Sentence diagrams

There are many parts of speech in what is called grammar.

Sentence diagrams are a graphical way to show the semantic and syntactic structure of a sentence.

The abstract syntax is the parsed representation of the actual text such as "subject verb".
The concrete syntax is the syntax of the actual text such as "children nap".

Due to differences in human opinion, speech, etc., there are many ambiguities that arise in the parsing of text. This can be the result of intentional ambiguity, deception, jokes, etc.

More: Word and parsing ambiguity

Some people think that a sentence diagram will help understand the meaning. There is no magic in a sentence diagram.
Seldom can a fixed grammar create a sentence diagram. Even then, each POS (Part of Speech), with inflections, needs to be decided.

Thus, a sentence diagram is usually created by deciding what the parse should be and then creating a visual representation as a sentence diagram.

Language processing experts such as Chris Manning, director of the of Stanford Artificial Intelligence Laboratory (SAIL), claim a limit of about 94% on the accuracy with which computers can identify the proper POS of a word in a sentence using NLP (Natural Language Processing), computational linguistics, etc.

Why is there a limit on the accuracy computers can attain on part of speech identification?

Experts can only agree on a given part of speech for a word about 94% of the time.

More: Philippians 3:5-8 not by works

The leading NLP group is at Stanford University, at https://www-nlp.stanford.edu. Their software is summarized and available from http://www-nlp.stanford.edu/software/index.shtml.

Here is an example sentence and the parse output from the Standford parser. This article contains a discussion of the history of commercial and academic efforts to automate patent classifications.…

(ROOT (S (NP (DT This) (NN article)) (VP (VBZ contains) (S (NP (NP (DT a) (NN discussion)) (PP (IN of) (NP (NP (DT the) (NN history)) (PP (IN of) (NP (UCP (JJ commercial) (CC and) (JJ academic)) (NNS efforts)))))) (VP (TO to) (VP (VB automate) (NP (NN patent) (NNS classifications)))))) (. .))) det(article-2, This-1) nsubj(contains-3, article-2) root(ROOT-0, contains-3) det(discussion-5, a-4) nsubj(automate-15, discussion-5) det(history-8, the-7) prep_of(discussion-5, history-8) amod(efforts-13, commercial-10) conj_and(commercial-10, academic-12) amod(efforts-13, academic-12) prep_of(history-8, efforts-13) aux(automate-15, to-14) xcomp(contains-3, automate-15) nn(classifications-17, patent-16) dobj(automate-15, classifications-17) (ROOT (S (NP (PRP It)) (ADVP (RB also)) (VP (VBZ suggests) (NP (NP (JJ new) (NNS approaches)) (PRN (-LRB- -LRB-) (VP (VBG adding) (NP (JJ additional) (JJ structured) (NN language)) (PP (TO to) (NP (DT the) (NN text)))) (-RRB- -RRB-)) (SBAR (WHNP (WDT that)) (S (PRN (-LRB- -LRB-) (S (NP (PRP it)) (VP (VBZ asserts))) (-RRB- -RRB-)) (VP (VBP lead) (PP (TO to) (NP (ADJP (RB statistically) (JJ meaningful)) (NNS improvements)))))))) (. .))) nsubj(suggests-3, It-1) advmod(suggests-3, also-2) root(ROOT-0, suggests-3) amod(approaches-5, new-4) dobj(suggests-3, approaches-5) nsubj(lead-20, approaches-5) dep(approaches-5, adding-7) amod(language-10, additional-8) amod(language-10, structured-9) dobj(adding-7, language-10) det(text-13, the-12) prep_to(adding-7, text-13) nsubj(asserts-18, it-17) parataxis(lead-20, asserts-18) rcmod(approaches-5, lead-20) advmod(meaningful-23, statistically-22) amod(improvements-24, meaningful-23) prep_to(lead-20, improvements-24)

Here is the output from the Stanford POS Tagger, a "maximum-entropy (CMM) part-of-speech (POS) tagger for English, Arabic, Chinese, French, and German, in Java ".

This_DT article_NN contains_VBZ a_DT discussion_NN of_IN the_DT history_NN of_IN commercial_JJ and_CC academic_JJ efforts_NNS to_TO automate_VB patent_NN classifications_NNS ._. It_PRP also_RB suggests_VBZ new_JJ approaches_NNS -LRB-_-LRB- adding_VBG additional_JJ structured_JJ language_NN to_TO the_DT text_NN -RRB-_-RRB- that_WDT -LRB-_-LRB- it_PRP asserts_VBZ -RRB-_-RRB- lead_NN to_TO statistically_RB meaningful_JJ improvements_NNS ._.

More: Natural Language Parsing examples

The NLTK (Natural Language Tool Kit) is available for natural language parsing.

"NLTK is a leading platform for building Python programs to work with human language data. It provides easy-to-use interfaces to over 50 corpora and lexical resources such as WordNet, along with a suite of text processing libraries for classification, tokenization, stemming, tagging, parsing, and semantic reasoning." https://nltk.org

Here is an example sentence and the parse tree from NLTK.

This article contains a discussion of the history of commercial and academic efforts to automate patent classifications.…

More: Natural Language Parsing examples


1. Default precedence.	2. Non-default precedence

What is the value of the expression 2*3+5?

Math people tend to think in terms of text and parentheses.

Computer people tend to think in terms of trees without parentheses.

Changing the root changes the implicit parentheses.

A syntax tree, similar to a sentence diagram does not help decide the meaning since the syntax tree depicts the meaning as decided by the precedence rules. The usual mathematical rule is that multiplication is done first and then addition. Natural language does not have such default rules.

Without rules on operator precedence, the value is ambiguous. Mathematics tends to use parentheses to disambiguate. Computer science tends to use trees since trees are used internally for computational ordering. The parentheses disappear inside the computer. Here are two ways to parse 2*3+5.

