, it may be perfectly acceptable to leave out descriptive terms in real life conversations. In other cultures, it’s not uncommon to not use descriptive terms, at least in casual conversation.
But when it comes to formal notation, it’s a different matter altogether, and in some cultures, it can be a bit confusing.
In the case of Google Maps, it’s not as common because it’s so searchable, but in cultures where it’s common to label points with descriptive terms, it often leads to confusing search results.
(This is especially confusing and frustrating in OSM, which doesn’t have the best search capabilities.)
This is why Google Maps often has photos or reviews of other places in the wrong places.
I’ll have to ask Google about their corporate philosophy, but I’m curious as to why OSM follows this philosophy.
Maybe it’s because it’s such a common practice in the English-speaking world that we just naturally follow it?
If so, are station names and river names the only things in the English-speaking world that don’t have descriptive terms?
(I’ve never seen a map anywhere else that shows “OO Police Station”, “OO Fire Station”, or “OO Elementary School” as “OO”. Does this happen in the English-speaking world?)
And in some cultures, it’s fine to write “Seoul Station” as “Seoul” if it’s mutually intelligible, but if you say “Seoul” to a Korean, at least, everyone will think “Seoul City,” which means it’s not at all common, universal, or reasonable.
Some people say that it’s because it’s distinguishable by tags, but if that’s the case, then we shouldn’t use descriptive terms for other place names to be consistent.
However, as in the example above, the “Charleville-Mézières station” in France is clearly labeled “Gare de Charleville-Mézières” on the signboard outside the station, but only “Charleville-Mézières” on OpenStreetMap and Google Maps.
This leads to a discrepancy where locals in France officially call it ‘Gare de Charleville-Mézières’, but Google Maps and OpenStreetMap call it ‘Charleville-Mézières’.
Again, I’m not saying that it’s distinguishable or not, I’m saying why it’s written that way, and that it’s uncomfortable and confusing for people from some cultures.

Can anyone clarify this for me?

P.s. I apologize for my poor English and the length of this post. I did some back-translation and it seems that the self-translation feature has quite a few translations that are a bit different from my meaning. If there are any sentences that don’t make sense, please use DeepL or Google Translate as much as possible.
1 Like

I don’t think I can, because you summed it up quite nicely. The only thing I can add to this is that this inconsistency is very, very common, even among the operators of these establishments. So over here, in Germany, if you want to ask someone for directions to a hotel or restaurant named “X”, you would certainly add the type to your question: “excuse me please, where can I find the hotel X?”. But this hotel would probably only have “X” written on their building, but “hotel X” inside in the lobby. On the invoice, they would then write “hotel X”, but the restaurant inside the hotel is then “restaurant in the X”, and so on and so forth.

The only thing that’s consistent about the naming is its inconsistency. If it’s common in your country to always add “hotel” to every hotel, then do it for the objects in your country. Over here, it’s not that easy :wink:

5 Likes

I suppose one thing influencing the general approach is that if you have descriptive terms in the name, it’s difficult to remove them programmatically if you want to display the “bare” name". Whereas if you start from the bare name. you can add labels (or icons) based on the other tags.

In the screenshot, the town and railway station of Pizarra are distinguished by the labelling in the search results, and by the blue square representing a station in the map. Do you feel this is confusing, or is the confusion in other contexts?

Also note that in this case, the river is labelled as River (Río). This seems to be standard practice for mapping in Spain. (At least in Spanish-speaking parts - from a quick glance it seems this might be different in Basque areas, but I haven’t looked at this in detail).

I suppose one thing influencing the general approach is that if you have descriptive terms in the name, it’s difficult to remove them programmatically if you want to display the “bare” name". Whereas if you start from the bare name. you can add labels (or icons) based on the other tags.

generally we solve it by adding variants, name is for the most common name, but there is short_name, official_name and a whole plethora of more name keys to catch subtleties (at least this is the hope). Names can have descriptive terms in them and it is fine to add them as well to the name tag, “station” is not uncommon as part of a name, as is hotel or restaurant, ultimately it has to be decided individually, that’s the nature of names.
E.g. https://www.openstreetmap.org/relation/6036811

2 Likes

Sorry if this a nitpick, but I think it’s an important one: in OSM we generally don’t “label points”, instead we include the names that are used in the real world.

In case of OpenStreetMap it’s because we use the names that are used in the real world by people who live nearby.

No, definitely not.

