Verse statistics¶

VerseStats(source, seseo=False)

The verse statistics of a text: the scansion of the lines, the meter and the number of metrical syllables, the rhythmic stresses and the stress profile, the types of the endecasílabo, the endings of the lines, the rhyme, its scheme and the strophes. The data source is a text with its line breaks or a Doc object of the spaCy library; the statistics are computed when the object is created, and no trained model and no dictionary are needed.

The meter of Spanish verse is syllabic: a line is measured by its metrical syllables, and the stresses shape its rhythm. The text is split into lines and stanzas (by blank lines), the words are split into syllables and stressed by syllabify and word_stresses, and the syllables of a line are counted with the synalephas and the law of the final stress; a compound verse, the alejandrino above all, is read as two hemistichs (algorithm). A text with letters but without Spanish syllables (another alphabet) gives empty statistics - n_lines 0, meter None, shares nan.

Raises:

SourceTypeError - a source that is neither a string nor a Doc;
SourceError - a source without letters (empty, digits and punctuation only);
ParameterError - a seseo that is not a bool.

CodeResult

from ests import VerseStats

# The beginning of the sonnet I by Garcilaso de la Vega
text = """Cuando me paro a contemplar mi estado
y a ver los pasos por do me han traído,
hallo, según por do anduve perdido,
que a mayor mal pudiera haber llegado."""

vs = VerseStats(text)
vs.print_stats()

                  Statistic                  |    Value
------------------------------------------------------------
Number of lines                              |      4
Number of stanzas                            |      1
Meter                                        | endecasílabo
Number of metrical syllables                 |      11
Share of lines off the meter                 |     0.00
Share of lines without the rhythmic stresses |     0.00
Share of rhymed lines                        |     1.00
Share of oxytone endings (aguda)             |     0.00
Share of paroxytone endings (llana)          |     1.00
Share of proparoxytone endings (esdrújula)   |     0.00

Parameters¶

Parameter	Type	Default	Description
`source`	`str` / `Doc`	-	Data source
`seseo`	`bool`	`False`	Pronounce `c` and `z` before `e` and `i` as `s` in the rhyme, as in Andalusia and in America: `voz` - `dos`

Attributes¶

Attribute	Type	Description
`lines`	`tuple[str]`	Lines with Spanish words
`stanzas`	`tuple[tuple[str, ...], ...]`	Lines by stanza
`syllables`	`tuple[tuple[str, ...], ...]`	Syllables of every line as the verse reads them, a synalepha marked with `‿`
`stresses`	`tuple[tuple[int, ...], ...]`	Indices of the stressed syllables of every line in `syllables`, from zero
`caesuras`	`tuple[int / None, ...]`	Index of the first syllable of the second hemistich of every line in `syllables`, `None` for a simple verse
`patterns`	`tuple[str, ...]`	Patterns of the lines of `+` (a stressed metrical syllable) and `-` (an unstressed one), as long as the line in metrical syllables
`mean_line_len`	`float`	Mean length of a line in metrical syllables
`c_stressed_vowels`	`dict[str, int]`	Distribution of the stressed vowels
`seseo`	`bool`	Whether the rhyme is read with seseo
`c_feet`	`dict[int, int]`	Distribution of the lines by metrical syllables (meter)
`stress_profile`	`tuple[float, ...]`	Share of the stressed lines of the meter by metrical syllable (meter)
`c_rhythms`	`dict[str, int]`	Distribution of the endecasílabos by type (rhythmic stresses)
`rhyme_schemes`	`tuple[str, ...]`	Rhyme schemes of the stanzas (rhyme)
`c_rhymes`	`dict[str, int]`	Distribution of the rhymed lines by the kind of the rhyme: `consonante`, `asonante` (rhyme)
`strophes`	`tuple[str / None, ...]`	Names of the stanzas, `VERSE_STROPHES` (strophes)
`form`	`str / None`	Form of the poem: `soneto`, `romance`, the strophe of all its stanzas, `silva` (strophes)
`c_clausulas`	`dict[str, int]`	Distribution of the endings of the lines by type: `aguda`, `llana`, `esdrújula`, `sobresdrújula` (endings)

stresses indexes syllables, and the two agree with patterns up to the last stress of a simple verse; in a compound verse each hemistich takes its own length, so after an aguda or an esdrújula at the caesura patterns and syllables part, and caesuras tells where the second hemistich begins.

