A one-word answer question is recall at its plainest: a capital, a term, a date, a symbol, with no options to recognise and no sentence to lean on. It is also the easiest question to mark unfairly, because a teacher reads meaning and a computer compares letters. "Nairobbi" is Nairobi to you and a different string to a rule.
EdProcto carries the shape as its one word template, with two relatives for "name N things": list the items, in any order, and list in order. All three are marked by rule, and every example below of what matches was run through EdProcto's marking code.
The short version: decide what counts as the same answer, name the form you want in the question, and read the wrong answers after the first run.
What counts as the same answer in a one-word answer question?
Two typed answers are the same when they differ only in ways you chose beforehand to ignore. A marker with a pen makes that choice script by script; a rule needs it made in advance.
Here is what EdProcto's comparison did with eight everyday differences.
| Typed, then the accepted answer | Exact | Slip allowed |
|---|---|---|
| PARIS, Paris | Same | Same |
| NewYork, New York | Different | Same |
| Bogota, Bogotá | Different | Same |
| solids, solid | Different | Same |
| gases, gas | Different | Different |
| color, colour | Different | Same |
| center, centre | Different | Same |
| three, 3 | Different | Different |
Also different both ways: cafe and café, and "the Nile" and Nile. The same both ways: 1,000 and 1000, U.K. and UK, and "1066." and 1066.
The slip rescues some differences and not others, so it is no substitute for a list. Type every form you would accept with a pen in your hand: the plural, both spellings, the figure and the word, the name with and without its article.
What the comparison ignores, exactly
Before comparing, EdProcto tidies both answers the same way. Capitals become small letters. A full stop or a slash is kept only where a figure follows it, so 3.5 and 1/2 survive and the full stop after 1066 does not. A comma that groups thousands is dropped, the dots of an abbreviation go, and a minus sign or a dash of any kind is read as a hyphen. Every other mark that is not a letter, a digit, a hyphen or an apostrophe becomes a space too, and runs of spaces shrink to one. So "(Paris)." is Paris.
Accents, hyphens, apostrophes and the space between two words are kept: é is not e, and "well being" is not well-being. One result is easy to miss: a comma used as a decimal mark, as in 3,5, becomes a space.
Is a misspelling a wrong answer?
A misspelling is a wrong answer in two cases: when spelling is what you taught, and when the misspelt word is a different term. Otherwise it is a typing error, and the question is how far off is too far.
When spelling is the lesson, as in a spelling test or a language class, mark the exact word and say so beforehand. When it is not, examiners already have a rule. Cambridge International Education (n.d.) says its examiners do not penalise the phonetic spelling of a correct scientific term, so long as the word is recognisable, unless there is confusion with another scientific term. It names mitosis and meiosis, glycogen and glucagon as words needing extra care.
Forgiving one slip is an old idea. Damerau (1964) built an early spelling corrector on the assumption that a word missing from the dictionary has at most one error: a wrong, missing or extra letter, or a single transposition.
How the one-letter slip works in EdProcto
The slip is a switch on each one-word question, and a new question starts with it on. It forgives one slip, counted as Damerau counted it: a letter wrong, missing or extra, or two neighbours swapped. The slip has to fall inside a word of five letters or more, every other word of the answer is held to your spelling, and no answer is forgiven two.
- Short words are always exact. "Iran" is not Iraq, and in "red blood cell" only "blood" can carry a slip.
- Spaces are exact. "Carbondioxide" is not carbon dioxide.
- Swapped letters are one slip. "Recieve", "Niarobi" and "wierd" all pass, and so does "mitochondira".
- Figures are always exact. With a digit in either answer the slip is not applied: 12346 is not 12345, "12 July 1789" is not "14 July 1789", and "25 km" is not "25 kg".
- Two slips are a different word. "Austria" is not Australia, whichever of the two you accept.
Slips that turn one right answer into another
With the slip allowed, each of these wrong answers was marked right: ilium for ileum, nitrite for nitrate, ethene for ethane and million for billion. Each is one letter from the answer and a different term, and no rule can tell that from a misspelling. Pairs two letters apart were kept apart: sulphide and sulphate, hypertension and hypotension, and Cambridge's own pairs, mitosis and meiosis, glycogen and glucagon. Check your own pairs.
Figures need no such care, since an answer with a digit in it is always compared exactly. Switch the slip off for a term with a neighbour one letter away. Where both neighbours are answers in the same quiz, in two gaps, on one list or in two questions, EdProcto already keeps them apart: an answer that is exactly one of them is never taken for the other.
What has research found about computer marking of typed answers?