For example, a sandwich fast-food restaurant called “Mr. Sub” is named “Mr. Sub” in OSM, not “Mr. Sub Sandwich Restaurant”.

A while back there was a major edit to have all Starbucks shops named “Starbucks” in OSM even though their sign outside might read “Starbucks Coffee”, and it’s definitely not something like “Starbucks Coffee, Pastry, Sandwich, and Juice Shop” even though that’s a valid description.

If you’re on the train, do the announcements say that the train is going to “Seoul” or to “Seoul Station”? If you go to buy a train ticket, do you ask for a train ticket to “Seoul” or to “Seoul Station”?

If it’s the latter, then perhaps that station should be called “Seoul Station” in OSM. But in many languages and locales, that’s not the case - in Polish I’d ask for a ticket to Gdynia or maybe “Gdynia Główna” to distinguish from other stations in Gdynia, not to “dworzec Gdynia”. But if I’m in the city, I’d say I’m going to the train station (“dworzec”), usually without naming it at all, or specify which station with “dworzec główny”.

Nothing describing the real-world can ever be fully consistent, because the real world is not fully consistent. The only debate is how much inconsistency we’re comfortable with, and how to attempt to resolve it.

2 Likes

I believe that it’s more complex than this. I happen to be from a city not far from Charleville-Mézières and nobody there says that they go to “gare de Charleville-Mézières” or to “Charleville-Mézières”. They go to “gare de Charleville” or to “Charleville”.

My point here is that there are many groups and many situations: local people in a train, tourists in a train, local going to the station, etc. Each of these groups may predominately use a different name from the others, and we need to find the name that will cause the least surprise and frustration to the most significant group when they use a map or an app.

On top of that, the name of the city itself can depend on context. Most style guides for written prose such as encyclopedia and newswire articles would expect “New York City”, “Washington, D.C.”, and “Quebec City”. But high-quality map publishers would label “New York”, “Washington”, and “Québec”, respectively; the rest is obvious from spatial context or typography. Thus, an article in National Geographic may refer to “Washington, D.C.”, while the locator map that accompanies the article labels the same city as just “Washington”.

For data consumers that get place names from Wikidata, I’ve proposed an approach that would allow them to get map-optimized names instead of prose-optimized names. However, there aren’t enough examples of this kind of tagging in Wikidata to justify developing support for it in renderers. Maybe a bulk edit of CJK items could provide the necessary incentive. But I think we need more examples of conventional cartography from that part of the world to know what the expectations really are.

1 Like

People like to bash the German language for being overly complicated, especially because of our love for compound nouns[1]. But it can also be handy if the type of POI is part of the noun itself, because we cannot and don’t want to separate it. “Gare du Nord” is Nordbahnhof, we don’t change that to “Nord” only, but the “Bahnhof Berlin Zoologischer Garten” gets its “Bahnhof” (train station) removed. A lot of cities end in “stadt” or “stätt” (Eichstädt, Neustadt, Nordstadt), these get to keep 'em, but if Berlin was called “Berlin City”[2], then we’d remove the “city” . So if a sign says “Restaurant Dionysos”, then we can safely strip “Restaurant” off, but not from the “Aussichtsrestaurant Adelheid”. So much for theory, but of course there’s a lot of gray zone. Happy greetings from a big fan of compound nouns :smiley_cat:


  1. “X of Y of Z” becomes the single word ZYX in German ↩︎

  2. No, it is not, just trying to illustrate something here ↩︎

3 Likes

It’s definitely not just in the Anglosphere. Maybe I am simply replacing an Anglocentric lens with a Eurocentric one here.
In any case, in Greek, one does not use descriptive names for rivers (Πηνειός), mountains (Όλυμπος), islands (Νάξος), or most geographical features. By contrast, the names of mountains and sometimes islands can included in Anglo contexts (Mount Everest, or the Isle of Mann come to mind).
Train stations may or may not have the descriptive part in their proper name in Greece, usually on the condition that they have to be differentiated from another similarly named one.
Roadways rarely have their classification of “Motorway” or “National Road” or “Street” mentioned in their name either in maps or in everyday speech (with a few exceptions). Contrast American usage where one always speaks of Wall Street, never simply Wall, or the Long Island Expressway.

In any case, most of this in fact contradicts English speaking conventions as you can see. Each community has quite unique ways of mapping that are particular to local cultures, and OSM aims to reflect that as best it can.

1 Like

locally, almost everything is removed and it becomes just “Zoo” or “Bahnhof Zoo”.