CodeResult

vs.syllables[1]
vs.patterns
vs.stresses[0]
vs.c_stressed_vowels

('y‿a', 'ver', 'los', 'pa', 'sos', 'por', 'do', 'me‿han', 'tra', 'í', 'do')
('---+---+-+-', '-+-+---+-+-', '+--+--+--+-', '--++-+-+-+-')
(3, 7, 9)
{'a': 8, 'e': 3, 'i': 2, 'u': 2, 'o': 1}

Metrics¶

Every metric is an attribute of the object, computed when it is created; get_stats gives the metrics of the table, while the distributions, the schemes and the names of the cards are attributes only.

Metric	Measures	Range	Higher means
`n_lines`, `n_stanzas`	Number of the lines with Spanish words and of the stanzas	0 and above	A longer text
`meter`	Syllabic meter: the name of the verse by its metrical syllables	`bisílabo` to `octodecasílabo` (`VERSE_METERS`), `None` without a meter	-
`n_feet`	Metrical syllables of the meter	2-18, `None` without a meter	Longer lines
`p_deviations`	Share of the lines off the meter	0-0.1, `nan` without a meter	More lines off the meter
`p_pyrrhics`	Share of the lines of the meter without its rhythmic stresses	0-1, `nan` without a meter	More lines without the rhythmic stresses
`p_rhymed`	Share of the rhymed lines	0-1, `nan` without lines	More rhymes
`p_masculine`, `p_feminine`, `p_dactylic`	Shares of the endings aguda, llana and esdrújula	0-1, `nan` without lines	More endings of the type

Meter and metrical syllables¶

meter · n_feet · c_feet · p_deviations · Quilis, 1984

detect_meter(text)

The meter of a poem by the algorithm: the name of the verse by its metrical syllables - octosílabo, endecasílabo, alejandrino... (VERSE_METERS), and n_feet its number of metrical syllables. c_feet is the distribution of the lines by metrical syllables, and p_deviations the share of the lines off the meter.

The meter is the length of most lines, and it is not determined (None) if more than a tenth of the lines (VERSE_MAX_DEVIATIONS) stay off it after the fitting - a polymetric poem (a silva, a sonnet with an estrambote of heptasílabos), free verse, prose - or if the text has a single line (VERSE_MIN_LINES). The lines of a poem without a meter are fitted to its common lengths - the ones of at least two lines and a tenth of them, as the 7 and the 11 syllables of a lira or a silva - if these take more than half of the lines; otherwise, as in free verse and prose, the lines keep their plain readings.

Parameter	Type	Default	Description
`text`	`str`	-	Text of a poem

The errors are those of VerseStats.

CodeResult

from ests.verse_stats import detect_meter

# The beginning of the Romance del prisionero
detect_meter(
    "Que por mayo era, por mayo,\ncuando hace la calor,\ncuando los trigos encañan\ny están los campos en flor,"
)
# The sonnet of Garcilaso at the top of the page
vs.c_feet, vs.p_deviations

'octosílabo'
({11: 4}, 0.0)

A compound verse is read by its hemistichs, each by the law of the final stress: in the alejandrino of 7 + 7 syllables La princesa está pálida (an esdrújula) and que ha perdido el color (an aguda) count 7 each.

