A taxonomy of comma splices

Working in academic ESL, I see all manner of comma splices – ones that immediately strike me as non-native, ones that seem to reflect English acquired via Facebook, and ones that could be made by professional writers. I’m 90% sure I saw some in the David Sedaris book I just finished, a COCA search finds a bunch under every genre (try the search terms “, _nn* [be] .”), and as a few survey respondents have pointed out, Dickens’ A Tale of Two Cities begins with a Guinness-worthy streak of them.

I set out trying to separate the types of comma splices I was seeing, assuming that they had some kind of logic to them. After all, all of the people I’ve seen splicing with commas know that periods (full stops) exist and are presumably using commas for some reason. What follows below is a list of the categories of comma splice that I saw with examples written by me, followed by the results of a survey I did in an attempt to figure out which type of comma splice is the least offensive. My Twitter followers are mostly language teachers (as are the LinkedIn groups that I posted the survey in), so their reactions ought to be interesting.

For the record, and just to show that I’m not just asking my Twitter followers to figure out what a comma splice is, a comma splice is two independent clauses separated by a comma and only a comma. Take out the comma and it’s a run-on, an error that probably deserves its own survey. You can fix comma splices in three ways:

  1. Add a conjunction (preferably and, so, or but – many teachers use the acronym FANBOYS, but honestly, how often do you see “for” as a conjunction?)
  2. Change the comma to a semicolon, or
  3. Change the comma to a period.

See? Here are the categories.

Elaborative comma splices

The price does not reflect the cost of production, it reflects supply and demand.

Similar to the corrective comma splices below, this type of comma splice seeks to add information to the previous independent clause, in this case by adding information. I flubbed the writing of this one a bit by leaving out the word “just”; feel free to ignore this item. Run your own survey if you like.

Explanatory comma splices

The chain went bankrupt, its sales had been falling for years.

Here, the independent clause after the comma explains the one before; it gives a reason. Anecdotally, these are very common among ELLs.

Juxtapositional comma splices

Dogs are descended from wolves, house cats came from feral cats.

Here, the comma separates two independent clauses that are meant as opposites of each other. In style manuals, cases like these would call for a coordinating conjunction like “but” or “yet”, or a subordinating conjunction like “while” or “although”. Many readers, not to mention writers, seem to find them unobjectionable with no conjunction at all.

Corrective comma splices

Richards was not the director, Denton was.

Here, the second clause supplies the correct information whose existence is implied in the first.

Incidentally, the range of relationships between clauses that one sees in comma splices makes one feel the poverty of meanings in the English palette of conjunctions, and particularly the coordinating conjunctions.

Chronological comma splices

The new phone model came out, planning for the next model began soon after.

In this comma splice, the connection between the clauses is chronological; the first precedes the second.

I may have tripped an additional wire in readers’ minds here by making the subject of the second clause a gerund, which could be interpreted as the start of a participial phrase, triggering a garden path (see below). This wouldn’t be a graceful participial phrase, but of course people don’t always respect “the rules” as those go either.

Non-sequitur comma splices

The air is thick with moisture, the office has a meeting at noon.

Not much to say except that this sentence was as unpopular as you might imagine.

Appositive comma spices

The article was written by Malcolm Gladwell, he is a Canadian journalist.

Here, the second clause also supplies more information about something in the first clause by defining a noun. I’m calling them appositive comma splices because without a subject or verb they would simply be appositive phrases.

Conjunctive adverb comma splices

There is a water shortage, therefore citizens are being asked to take shorter showers.

Here, the comma comes before a word with a similar meaning to a conjunction but in a different grammatical class. Called “linking words”, “transitions”, or “conjunctive adverbs” by writing teachers, these words most commonly come at the start of a complete sentence or between commas, as with some other adverbials, in the middle of an independent clause set apart by commas. To illustrate what I understand is the conventional usage of these words:

  • There is a water shortage. Therefore, citizens are being asked to take shorter showers.
  • There is a water shortage; therefore, citizens are being asked to take shorter showers.
  • There is a water shortage. Citizens, therefore, are being asked to take shorter showers.

As an academic writing teacher, my radar is always up for comma splices with these words (which we usually teach explicitly as a counterpart to coordinating and subordinating conjunctions), and to be honest I’m a bit disappointed that so many people disagree with me. Being a responsible writing teacher, though, I need to respect the vox populii. My job is not to make my ELLs the most comma splice-averse people on campus.

Quasi-participial phrase comma splices

The customers are waiting in line, they are holding their baskets.

