Contribute

This dictionary has no words of its own. Every entry here comes from Wiktionary, which anyone can edit — so a correction made there reaches this site, and everything else in the world built on Yorùbá Wiktionary, at the same time.

Why corrections go to Wiktionary, not to us

We did not write this dictionary. We take Wiktionary's Yorùbá data (through Kaikki, which does the first cleanup pass) and rebuild the reading and searching around it. The words, the definitions and the etymologies are somebody else's work, contributed by people who mostly do not know each other.

So when we want to improve what you see here, we do not patch it here. We go and edit Wiktionary, and wait for the next refresh. That is not a limitation we put up with — it is the point. A fix typed into our site would help our readers. The same fix typed into Wiktionary helps every reader, every app, every researcher and every language project working from that data, most of whom we will never hear about.

Yorùbá Wiktionary is not ours and it is not anybody's. It is community-driven and community-owned, and it will only ever be as good as the people who show up to it. There is a great deal of Yorùbá in it already, and a great deal missing.

What this site is good for: finding the gaps

Wiktionary is hard to browse. You have to spell a word a particular way to find it, results arrive mixed in with every other language, and the relationship between two words is often recorded on only one of them.

We fixed those for reading, and the same work makes gaps visible. Search a Yorùbá word however you write it, or search from English. Open a word and see what it is made of, and what is made from it. Anything conspicuously absent from those lists is a gap, and now you can see it without knowing where to look.

Some of them we can count for you. The Data quality panel in the menu holds around 2,300 of them, checked on every rebuild and sorted by how much judgement each one needs.

Learning Wiktionary's conventions

Editing Wiktionary takes some investment. It has its own markup, its own layout rules and its own templates, and a well-formed entry looks nothing like a sentence you would type into a document. None of it is difficult; there is just a bit of it.

These are the pages worth reading, roughly in order:

• Wiktionary:Tutorial — the official walkthrough for a first-time editor, with practice areas where nothing you do matters.
• Wiktionary:Entry layout — what goes where in an entry, and in what order. This is the one to keep open while you edit.
• Wiktionary:Yoruba entry guidelines — the Yorùbá-specific rules, chiefly about which spelling a word is filed under. Underdots in the page title, tone marks on the headword line.
• Wiktionary:Welcome, newcomers — the front door, if you would rather start with the community than the markup.

What a well-formed Yorùbá entry looks like

The fastest way to learn the conventions is to read one entry that follows all of them. odo is the best-formed Yorùbá page we have found, and it is worth opening in another tab.

1. The page is a spelling. The sections are the words.

Seven different Yorùbá words are written odo once tone marks are removed. They are not senses of one word — they are unrelated words that collided. Wiktionary gives each one a numbered etymology section, and everything about that word lives inside its section.

==Yoruba==

===Etymology 1===   òdo   zero
===Etymology 2===   òdo   pig
===Etymology 3===   odò   river
===Etymology 4===   odò   a hardwood tree
===Etymology 5===   odo   the core of a syllable
===Etymology 6===   odo   white yam
===Etymology 7===   odó   pounding mortar

2. Each section gets a name

A numbered section cannot be pointed at from anywhere else — "the third one" changes the moment somebody inserts a section above it. So each one opens with {{etymid}}, which gives that meaning a permanent name:

===Etymology 3===
{{etymid|yo|river}}

All seven sections on odo carry one: zero, pig, river, mansonia, syllable, yam, mortar. That is what makes the rest of this possible, and it is the single most useful thing anyone can add to a crowded Yorùbá page.

3. Senses nest inside the section

One word can mean several things without being several words. odò, the river, is also the lower part of something, and also the south — one etymology, three senses, numbered inside the one section.

====Noun====
{{yo-noun|odò}}

# [[river]]
# [[lower]] or [[inner]] [[part]]
# [[south]]

Getting this wrong in either direction is the commonest fault on a Yorùbá page: senses of one word split into separate etymology sections, or genuinely separate words flattened into one list.

4. Derived terms belong to the section, not the page

This is the part most easily missed, and on odo it is doing real work. Each section lists the words built from that meaning, underneath it:

===Etymology 1===   {{etymid|yo|zero}}
  =====Derived terms=====
  * olódo   "a dunce, one who receives poor grades"

===Etymology 3===   {{etymid|yo|river}}
  =====Derived terms=====
  {{col3|yo |etídò<t:river side> |olódò<t:river spirit> |ojúdò<t:midstream> …}}

===Etymology 7===   {{etymid|yo|mortar}}
  =====Derived terms=====
  {{col2|yo |ìyá-odó<t:pounding mortar> |odókódó<t:any mortar> …}}

Read those three lists together. olódo is a dunce, from zero. olódò is a river spirit, from river. olódó makes and sells mortars, from mortar. Three words that differ only in tone marks, each belonging to a different meaning of the same spelling.

Put all three in one list at the bottom of the page and that information is gone — every one of them would appear to come from every meaning. Attached to their own sections, each says exactly where it came from.

5. A compound says which meaning it came from

The last piece is the other direction: a word built from odo naming which odo it means. dódò, "to arrive at a river", is dé ("to arrive") plus odò ("river"), and its etymology says so by name:

From {{compound|yo|dé|odò|t1=to arrive at|id1=arrive|t2=river|id2=river}}.

id1=arrive points at the {{etymid|yo|arrive}} on the page de. id2=river points at the one on odo. Both exist, so both resolve, and this dictionary can state where dódò came from instead of guessing — which is why its entry here shows the right two words.

The t1= and t2= arguments are the meanings shown to a reader. The id1= and id2= arguments are the machine-readable half, and they are the ones almost always missing.

