Background: Every properly formed English sentence, not counting interjections or sentences pasted together by conjunctions or certain punctuation, is one of only ten possible patterns.

http://www.kent.edu/writingcommons/resources/upload/grammaticalsentencepatterns.pdf
http://www.oxfordtutorials.com/Ten_Sentence_Patterns.htm
http://banal-atrocities.blogspot.com/2008/08/sentence-patterns-1-5.html
http://banal-atrocities.blogspot.com/2008/08/sentence-patterns-6-10.html

This is a striking fact that has long fascinated me. First, the number 10 is intriguing in itself, since it’s the same number as the number of fingers that humans have. Is it just a coincidence that the standard language of our species evolved to have exactly the same number of fingers that we used to count things while we were evolving? Second, the fact that seemingly every complex concept in the universe could be represented using a framework of only ten forms must be significant. That train of thought immediately leads to a number of questions: Why exactly ten? Can this number be reduced to a smaller number of forms by either elimination of redundancy, or by generalization? Does this number hold for languages other than English? If so, what are the quantities of sentence forms associated with other languages, and where is there a list of these language names and numbers? Did this number ten evolve just because of the human brain limitation to the “magic number 5” simultaneous concepts, since we either couldn’t count beyond 5 using fingers of one hand to represent/reinforce our knowledge, or since we couldn’t disambiguate an uttered sentence of too many possible simultaneous resolutions? If we extended this number by consciously enhancing the grammar we currently use only by evolutionary default, can we thereby represent concepts we couldn’t represent before, or at least do so more efficiently? If we make such any such enhancements to our standard grammar, will that shed any light on natural language understanding or new ideas for AI?

For later reference of discussion, the following is a list of the ten sentence patterns. You can view specific examples of sentences in these patterns (which is very useful to do) at the above links. I enhanced this list to include each pattern’s equivalent passive form (#p), if it exists, as well as the usual active form (#a) since I was unable to find such a list online, and I’ve already run into one person online who was stumped while trying to match a passive sentence to one of those ten forms (http://www.fluther.com/151135/which-of-the-10-sentence-patterns-is-this/). Since this thread is going to get long, I think I’ll break it up into smaller posts.

be = some form of the verb “to be”
ADJ = adjectival
ADV = adverbial
ADV/TP = adverbial of time or place
NP = noun phrase

(#1a) - active form
NP be ADV/TP
(#1p) - passive form
(none)

(#2a)
NP be ADJ
(#2p)
(none)

(#3a)
NP1 be NP2
(#3p)
(none)

(#4a)
NP linking-verb ADJ
(#4p)
(none)

(#5a)
NP1 linking-verb NP2
(#5p)
(none)

(#6a)
NP intransitive-verb
(#6p)
(none)

(#7a)
NP1 transitive-verb NP2
(#7p)
NP2 be transitive-verb [by NP1]

(#8a)
NP1 transitive-verb NP2 NP3
(#8p)
NP2 be transitive-verb NP3 [by NP1]

(#9a)
NP1 transitive-verb NP2 ADJ
(#9p)
NP2 be transitive-verb ADJ [by NP1]

(#10a)
NP1 transitive-verb NP2 NP2
(#10p)
NP2 be transitive-verb NP2 [by NP1]

Years ago when reading through “Understanding English Grammar” (Martha Kolln and Robert Funk), which is the first place I learned about the ten sentence patterns (why didn’t any of my teachers teach us this before I got to college?!), I came up with the idea of generalizing these ten patterns into a single pattern, in order to get a better grasp of how the patterns related to each other, especially in a visual/diagrammatic way. For now I’ll show my original representation method, since my later methods and insights are probably very publishable, so I’m reluctant to disclose much about them.

I ignored the passive forms for this study. Note that sentences with “to be” are essentially describing only static objects or static scenarios. I represented those as nodes/circles. I represented non-be verbs with arc/arrows. This is standard practice for representing things in graph theory.

Together, this produces less than 10 basic diagrams containing only circles and arrows. The circles, according to formal English terminology, will correspond to the three different types of objects: subject, direct object, and indirect object. Even in the most complicated pattern, which is pattern #8, there are only three objects involved (NP1, NP2, NP3). So I consider the subject (the first “O”) to be transferring a direct object (the second “O”) to a third object (the third “O”), with the transfer represented by the verb arrow and the direct object—the object being transferred—lying alongside the arrow. This is shown below.