Like the appositive comma splice above, the second independent clause could easily be reduced down to a familiar and well-accepted construction, the participial phrase.

(side note: I absolutely hate the term “reduced adjective clause” for participial phrases. I don’t think the term bootstraps learners’ understanding enough to justify the additional metalanguage and I don’t think it captures at all how fluent speakers think of them.)

Rephrasing comma splices

It is wasteful, it is inefficient.

Here, the second phrase is semantically very similar to the first. Anecdotally, I don’t see these very much in academic writing classes.

Garden path comma splices

The manager adjourned the meeting, switching off the projector caused the room to go dark.

This is the least comfortable for me to read, and is the reason that, if I remember correctly, Steven Pinker rejects any emails written with comma splices. Here, the comma splice encourages a misprediction by the reader as to how the sentence will proceed – the reader imagines that the manager adjourned the meeting by or while switching off the projector, only to trip over another finite verb in “caused”, which forces them to reappraise “switching” as a gerund instead of as a present participle, leading them up the garden path, so to speak.


The above were presented to anyone who followed my link (mostly Twitter followers, who are mostly language teachers) with a 1-5 Likert-like scale of acceptability, 1=totally unacceptable, 3=don’t know/can’t decide, 5=totally acceptable. The following blurb was presented as well:

The following are all examples of comma splices. Please rate their acceptability as if you were a college writing teacher and they were produced by freshman writing students for a take-home writing assignment that has a grading rubric that includes grammatical accuracy.
For “totally unacceptable” answers, there would be a penalty in points for the error in question. For “totally acceptable” answers, the sentence in question could stand as an example of correct usage and could be recommended to other students.

46 people answered the survey, which again makes me wish I’d had Twitter when I was doing my MA research.

First major result

People vary widely in how they view the comma splices above. The mean score across all items was 2.33, slightly disapproving, and the mean standard deviation was 1.26. The median across all answers was 2 and the mode (most common answer) across all answers was 1.

Second major result

People who accept one of the sentences above tend to accept others as well; people who don’t don’t. All items are positively correlated with all other items, although some more than others. The average correlation of every item with every item (including itself) was 0.407.

Other results


The mean scores were:

  1. The price does not reflect the cost of production, it reflects supply and demand. (elaborative, but reads like corrective) Mean: 3.20, Median 4, Stdev 1.55; notably correlated with #2 (explanatory) and #3 (juxtapositional) at about 0.51 each, and #4 (corrective) at 0.59. This is one of 3 of my sentences that scored above 3, i.e. more accepted than not. 2 of them, this one and #4, have the same “not A, B” structure (although I didn’t intend for this one to), making that an early candidate for a characteristic of “acceptable” comma splices. It is probably good news that these two are more correlated with each other than most.
  2. The chain went bankrupt, its sales had been falling for years. (explanatory) Mean 1.84, Median 2, Stdev 1.09; notably correlated with #1. Early on, when I only had about 20 respondents, this one was correlated with #5 (chronological), #7 (appositive), #9 (quasi-participial phrase), and #11 (garden path). I guessed that people who accept (or not) one type of comma splice tend to accept (or not) others as well, and these hills in the correlations might even out with more survey participants. It turns out that is what happened.
  3. Dogs are descended from wolves, house cats came from feral cats. (juxtapositional) Mean 2.40, Median 2, Stdev 1.48; notably correlated with #1. This one fell slightly under mean acceptability. I suppose it screams a bit too strongly for a semicolon.
  4. Richards was not the director, Denton was. (corrective) Mean 3.71, Median 4, Stdev 1.41; notably correlated with only #1, which shares its “not A, B” structure. This one had the highest score overall and was one of 3 (#1, this one, and #8) that had a mode of 5, “totally acceptable”.
  5. The new phone model came out, planning for the next model began soon after. (chronological) Mean 1.96, Median 2, Stdev 1.17; notably correlated with #6 at 0.654 (see below).
  6. The air is thick with moisture, the office has a meeting at noon. (non-sequitur) Mean 1.38, Median 1, Stdev 0.89; notably correlated with #5. This one had the lowest score, as one would expect. It seems that chronological comma splices are disapproved of by the same people who disapprove of comma splices whose clauses are completely unrelated.
  7. The article was written by Malcolm Gladwell, he is a Canadian journalist. (appositive) Mean 1.66, Median 1, Stdev 1.16; not notably correlated with any other item.
  8. There is a water shortage, therefore citizens are being asked to take shorter showers. (conjunctive adverb) Mean 3.25, Median 4, Stdev 1.60; notably correlated with #9 at 0.557. I’m not sure what makes people who accept this sentence also accept the next (and vice versa), as their clauses have very different relationships with each other. To me, this item is unique in that the relationship between the two clauses is actually transparent and given literally by a conjunction-like word (therefore). I don’t see a connection between this item and any other, but it seems that some people do.
  9. The customers are waiting in line, they are holding their baskets. (quasi-participial phrase) Mean 1.93, Median 2, Stdev 1.08; notably correlated with #8 (0.557) and #10 (0.516). Even more mysterious than the lack of relationship (at least to me) with these two items is the fact that this item is much less accepted than either of them.
  10. It is wasteful, it is inefficient. (rephrasing) Mean 2.68, Median 2, Stdev 1.55; notably correlated with #9. This is one where respondents might also be reacting to the redundancy in the two clauses. I said in my blurb to treat the assignment as a first-year college English assignment, and respodents might have struggled with thinking of a college writing assignment where saying the same thing twice, even with orthodox correct grammar, would be acceptable.
  11. The manager adjourned the meeting, switching off the projector caused the room to go dark. (garden path) Mean 1.57, Median 1, Stdev 0.93. I would actually be worried if this one were approved of by most people.