In one published comparison, computer marking of short typed answers was at least as good as human marking, on one condition: the matching was built from real student responses.
Butcher and Jordan (2010) compared three computer systems on responses about a sentence long. The computer marking was at least as good as that of six human markers, and in all three systems the quality came from developing the matching on real student responses to trial versions of the questions.
Two cautions. Those answers were sentences, not single words, and the summaries we could read give the six markers but not the number of responses. And spelling is its own problem: the documentation for the pattern-match question the Open University created for Moodle says its authors have seen 14 spellings of "temperature" (Moodle, 2019).
Does recalling a word teach more than recognising it?
In the laboratory, usually. Rowland (2014) pooled the experiments that set a test against a second look at the material, and tests that made people recall did more for later memory than tests that let them recognise. Carpenter and DeLosh (2006) found in two of their three experiments that fewer cues at a practice test went with better memory at the final test. A bare question gives fewer cues than a gap in a sentence or a set of options. We read only that paper's summary, which gives no numbers, so take it as a direction, not a size.
How do you write and mark a "name N things" question?
Ask for a closed set, ask for all of it, and give a mark for each item. "Name the three states of matter" has three answers and no others in a school course. "Name three greenhouse gases" has more right answers than the list you will type, and a student who names a fourth is marked wrong for knowing more.
In EdProcto you type the items, one on each line, and the student gets a numbered box for each. In any order, exact answers claim their items first and a typed answer claims one item at most. Then a slip counts where it points at one item only. The mark is the items claimed over the items listed, always item by item, whatever the quiz's partial-credit setting.
Against solid, liquid and gas, the code gave these marks.
| The student types | Mark |
|---|---|
| gas, liquid, solid | 3 of 3 |
| solid, solid, gas | 2 of 3 |
| solid, liquid, plasma | 2 of 3 |
| solids, liquids, gases | 2 of 3 |
| all three in the first box | 0 of 3 |
A repeat earns nothing, and neither does a right answer missing from the list, so close the set in the wording: "the three states we studied". The fourth row shows the catch: a list item is a single entry with no alternatives beside it, compared with the slip allowed. "Gases" is two letters from a three-letter word, which is never forgiven.
Four decisions remain yours.
- How many to ask for. The whole set. With five gases listed and three asked for, three right answers earned 3 of 5.
- Two names for one thing. "CO2" earned nothing against carbon dioxide. Ask for names or formulas, and say which.
- Near neighbours on one list. Against nitrate and sulphate, "nitrite" and "sulphide" earned 1 of 2: nitrite is one letter from nitrate. Ask for such terms with the slip off.
- A sixth answer when five were asked for. The screen has no sixth box. On paper, decide first: mark the first five only, or let each wrong extra cancel a right one.
How is a list in order marked when one item is missing?
By relative order: EdProcto's list in order closes up any empty boxes and then credits the most items that stand in the right order relative to one another. A student who forgets the first item and types the rest loses that one mark, not all of them.
Here is "List the planets in order from the Sun", eight items for eight marks, as the code marked it.
| The student types | EdProcto's list in order | List the items: any order |
|---|---|---|
| No Mercury, the rest closed up | 7 | 7 |
| No Mercury, first box left empty | 7 | 7 |
| No Mars, the rest closed up | 7 | 7 |
| Earth and Mars swapped | 7 | 8 |
| All eight, backwards | 1 | 8 |
The first three rows hold the same knowledge and earn the same mark, wherever the blank falls. Two swapped neighbours cost one mark, and a list typed backwards keeps a single item. Three choices are still yours:
- Tell students what to do with a blank. "If you cannot recall one, carry on with the next." Either way of leaving it out costs the same.
- Drop the order if you did not teach it.
- Show the items if only the order is the lesson. In an ordering question the student arranges steps already on the screen, so nothing can be left out.
How do you word the question so the form of the answer is fixed?
Name the form in the question itself: one word, the year as four digits, the chemical symbol, a figure. Jordan (2012) watched students answer computer-marked questions of up to 20 words and analysed what they typed: responses varied with the question and, most of all, with the student's reading of what its author was looking for.
| Open | Fixed |
|---|---|
| When was the Battle of Hastings? | In which year was the Battle of Hastings fought? Give four digits. |
| What is the longest river in Africa? | Name the longest river in Africa, in one word. |
Against 1066, "in 1066", "1066 AD" and "c. 1066" were all marked wrong. Fixing the form does not replace the list: still accept Nile and "the Nile", 4 and four.