And one loose end, on the best page we could find

Etymology 7 of odo derives the mortar from a prefix:

equivalent to {{af|yo|o-|dó|t1=nominalizing prefix|id1=nominalizing prefix|t2=to pound}}

That id1 points at a name on the page o-, and o- has no Yorùbá section at all. The pointer is correct and its target has never been written, so it goes nowhere.

We are not saying that to criticise a page we have been praising. It is the most realistic thing on this page: the best-formed Yorùbá entry in Wiktionary still has a loose end, that loose end is one short section somebody could write this afternoon, and it is invisible unless you go looking. That is what contributing here mostly is.

It is also not a one-off. Our data quality report finds nine pointers aimed at names that were never created, and odo's is one of them.

A pile of work, gathered in one place

Every time this site rebuilds it checks the data against itself and writes down what does not add up. That is the Data quality panel in the menu, and it currently holds about 2,300 things worth fixing.

The useful part is that they arrive sorted rather than as one large number. Each row names the word and the page it sits on, says what is wrong and why it matters, gives the wikitext to add, and links straight to the section of Wiktionary you would be editing. Rows are ordered by how much judgement each needs, so the quick ones with the widest reach are at the top. Some of what is in there now:

• 33 cross-references with the wrong tone marks. The word being pointed at exists — the reference just spells it with different tones, so a link the reader should get never appears. Two words are involved and either could be the wrong one, so check both before you type: usually the reference is at fault, sometimes the entry it points at.
• Six with a missing or extra underdot. The same thing, one dimension over.
• Nine pointers aimed at a name nobody ever created, including odo's. Somebody did the work and the target page never got the matching name, so careful writing resolves to nothing.
• 51 pages where several meanings share a spelling and not one of them has a name. The other half of the same problem: words are built from these pages and have no way of saying which meaning they mean.
• 374 entries with no pronunciation. This is the IPA line rather than a recording, and {{yo-IPA}} works most of it out from the tone-marked spelling, so it is bulk work rather than research — as long as the tones on the page are right.
• 775 words whose main spelling the source never confirms. No headword template gave a tone-marked spelling, so we fall back to the page title, which is usually untoned. The spelling shown may well be correct; the tones are simply unverified, and settling them needs somebody who knows the word.
• 1,046 cross-references to words Wiktionary does not have. The largest real gap in the dictionary. Every one is a word somebody has already said is worth having, with no entry behind it yet.

The return on any of them is immediate and it is not only ours. An entry fixed today is live on Wiktionary at once, in this dictionary at the next refresh, and in every other project reading that data whenever they next pull it.

What we are working on now: names for meanings

Two of those rows are what Speak Nigeria is working through at the moment: the 51 pages carrying several meanings and no names, and the nine pointers aimed at names that were never written. Both are the second step of the odo walkthrough above — {{etymid}} — missing from one side or the other. It is a small piece of a much larger job, and it is bookkeeping rather than research. Somebody has usually written down which meaning a word came from already, in plain words a reader can follow. What is missing is the template that lets a machine follow it too.

Seven different words are spelled pa: to kill, to tell, to rub, to gain, to be in a state, to be tight, to be bald. Fifty-four words here are built from one of those seven. Open one and the answer is usually sitting in the etymology already — t1=kill, clear, which tells a person exactly which pa is meant. It tells a computer nothing, because no part of it points at a particular section of the pa page. So this dictionary guesses, and the first guess is often wrong, which is why pàdé, "to meet", currently appears to come from pa meaning "to kill".

We picked this up because of what the dictionary is for. Yorùbá builds long words out of short ones, that is the most useful thing a learner can see, and it is the part most often left out of how the language gets taught and written down. A reader who can follow the links can explore. When a link lands on the wrong meaning, or goes nowhere, the exploring stops, and a learner has no way of telling a real connection from a broken one.

Two lines fix it, and neither works without the other. A name on the section:

{{etymid|yo|kill}}

And a pointer from each word built on it:

{{compound|yo|pa|kó|t1=kill, clear|id1=kill}}

A name nothing points at has no effect; a pointer to a name that does not exist has no effect either — as odo's prefix shows. So every page in the queue below lists both halves.

Choosing a name for a meaning

The name is yours to pick. It identifies one etymology section, it does not have to describe every sense inside it, and it only has to be unique on its own page.

• One or two plain words, lowercase. On odo: zero, pig, river, mortar.
• Enough to tell it from the others on the page. On pa, kill and tell are enough. verb would not be.
• Broad enough for the whole section. The names suggested below come from each section's first definition, which is a starting point rather than a rule. Section 6 of ta covers "to shoot", "to sting", "to be spicy", "to kick" and "to pick" — shoot works, and something broader would be better.
• Hard to change later. Renaming breaks every pointer aimed at it and nothing warns you. Pick something that will still fit when the meaning is written up more fully.
• Read by other people. Wiktionary keeps lists of words sharing a part, such as "Yoruba terms prefixed with a-". A name splits that list by meaning, and your name is what appears on it.

Checking a suggestion before you type it

Where we can, we suggest which name a word should point at. Each suggestion says the same thing: this word records its part as meaning X, and section N of the target page covers X. Open both pages and confirm that.

Two things can be wrong with it. The meaning recorded on the word may itself be vague or mistaken — somebody else wrote it and we take it at face value. Or the section may contain those words incidentally rather than actually meaning them.

There is a third case which is not a mistake in the suggestion. Sometimes the recorded meaning matches no section at all. pàdé records its pa as "to do; action verb", and none of pa's seven sections says that. No pointer is right there: either the meaning written on pàdé is wrong, or pa is missing a meaning it ought to list.

When you are not sure, leave it. A word with no pointer shows here as a gap. A word with the wrong pointer shows as a fact, and other tools will believe it.

This site has no Wiktionary account and makes no automated edits, now or later. Everything below is text for a person to check and type.

The queue