CodeResult

# The first stanza of Sonatina by Rubén Darío (Prosas profanas, 1896)
sonatina = (
    "La princesa está triste... ¿qué tendrá la princesa?\n"
    "Los suspiros se escapan de su boca de fresa,\n"
    "que ha perdido la risa, que ha perdido el color.\n"
    "La princesa está pálida en su silla de oro,\n"
    "está mudo el teclado de su clave sonoro;\n"
    "y en un vaso olvidada se desmaya una flor."
)
vs_sonatina = VerseStats(sonatina)
vs_sonatina.meter, vs_sonatina.n_feet
vs_sonatina.syllables[3]
vs_sonatina.patterns[2], vs_sonatina.patterns[3]
vs_sonatina.stresses[3], vs_sonatina.caesuras

('alejandrino', 14)
('la', 'prin', 'ce', 'sa‿es', 'tá', 'pá', 'li', 'da', 'en', 'su', 'si', 'lla', 'de', 'o', 'ro')
('+-+--+-+-+--+-', '--+-++---+--+-')
((2, 4, 5, 10, 13), (7, 7, 7, 8, 7, 7))

Rhythmic stresses¶

p_pyrrhics · stress_profile · c_rhythms · Navarro Tomás, 1972

The rhythmic stresses (VERSE_RHYTHMS) are the ones a meter asks for besides the last stress, which every line has by the law of the final stress: the endecasílabo is stressed on the 6th syllable (a maiore) or on the 4th and the 8th (sáfico) or the 7th (dactílico); a compound verse on the last stress of its first hemistich. p_pyrrhics is the share of the lines of the meter without them, and a meter with the last stress alone, the octosílabo among others, has p_pyrrhics 0.

\[ \frac{\textrm{Number of lines of the meter without its rhythmic stresses}}{\textrm{Number of lines of the meter}} \]

stress_profile is the share of the lines of the meter stressed on each metrical syllable; the syllables are counted from zero, so the 6th syllable is the number 5. Without a meter p_pyrrhics is nan and the profile is empty. c_rhythms is the distribution of the endecasílabos by type:

Type	Stresses	Name
`1-6-10`	1st, 6th	enfático
`2-6-10`	2nd, 6th	heroico
`3-6-10`	3rd, 6th	melódico
`4-6-10`, `5-6-10`	the first stress on the 4th or the 5th, 6th	a maiore
`6-10`	6th, no stress before it	a maiore
`4-8-10`	4th, 8th, no 6th	sáfico
`4-7-10`	4th, 7th, no 6th nor 8th	dactílico, de gaita gallega

A line with the 6th syllable stressed takes its type from its first stress; 4-6-10 counts as sáfico in some treatises. A line with none of the three sets is left out of c_rhythms and counted in p_pyrrhics.

CodeResult

vs.p_pyrrhics, vs.stress_profile
vs.c_rhythms

(0.0, (0.25, 0.25, 0.25, 1.0, 0.0, 0.25, 0.25, 0.75, 0.0, 1.0, 0.0))
{'4-8-10': 2, '4-7-10': 1, '3-6-10': 1}

Rhyme¶

p_rhymed · rhyme_schemes · c_rhymes

rhyme_scheme(text, seseo=False)

The rhyme scheme of a poem: the schemes of the stanzas separated by spaces. The object has the schemes by stanza in rhyme_schemes, the share of the rhymed lines in p_rhymed and the distribution of the rhymed lines by the kind of the rhyme, consonante or asonante, in c_rhymes.

A line rhymes by its ending from the last stressed vowel: in full (consonante) if the sounds are the same, by assonance (asonante) if the stressed and the last vowels are. The sounds are the ones of the ear: b and v, c and z (s with seseo), ll and y are one sound, h is silent, the tap r and the trill rr differ. The assonance leaves out the vowels between the stressed and the last one (pálido - a, o), takes the stressed vowel of a diphthong (cielo - e, o; gracia - a, a), reads a final i and u as e and o (fácil - a, e) and gives an aguda its stressed vowel alone. A line rhymes with the lines of the window of RHYME_WINDOW (4) lines before it, across the stanzas, so that the quatrains of a sonnet rhyme ABBA ABBA.