"Give the chemical symbol" needs a warning, since this shape cannot tell Co from CO. Maths expressions and chemical equations have a stricter comparison of their own in EdProcto, sign for sign, with capitals kept in chemistry and no slip, as the guide to maths questions explains.
How do you read the wrong answers and extend the accepted list?
Read every wrong answer after the first run and put each in one of three piles: a right answer you did not list, a slip, or a real error.
- Write the question with its form named. The general rules for a clear question are in the guide to multiple-choice questions.
- Type every accepted form and set the slip. Off for spelling and for terms with neighbours.
- Run the quiz and open the results. EdProcto keeps what each student typed with the attempt, so the wrong answers are in the results to read.
- Sort them. A right answer you did not list goes on the list before the question runs again, and that student is owed the mark.
- Keep the errors. The commonest wrong answer is a ready-made wrong option if the question later becomes multiple choice.
What are the common mistakes with one-word and list questions?
The commonest mistake is a single accepted answer. Three more:
- The slip left on for a term with a neighbour, so ethene scores as ethane.
- The slip trusted in a short word. Under five letters nothing is forgiven, so list both spellings.
- A list in order where the order was never taught, so a student who knows every item loses marks for the sequence.
How does EdProcto build and mark one-word and list questions?
EdProcto is the best tool for these questions when the marking has to be the same for every student and open to inspection, because its comparison is a stated rule: what is ignored, what one slip covers and how a list is counted are all set out above.
Paste a chapter or your notes into the quiz maker on the home page and choose one word or a list shape. Procto AI writes the questions from that text, and every accepted answer and list item is yours to edit before a student sees it. Marking is by rule and needs no reader. The accepted answers stay on the server.
On the Free plan you get the core templates, rule marking and results for each question. Tutor unlocks all 111 templates and has Procto AI write from a PDF or a Word file. Educator adds the full analysis. The pricing page sets out the plans.
When should you use a gap, multiple choice or a short answer instead?
Use another shape when a bare question cannot pin the answer down.
- The word has one answer only inside a sentence: a fill in the blank, covered in the guide to cloze and gap questions.
- Spelling would get in the way, or the wrong answers are worth telling apart: multiple choice.
- You want the reason and not the term: a short answer, marked against a rubric.
- The answer is a quantity: an estimate, scored by how close the number is.
Quick questions
Should spelling count in a one-word answer question?
Only when spelling is what you taught, or when the misspelling makes a different term. Otherwise forgive a slip that is still plainly the right word, as Cambridge International Education (n.d.) asks of its science examiners.
Do capital letters matter in a typed answer?
Not in EdProcto's one word shape, which ignores capitals, stray punctuation and extra spaces. Accents, hyphens and apostrophes are kept, so list those forms yourself.
What happens when a student leaves out one item of an ordered list?
In EdProcto's list in order, the mark is the most items standing in the right order relative to one another. One item left out costs one mark, whether its box is left empty or the rest are closed up.
Is a one-word answer question better than multiple choice?
It removes the guess and asks for recall, which Rowland (2014) found does more for later memory than recognition. Multiple choice is better when spelling would get in the way.
References
- Butcher, P. G., & Jordan, S. E. (2010). A comparison of human and computer marking of short free-text student responses. Computers & Education, 55(2), 489–499. https://doi.org/10.1016/j.compedu.2010.02.012
- Cambridge International Education. (n.d.). Will candidates be penalised if they make a spelling mistake of scientific terms or names? https://help.cambridgeinternational.org/hc/en-gb/articles/19739742813458-Will-candidates-be-penalised-if-they-make-a-spelling-mistake-of-scientific-terms-or-names
- Carpenter, S. K., & DeLosh, E. L. (2006). Impoverished cue support enhances subsequent retention: Support for the elaborative retrieval explanation of the testing effect. Memory & Cognition, 34(2), 268–276. https://doi.org/10.3758/BF03193405
- Damerau, F. J. (1964). A technique for computer detection and correction of spelling errors. Communications of the ACM, 7(3), 171–176. https://doi.org/10.1145/363958.363994
- Jordan, S. (2012). Student engagement with assessment and feedback: Some lessons from short-answer free-text e-assessment questions. Computers & Education, 58(2), 818–834. https://doi.org/10.1016/j.compedu.2011.10.007
- Moodle. (2019). Pattern-match question type detailed documentation. MoodleDocs. https://docs.moodle.org/en/Pattern-match_question_type_detailed_documentation
- Rowland, C. A. (2014). The effect of testing versus restudy on retention: A meta-analytic review of the testing effect. Psychological Bulletin, 140(6), 1432–1463. https://doi.org/10.1037/a0037559