(#1a)
NP be ADV/TP
O

(#2a)
NP be ADJ
O

(#3a)
NP1 be NP2
O

(#4a)
NP linking-verb ADJ
O

(#5a)
NP1 linking-verb NP2
O———->O

(#6a)
NP intransitive-verb
O———->

(#7a)
NP1 transitive-verb NP2
O———->O

(#8a)
NP1 transitive-verb NP2 NP3
O———->O
    O

(#9a)
NP1 transitive-verb NP2 ADJ
O———->O

(#10a)
NP1 transitive-verb NP2 NP2
O———->O

Using this method of representation, the general pattern of possibilities becomes obvious: there exist only 3 objects/circles at most, plus one verb/arrow, and any number of them (including all of them, in theory) can be missing. Some of these diagrams don’t correspond to any of the ten sentence patterns, but note that all of them can be described in English, usually easily, nevertheless. Below is a representative list of the possible permutations. My claims should still hold even though I didn’t take care to ensure that every permutation was listed here. If the verb is missing, then that merely corresponds to a description of the listed static objects.

Represented as vector element permutations: (For brevity I’m using “N” for “NP”, and “-” for “absent”.)

[N1, N2, N3, V]

[N1, N2, N3, -]
[N1, N2, -, V]
[N1, -, N3, V]
[-, N2, N3, V]

[N1, N2, -, -]
[N1, -, N3, -]
[-, N2, N3, -]

[N1, -, -, V]
[-, N2, -, V]

[-, -, N3, V]

[N1, -, -, -]
[-, N2, -, -]

[-, -, N3, -]

etc.

Represented as illustration permutations:

O———->O
    O

O———->O

O———->

———->O

———->

O       O
    O

O       O

O

      O

etc.

I think I’ll stop there in my description of this train of thought, though as I mentioned it got more involved and more interesting as I went along. To generalize this presentation so far, I’ll just note:

1. The vector representation of permutations, where each element can be present or absent, namely [N1, N2, N3, V], seems to effectively generalize all ten sentence patterns, though it adds many patterns that are either not used, or can be described using other sentence patterns. This is a minor insight in that the set of 10 patterns we use are merely a spanning set of patterns that suffice to represent the most commonly needed concepts for language / thought conveyance, somewhat analogous to finding a small set of logic gates from among AND/OR/NOT/NOR/XOR/etc. that will span all possible logic functions. This in turn implies that there is neither anything special about the number 10, nor that those particular sentence patterns are special. The more important observation is that a vector of only 4 elements, where each element can be a (grammatically correct) combination of words from a dictionary, suffices to describe anything that can be conceived by human beings, at least as far as I can tell.

2. The above observation can be obviously generalized to an arbitrary number of objects, though for all practical purposes, 3 objects always suffice. A similar situation occurs in semantic nets, where a relationship involving three objects can be described with a semantic net, but only with difficulty, after the simpler relationships have been clustered and/or generalized.

3. Even with the above generalizations, that doesn’t answer questions about what we need to actually do with the (4-element vector) information so gleaned. I suppose the above scheme makes it easier to understand a situation, say for a chatbot that needs to understand that an object has changed its location or position status due to a verb whose action was “give” or “won” or “mailed”, but that doesn’t necessarily shed more light on natural language understanding or AI. Or does it?

So that I’m not operating at too abstract a level, I’d better post some concrete examples of these sentence patterns to anchor all these generalizations in reality. Here is such a list per sentence pattern, all in active form, and all examples I found online. (The sentence pattern descriptions shown here vary slightly from the descriptions I gave above, due to being pulled from a different web site, but the correlation/meaning is obvious.)

(#1)
NP1 + V-be + ADV/TP
Fido is in his kennel.
My friends are at the library.
My friends are here.
My grandmother was there.
She is in a bad mood.
The children are outside.
The school bus is late.
The students are here.
The students are upstairs.
The supervisor was in a good mood today.
The team is outside.

(#2)
NP1 + V-be + ADJ
Fido is tired.
His clear tenor voice was quite lovely.
My mother is off her rocker.
The child is stupid.
The children are smart.
The dress was pretty.
The students are diligent.
The team is hard working.

(#3)
NP1 + V-be + NP1
Fido was a beautiful dog.
Mr. James has been a teacher for forty years.
Mrs. Brown was a bad cook.
That team is the Raiders.
The astronaut is an old man.
The child is a genius.
The children are scholars.
The students are scholars.
The widow is a talented knitter.

(#4)
NP1 + LV + ADJ
Fido seems anxious.
Marianne looks like her mother.
My sister Sam seemed on edge.
The cake on the table looks delicious.
The child seems honest.
The children seem diligent.
The cookies smelled delicious.
The students look diligent.
The students seem diligent.
The teacher thought him stupid.