The assonance counts in two cases. A poem whose rhyme is the assonance - at least a third of its lines with no full rhyme, and at least four of them, share one (RHYME_MIN_ASSONANCE, RHYME_MIN_ASSONANT_LINES) - rhymes by the assonances of all its lines, which take in the full rhymes of the same vowels, as an assonant sonnet does. Otherwise the assonance counts on alternate lines with unrhymed lines between them: a run of at least four lines two apart with three in four of the lines between them free of a full rhyme (RHYME_MIN_FREE), or all the even lines of a stanza of an even number of lines, as the second and the fourth line of a copla. So a romance rhymes even when its assonance changes from part to part or its assonant lines also rhyme in full two by two. An assonance of neighbouring lines is no rhyme: in a sonnet of -ado and -ano the two groups stay apart.

The rhyme scheme follows the Spanish usage: the letters go in the order of the rhyme groups of the whole poem, upper-case for the lines of arte mayor (9 syllables and more, VERSE_ARTE_MAYOR) and lower-case for the lines of arte menor, a hyphen for an unrhymed line: a sonnet is ABBA ABBA CDC DCD, a redondilla abba, a lira aBabB, a romance -a-a-a-a. When the 26 letters run out, the letter of a group that no line in the window can join any more is taken again.

Parameter	Type	Default	Description
`text`	`str`	-	Text of a poem
`seseo`	`bool`	`False`	Pronounce `c` and `z` before `e` and `i` as `s` in the rhyme

The errors are those of VerseStats.

CodeResult

from ests.verse_stats import rhyme_scheme

# Sor Juana Inés de la Cruz
rhyme_scheme(
    "Hombres necios que acusáis\na la mujer sin razón,\nsin ver que sois la ocasión\nde lo mismo que culpáis."
)
# The sonnet of Garcilaso at the top of the page
vs.rhyme_schemes, vs.p_rhymed, vs.c_rhymes

# The beginning of La tierra de Alvargonzález by Antonio Machado
romance = """Siendo mozo Alvargonzález,
dueño de mediana hacienda,
que en otras tierras se dice
bienestar y aquí, opulencia,
en la feria de Berlanga
prendóse de una doncella,
y la tomó por mujer
al año de conocerla."""
rhyme_scheme(romance), VerseStats(romance).c_rhymes

'abba'
(('ABBA',), 1.0, {'consonante': 4})
('-a-a-a-a', {'asonante': 4})

Strophes and form¶

strophes · form

A stanza is named by its lines, their lengths, its scheme and its rhyme (VERSE_STROPHES, the first that fits), None if no strophe fits:

Strophe	Lines	Lengths	Schemes	Rhyme
pareado	2	any	aa	any
terceto	3	arte mayor	aba, a-a, abc, aab, abb	any
tercerilla	3	arte menor	aba, a-a	any
cuaderna vía	4	14 each	aaaa	consonante
cuarteto, serventesio	4	arte mayor	abba; abab	consonante
redondilla, cuarteta	4	arte menor	abba; abab	consonante
seguidilla	4	7-5-7-5	-a-a	any
copla	4	8 each	-a-a	asonante
lira	5	7-11-7-7-11	ababb	consonante
quinteto, quintilla	5	arte mayor; arte menor	ababa, abaab, abbab, aabab, aabba	consonante
estrofa manriqueña	6	8-8-4-8-8-4	abcabc	consonante
sexta rima	6	arte mayor	ababcc	consonante
sexteto	6	arte mayor	aabccb, abcabc, abbacc, aabbcc	consonante
sextilla	6	arte menor	aabccb, ababcc, abcabc, aabaab	consonante
octava real	8	11 each	abababcc	consonante
décima	10	8 each	abbaaccddc	consonante

A stanza of a poem with no meter is fitted to the syllables of a strophe that asks for them, so that a single lira gets its 7 and 11 syllables. The form of a poem is soneto for 14 lines of a meter of arte mayor in full rhyme with the quatrains ABBA or ABAB and the tercets on rhymes of their own; romance for octosílabos from 8 lines with three in four of the even lines of the stanzas rhymed by assonance, which may change, and three in four of the odd ones unrhymed; the strophe of all the stanzas; silva for heptasílabos and endecasílabos with no meter, longer than a strophe (more than 10 lines) and with no pattern of lengths repeated by all its stanzas; otherwise None.