A property is "distributive" if it can be applied to both operands.


2. Non-default precedence	2. Multiplication distributed

The same can happen in natural language.

Concrete syntax:

good food and drink

The word "good" is an adjective that describes "food" and/or "drink". A textual way to show the abstract syntax tree is with parentheses. Here are two ways to parse the phrase. The second form can be distributed.

Abstract syntax:

1. (good food) and drink 2. good (food and drink) 2. (good food) and (good drink)

The first way is the usual way. Does this mean that the "drink" is "not good"? The phrase does not say. That part is ambiguous. The context of the phrase may change the parsing and the meaning. ... more to be added ...

Revelation 17:4 And the woman was arrayed in purple and scarlet colour, and decked with gold and precious stones and pearls, having a golden cup in her hand full of abominations and filthiness of her fornication: [kjv]

και η γυνη ην περιβεβλημενη πορφυρουν και κοκκινον και κεχρυσωμενη χρυσιω και λιθω τιμιω και μαργαριταις εχουσα ποτηριον χρυσουν εν τη χειρι αυτης γεμον βδελυγματων και τα ακαθαρτα της πορνειας αυτης [gnt]

In some, but not all, cases the inflections in Greek can help disambiguate a parse.

Concrete syntax:

precious stones and pearls

Abstract syntax:

1. (precious stones) and pearls 2. precious (stones and pearls) 2. (precious stones) and (precious pearls)

In this case, masculine "stone" is singular but translated as plural in the KJV (King James Version). The word "pearls" is masculine and plural. Thus, the first abstract syntax fits better.

What is the one "valuable stone" or "honored stone" (singular)?
What are the "pearls"? Why are they not described as "valuable"?

Revelation 17:4 And the woman was arrayed in purple and scarlet colour, and decked with gold and precious stones and pearls, having a golden cup in her hand full of abominations and filthiness of her fornication: [kjv]

και η γυνη ην περιβεβλημενη πορφυρουν και κοκκινον και κεχρυσωμενη χρυσιω και λιθω τιμιω και μαργαριταις εχουσα ποτηριον χρυσουν εν τη χειρι αυτης γεμον βδελυγματων και τα ακαθαρτα της πορνειας αυτης [gnt]

Revelation 17:4
Strong	Used	Unique	Word	Part of speech
G2532	8954	8954	και	conjunction
G3588	19706	981	η	article nominative singular feminine
G1135	219	76	γυνη	noun nominative singular feminine
G1510	2435	313	ην	verb imperfect active indicative 3rd person singular
G4016	23	3	περιβεβλημενη	verb perfect passive participle nominative singular feminine
G4210	4	4	πορφυρουν	adjective accusative singular neuter
G2532	8954	8954	και	conjunction
G2847	6	3	κοκκινον	adjective accusative singular neuter
G2532	8954	8954	και	conjunction
G5558	2	2	κεχρυσωμενη	verb perfect passive participle nominative singular feminine
G5557	12	6	χρυσω	noun dative singular masculine
G2532	8954	8954	και	conjunction
G3037	60	10	λιθω	noun dative singular masculine
G5093	13	4	τιμιω	adjective dative singular masculine
G2532	8954	8954	και	conjunction
G3135	9	2	μαργαριταις	noun dative plural masculine
G2192	702	15	εχουσα	verb present active participle nominative singular feminine
G4221	31	24	ποτηριον	noun accusative singular neuter
G5552	18	6	χρυσουν	adjective accusative singular neuter
G1722	2735	2735	εν	preposition
G3588	19706	878	τη	article dative singular feminine
G5495	175	20	χειρι	noun dative singular feminine
G846	5547	166	αυτης	personal pronoun genitive singular feminine
G1073	11	1	γεμων	verb present active participle nominative singular masculine
G946	6	2	βδελυγματων	noun genitive plural neuter
G2532	8954	8954	και	conjunction
G3588	19706	826	τα	article accusative plural neuter
G169	31	6	ακαθαρτα	adjective accusative plural neuter
G3588	19706	1291	της	article genitive singular feminine
G4202	24	11	πορνειας	noun genitive singular feminine
G846	5547	166	αυτης	personal pronoun genitive singular feminine

Revelation 17:4

All

KJV: And the woman was arrayed in purple and scarlet colour, and decked with gold and precious stones and pearls, having a golden cup in her hand full of abominations and filthiness of her fornication:

Greek: και η γυνη η ην περιβεβλημενη πορφυρα πορφυρουν και κοκκινω κοκκινον και κεχρυσωμενη χρυσω χρυσιω και λιθω τιμιω και μαργαριταις εχουσα χρυσουν ποτηριον χρυσουν εν τη χειρι αυτης γεμον βδελυγματων και ακαθαρτητος τα ακαθαρτα της πορνειας αυτης

Titus 3:3 For we ourselves also were sometimes foolish, disobedient, deceived, serving divers lusts and pleasures, living in malice and envy, hateful, and hating one another. [kjv]

ημεν γαρ ποτε και ημεις ανοητοι απειθεις πλανωμενοι δουλευοντες επιθυμιαις και ηδοναις ποικιλαις εν κακια και φθονω διαγοντες στυγητοι μισουντες αλληλους [gnt]

The Greek does not appear to match the KJV translation.

KJV: ... serving various desires and pleasures.
Greek: ... serving desires and various pleasures

That is, the "divers" or "varied" appears to go with "pleasures" and not "lusts" or "desires". Another way to parse the sentence fragment is (by distribution) as follows.

Parse: ... serving various desires and various pleasures.

Good food and drink. What is wrong with the drink?
Good food and good drink.

More: The Christian Hedonism fallacy