Last, the item with the highest average correlations with the other items overall is #3 (chronological) at 0.469, followed closely by #9 (quasi-partipial phrase) at 0.462. In a nutshell, if someone accepts #3 or #9, he or she will probably be willing to accept other comma splices too, and vice versa – someone who rejects #3 or #9 probably rejects others.

Unsurprisingly, a comma splice in which the two clauses appear to have nothing to do with one another (#6) strikes almost everyone as wrong. The fact that this one scored lower when all of them are “technically” wrong is interesting, showing that semantic or discursive rather than grammatical congruence determines acceptability for most people. As is the premise of this entire project, all of these are equally wrong for exactly the same reason, at least according to the rules of grammar most of us use when giving feedback on student writing. If my Twitter followers are a sign of the attitudes that most grammatically-minded people take toward comma splices, then corrective comma splices may be bordering on safe for beginning writing students, and at least some conjunctive adverbs like therefore might be heading toward acceptability in the grammatical role usually reserved for conjunctions.


There was a space at the end for respondents to leave qualitative comments on what makes a comma splice acceptable or not. Here are some of them:

It seems that if the topics of the two parts are more closely aligned the error doesn’t seem quite as wrong. [That’s probably exactly what my students think]

It seems to me like if the comma could be replaced by a semi-colon or a colon it “sounds” more acceptable.

Something about rhetorical effect and whether the subject has changed. Dickens and the beginning of A Tale of Two Cities set some kind of precedent of acceptability.

Most of them should just be periods; some could be semicolons. [When you do a survey of different sentences which make the same language-related mistake in different ways, you always get a few respondents who refuse to look past the issue that is the baseline for the entire exploration. Ah well, uncovering this attitude is part of the purpose of a survey like this]

I love them all

I wish there was a little more info on the type of assignment. Going by the statement that there’s a grammar rubric, I’m thinking it’s straight up conventional rules, and they’re all “wrong” in that case. If it was at all a creative writing assignment, I’d give some of the ones that imply a series more leeway, like the wasteful/inefficient one. I think even more so if there was a third clause: It is wasteful, it is inefficient, it is wrong. [A very useful point!]

Super interesting survey! They’re more acceptable when the second clause describes the first clause of adds more information, e.g., Gladwell question. [Interestingly, most respondents disagreed.]

These are all independent clauses [See above]

There were all unacceptable …

I guess it might sound more acceptable if the second sentence seemed to be tacked on as an afterthought, like in the Denton example.

Continuation of a idea- more like an a positive. Or listing.

I think if one of the clauses is made unequal either by shortening it or using a pronoun, a comma splice might be acceptable. For e.g. “dogs are descended from wolves, house cats from feral cats.” (But in this case would it still be a comma splice?) [Not according to my sense of grammar orthodoxy since the second phrase lacks a verb]

As with anything in language matters, conventions are merely an agreement of that languages speakers. They are subject to change with each generation. If a new generation agrees that they should change conventions, then the convention will be changed. It appears that normally 75% agreement is a threshold for introducing the new convention. It would seem to me that young writers at the threshold percentage have changed the convention of the use of the comma, especially in the case of what traditionalists now call comma splices. [Curious where the 75% number comes from]

Thanks for reading! Feel free to share these results over Thanksgiving dinner.

2 thoughts on “A taxonomy of comma splices

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s