CodeResult

# The sonnet I by Garcilaso de la Vega
sonnet = """Cuando me paro a contemplar mi estado
y a ver los pasos por do me han traído,
hallo, según por do anduve perdido,
que a mayor mal pudiera haber llegado;

mas cuando del camino estó olvidado,
a tanto mal no sé por dó he venido;
sé que me acabo, y más he yo sentido
ver acabar comigo mi cuidado.

Yo acabaré, que me entregué sin arte
a quien sabrá perderme y acabarme
si quisiere, y aún sabrá querello;

que pues mi voluntad puede matarme,
la suya, que no es tanto de mi parte,
pudiendo, ¿qué hará sino hacello?"""
vs_sonnet = VerseStats(sonnet)
vs_sonnet.rhyme_schemes, vs_sonnet.strophes, vs_sonnet.form
vs_sonatina.strophes, VerseStats(romance).form

(('ABBA', 'ABBA', 'CDE', 'DCE'), ('cuarteto', 'cuarteto', 'terceto', 'terceto'), 'soneto')
(('sexteto',), 'romance')

Line endings¶

p_masculine · p_feminine · p_dactylic · c_clausulas

The type of the ending (cláusula) of a line by the syllables after its last stress: aguda - the stress on the last syllable, llana - one syllable after the stress, esdrújula - two, sobresdrújula - three or more (VERSE_CLAUSULAS, as stress_type of a word). c_clausulas is the distribution of the lines by type, and the shares are taken over all the lines: p_masculine of the agudas, p_feminine of the llanas and p_dactylic of the esdrújulas.

CodeResult

vs.c_clausulas, vs_sonatina.c_clausulas
vs_sonatina.p_masculine, vs_sonatina.p_feminine

({'llana': 4}, {'aguda': 2, 'llana': 4})
(0.3333333333333333, 0.6666666666666666)

Methods¶

accentuate() returns the text with the stresses marked: an acute accent (U+0301) is put after the stressed vowel of every stressed word that has no written accent; the words with a written accent keep it, and the unstressed words of the verse (VERSE_PROCLITICS) stay unmarked, except the last word of a line. The marks are combining characters apart from the written accents, so unicodedata.normalize("NFC", ...) turns them into accented letters. Only the lines with Spanish words are returned (as in lines), the stanzas separated by a blank line. Unlike the accentuate function, the last word of a line is always stressed.

CodeResult

print(vs.accentuate())

Cuando me páro a contemplár mi estádo
y a vér los pásos por do me hán traído,
hállo, según por do andúve perdído,
que a mayór mál pudiéra habér llegádo.

The methods get_stats, to_dataframe, print_stats and describe are those of the statistics interface of the anyTS core: get_stats() returns the metrics as a dictionary, and print_stats() prints them as a table with their descriptions (VERSE_STATS_DESC), as in the example at the top of the page. The verse statistics have no scales, so describe gives None for the statistics of get_stats.

CodeResult

vs.get_stats()

{'n_lines': 4,
 'n_stanzas': 1,
 'meter': 'endecasílabo',
 'n_feet': 11,
 'p_deviations': 0.0,
 'p_pyrrhics': 0.0,
 'p_rhymed': 1.0,
 'p_masculine': 0.0,
 'p_feminine': 1.0,
 'p_dactylic': 0.0}

Algorithm¶

A line of Spanish verse is measured by its metrical syllables, which are not the syllables of its words one by one:

