It's Pronounced "Reading"
language linguistics maps OpenStreetMap · 8 comments · 400 words · read ~556 times.
English is a funny old language. That my mother tongue doesn't bother with internal consistency doesn't bother me much - except when it comes to Text-To-Speech.
Using Google Maps to provide route guidance in the UK is a challenging affair. Driving through Reading, the computerised voice continually mispronounced is as "Reading".
Err... that is to say, it should have said "ˈrɛdɪŋ" instead it said "ˈriːdɪŋ" - that is, "red-ing" rather than "reed-ing".
Ok, hetronyms are a notoriously difficult to get right - even for humans. Without context, it's hard to know which pronunciation should be used.
But, I've heard my sat-nav pronounce "Woking" as "Wokk-ing" rather than the correct "Woe-king". Vexing to those living there, distracting for those relying on accurate directions.
A few weeks ago, my wife and I learned that a friend of ours was getting married in Canandaigua, New York. While not as linguistically complex as, say, Poughkeepsie, it caused some consternation in our household. How can you travel to a location you can't even pronounce?
That's when we stumbled across the curiously named "HowJSay.com" - it supplies audio samples of people speaking particular words. So now I know how Canandaigua is pronounced.
I wonder if there is any sensible way to crowd-source pronunciation for a mapping project like OpenStreetMap. Sadly, Google Maps doesn't have a way to contribute language changes and the less said about Apple Maps the better!
There have been some discussion on OpenStreetMap mailing lists (in 2008) and there is a proposed phonetics feature (from 2010).
This is not a minor problem, Wikipedia lists hundreds of UK place names with counter-intuitive pronunciation.
There's no direct harm in a sat-nav mispronouncing a town or street name - although it can be very annoying for anyone expecting the correct vocalisation.
So, should modern maps allow for the correct pronunciation of place names? I think they should. That then draws us to an interesting question about regional pronunciation. An Englishman, Scotsman, and American all walk in to a bar - they each pronounce the word "Edinburgh" very differently. Who is right?
I leave you with one of Richard Lederer's hetronymic poems to enjoy.
The crowd-sourcing of pronunciations is already underway; unsurprisingly, it's on Wikipedia. You an see it, for example, in the first line of the article on Reading: http://en.wikipedia.org/wiki/Reading,_Berkshire - anyone can edit Wikipedia, and I encourage interested parties to contribute there.
I'm about to suggest that we repeat it in Infoboxes (the info panels in articles, top right), so that it's more easily machine readable, and can be made available via DBpedia http://en.wikipedia.org/wiki/DBpedia etc.
There is stalled proposal on OSM, to include phonetics: http://wiki.openstreetmap.org/wiki/Proposed_features/Phonetics - one issue is that the IPA character set isn't unicode compatible. Suggestions for overcoming this are invited!
That aside, there is nothing to stop someone running a script to import IPA pronunciations from Wikipedia to OSM (or anything else); as it's under an open licence. If someone does that, it would be cool if they'd complete the Wikipedia tag at the same time: http://wiki.openstreetmap.org/wiki/Key:wikipedia - we sorely need someone to run a script to do that.
A sanity check, that the Wikipedia article is the correct one for a place in OSM, can be done by checking the former's coordinates and categories.
Oops. I missed that you had seen the OSM proposal. Sorry.
Also "isn't unicode compatible" should read "isn't ASCII compatible".
The Wikipedia infobox proposal is now at: http://en.wikipedia.org/wiki/Template_talk:Infobox_settlement#Pronunciation
My sat-nav, interestingly, manages to hold two pronunciations - "St Alb'ns" most of the time, but one or two junctions are "St All-bans". Very disconcerting...
Has anyone ever built an IPA-to-speech-synthesiser system? It feels (on paper) like it ought to be practical, and it would be the core element to doing this - we could harvest the IPAs from Wikipedia or elsewhere.
I've noticed this too with some place names. Really odd.
I mentioned this post in the OSM mailing list for GB; it's reuslted in some interesting comments there: http://lists.openstreetmap.org/pipermail/talk-gb/2013-January/014342.html
2 anecdotes come to mind:
1: During one of our holidays we decided to go camp in Belgium (yeah, I know!) and the place name was called Gesves. This was pre sat nav and our French was pretty poor (still is). We had a lot of fun asking for directions before someone understood where we wanted to go and told us the right pronunciation.
2: Recently been using my sat nav (Copilot Live) for trips to NL. The funny thing is that I downloaded the maps and both Dutch and English voices. Sadly the English voices are really bad at pronouncing Dutch place names. We were rolling on the floor (not safe in a car) at the sat nav pronouncing places like "Breda Oost" as "Breed Oost" instead of something like "Braidah Oh-wst". Also: very hard to change voice to Dutch while driving.
Andrew: I've just been informed of a way to set up JAWS to recognise IPA symbols. NVDA is an open-source alternative, but I can find nothing in its documentation about IPA. I've raised a ticket to add IPA support in NVDA.
Cristiano: I'd have detoured to Hellevoetsluis, just to hear how that was handled!.
It looks like JAWS will recognise IPA symbols, but read them as "transliterated" symbols rather than characters - which probably isn't much help here.