(#5)
NP1 + LV + NP1
At a very early age, Joan became a Buddhist.
Fido proved a champion.
Her friend seemed like quite a student.
The children became foster kids.
The children became scholars.
The students became scholars.
The yellow bird became an annoyance.

(#6)
NP1 + V-int
Boys sing.
Fido slept.
Harry jumped off the box.
In a few weeks my cousin will arrive.
In a few weeks my cousin will arrive with my uncle.
I wrote.
One of the thieves must have been hiding in the basement.
The boys in the choir from Detroit sing sweetly at Christmas.
The club members arrived.
The marathon runner ran.
The students rested.

(#7)
NP1 + V-tr + NP2
Fido chased squirrels.
Harry took off his raincoat.
JCCC students write amazingly well-written essays about themselves.
My pie won first prize at the State Fair.
Students write essays.
The archer shot an arrow into the target.
The car needs new tires.
The dog bit me on the leg.
The students studied their assignments.
The students study their books.
The woman passed the test.

(#8)
NP1 + V-tr + NP2 + NP3
Fido won Fred a prize.
I gave the teacher my essay.
I nervously gave my demanding English teacher my perfectly correct essay.
Kerry mailed Frank the package.
My boyfriend gave a diamond ring to me.
Smithers gave the employees a raise.
The judge awarded Mary the prize.
The players gave the other team the ball.
The students gave him their books.
The students gave their teacher an apple.

(#9)
NP1 + V-tr + NP2 + ADJ
Fido found it upsetting.
I want my tea sweet.
The jury found the defendant guilty.
The members find the club interesting.
The parents considered their child a genius.
The students consider him intelligent.
The students consider the teacher intelligent.
The teacher made the test easy.

(#10)
NP1 + V-tr + NP2 + NP2
Fido feared Bo the alpha dog.
Mary considers Jon a jerk.
Most people consider Jacobsen a loyal friend.
She considers her teacher the best.
The principal called me a bother.
The students consider the course a challenge.
The students thought it a challenge.
They named their dog Oscar.

What about time and place indicators? These aren’t interjections nor compound sentences.

Jan Bogaerts - Dec 20, 2012:

What about time and place indicators?

I’m not completely sure what you’re asking. Time and place information added as an aside typically occurs in adverbials / adverbial clauses or prepositional phrases such as “on Saturday” or “in school”, and adverbials are modifications of the verb. Each noun and verb of a sentence typically has modifiers hanging off it when a sentence is diagrammed, and adverbials would hang off the verb. Such a modifier of time or place could also hang off a noun, such as in “Rome in ancient times” or “the man in the car” without a verb being involved, though, which is called a noun post-modifier. In my representation such a diagrammed sentence would look something like 4 small trees, one growing out of each of the 4 elements of my 4-element vector, almost like a flower pot with 4 flowers growing in it.

See examples 29-30 at this page:
http://grammar.ccc.commnet.edu/grammar/diagrams2/one_pager1.htm

http://www.chompchomp.com/terms/prepositionalphrase.htm
http://learnenglish.britishcouncil.org/en/english-grammar/adverbials/adverbials-time
http://grammar.ccc.commnet.edu/grammar/adverbs.htm
http://en.wikipedia.org/wiki/Adverbial
http://www.criticalreading.com/noun_phrase.htm

Since you posted, I’ll use this as an opportunity to flag some minor errors I later found in my three posts:

2nd post:
“arc/arrows” should be “arcs/arrows”
“position status” should be “possession status”

3rd post:
“The parents considered their child a genius.” should be under pattern #10 instead of #9.

I also forgot to mention that another thing that makes the presence of exactly ten patterns interesting is that this is a great example of how human beings learn by example: subconsciously through many examples English speakers evidently learn the implied underlying skeleton/structure/rule that holds words together in acceptable order without being told those rules beforehand. That’s like the difference between learning in an artificial neural network versus explicit rules in an expert system. Only after subconscious subsymbolic learning do stored, generalized patterns rise to the level of attention whereby they could then be represented as rules.

By the way, here are some references to “universal gates”—logic gates (or sets of logic gates) from which all other gates can be represented, to which I alluded in my analogy in my 2nd post:

NAND logic
http://en.wikipedia.org/wiki/NAND_logic

NOR logic
http://en.wikipedia.org/wiki/NOR_logic#Making_other_gates_by_using_NOR_gates