Syllables and stresses. Every word is split into syllables by syllabify and stressed by the orthographic rules of word_stresses; an adverb in -mente has two stresses. The unstressed words of the verse (VERSE_PROCLITICS) take no stress: the articles, the prepositions except según, the conjunctions, the relatives (que, cual, donde, como, cuanto), the clitic pronouns (me, te, se, le, nos), the possessives before a noun (mi, tu, su, nuestro, vuestro), the titles before a name (don, fray, san), tan, aun and the interjections oh, ay, ah. The last word of a line is always stressed.
The plain reading. The final vowel of a word and the first one of the next make one syllable - a synalepha (cuan-do‿a-pe-nas, tie-rra‿y‿a-gua). A silent h lets it through (oh‿al-ma), while hi and hu before a vowel (hierba, hueso) and y before a vowel (ya, yo) are consonants and stop it; a consonant at the end of a word stops it too (el | al-ma).
The law of the final stress. A line counts up to its last stressed syllable and one syllable more: a line that ends in an aguda gets a syllable (cuan-do‿ha-ce-la-ca-lor - 6 + 1 = 7), a line that ends in an esdrújula loses one (el pá-ja-ro - 3).
The meter. The meter of a poem is the length of most plain readings; on a tie from three lines up, the tied length with a name that leaves the fewest lines off it wins. Every line is fitted to it with the fewest changes of its plain reading: a longer line joins two vowels of a hiatus in a word - a synaeresis (poe-ta), the ones with an accented i or u (dí-a) last; a shorter line breaks its synalephas from the end of the line - a hiatus, then splits a diphthong of a stressed syllable - a dieresis (su-a-ve, ru-i-do). A written diaeresis marks a hiatus (sü-a-ve) that no synaeresis joins. A line that cannot reach the meter keeps its plain reading and counts as off it. A poem without a meter fits its lines to its common lengths instead, if they take more than half of the lines.
Compound verse. A compound verse of two equal hemistichs (VERSE_HEMISTICHS: 5 + 5, 6 + 6, 7 + 7, 8 + 8, 9 + 9) is tried when more than half of the plain readings split into its hemistichs and its length equals the one of most lines or is one syllable off it; the reading with fewer lines off the meter wins, the compound verse on a tie. The caesura falls after a stressed word and blocks the synalepha, and each hemistich follows the law of the final stress on its own: the alejandrino La princesa está pálida | en su silla de oro is 7 + 7.

On the 60,209 lines of the 4,259 sonnets of SpanishSonnets, the length of a line agrees with the automatic scansion of DISCO in 97.0% of the lines and with rantanplan in 97.7%, and the stress of a metrical syllable of the lines of equal length in 97.4% and 99.6%.

The rhyme is described in its card. Against the automatic rhyme of DISCO (RhymeTagger), 99.1% of the pairs of rhyming lines we find are its pairs, and we find 97.9% of its pairs.

Accentuation¶

accentuate(text)

Marks the stresses of a text by the orthographic rules: an acute accent (U+0301) is put after the stressed vowel of every word without a written accent, both stresses of an adverb in -mente and of a hyphenated compound; the unstressed words of the verse (VERSE_PROCLITICS) stay unmarked. The text is normalized to NFC. For the lines of a poem, where the last word is always stressed, the VerseStats.accentuate method serves.

Parameter	Type	Default	Description
`text`	`str`	-	Text

CodeResult

import unicodedata

from ests.verse_stats import accentuate

text = "Yo soy aquel que ayer no más decía el verso azul y la canción profana"
unicodedata.normalize("NFC", accentuate(text))

'Yó sóy aquél que ayér nó más decía el vérso azúl y la canción profána'

Stanzas¶

split_stanzas(text)

Splits a text into stanzas and lines: the stanzas are separated by blank lines, and the lines without a Spanish syllable (numbers, asterisks, other alphabets) and the roman numerals of the parts of a poem (II, IV.) are left out. The text is normalized to NFC.

Parameter	Type	Default	Description
`text`	`str`	-	Text of a poem

CodeResult

from ests.verse_stats import split_stanzas

split_stanzas(
    "Cuando me paro a contemplar mi estado\n"
    "y a ver los pasos por do me han traído,\n\n* * *\n\n"
    "hallo, según por do anduve perdido,"
)

[['Cuando me paro a contemplar mi estado', 'y a ver los pasos por do me han traído,'],
 ['hallo, según por do anduve perdido,']]