data-trove-world-languages-endangerment-silence

Overview

Source: /home/coolhand/html/datavis/data_trove/data/quirky/silence_data.json

Saturn profiled 6,998 rows across 6 columns. The stats below are deterministic and machine-readable; the prose is a language-model interpretation of those stats (opt-in, added after the fact, never sees raw rows).

[2]:

!pip install saturn-dissect
import subprocess
subprocess.run([
    "saturn", "analyze", "/home/coolhand/html/datavis/data_trove/data/quirky/silence_data.json",
    "--findings", "data-trove-world-languages-endangerment-silence.json",
    "--llm", "anthropic:default",
])

Summary confidence: high

This dataset catalogues approximately 7,000 world languages, each with a name, geographic coordinates, speaker population, and endangerment status. The most striking finding is the extreme inequality in speaker populations: the median language has only 11,000 speakers while the maximum reaches nearly 1 billion, with over 16% of languages flagged as outliers — a classic long-tail distribution reflecting how a handful of dominant languages vastly outnumber the rest. Equally notable is the endangerment picture: while 44% of languages are classified as 'safe', a substantial share face real risk — 1,753 are 'definitely endangered', 327 are 'critically endangered', and 219 are already extinct. Top words in language names include 'sign', 'zapotec', 'mixtec', and directional qualifiers like 'southern' and 'northern', hinting at rich dialect clustering worth exploring geographically.

citing: row_count · column_count · stats.median · stats.max · stats.outlier_rate · top_values · top_words · stats.zero_rate

Out[4]:

saturn.schema() · 6 columns

column	kind	n	null%	unique	alerts
n	text	6,998	0.0%	6,998	near_unique one_word
lat	numeric	6,998	0.0%	4,048
lng	numeric	6,998	0.0%	5,560
p	numeric	6,998	0.0%	1,627	high_skew outliers
s	numeric	6,998	0.0%	6
ss	categorical	6,998	0.0%	7

Fig 1.

ss · Look at how the 'safe' majority compares to the combined endangered and extinct categories — nearly half of all languages are at some level of risk.

Show data table

Top values for ss (7 unique shown, of 7 total).
value	count	share
safe	3074	43.9%
definitely endangered	1753	25.1%
vulnerable	1160	16.6%
severely endangered	374	5.3%
critically endangered	327	4.7%
extinct	219	3.1%
unknown	91	1.3%

Fig 2.

p · Expect a heavily right-skewed distribution; the vast majority of languages cluster near zero speakers while a tiny number reach hundreds of millions.

Show data table

Histogram bins for p (median: 11000.0).
bin	count
0 – 2.411e+07	6943
2.411e+07 – 4.823e+07	28
4.823e+07 – 7.234e+07	5
7.234e+07 – 9.646e+07	12
9.646e+07 – 1.206e+08	0
1.206e+08 – 1.447e+08	3
1.447e+08 – 1.688e+08	1
1.688e+08 – 1.929e+08	0
1.929e+08 – 2.17e+08	0
2.17e+08 – 2.411e+08	2
2.411e+08 – 2.653e+08	0
2.653e+08 – 2.894e+08	0
2.894e+08 – 3.135e+08	0
3.135e+08 – 3.376e+08	0
3.376e+08 – 3.617e+08	0
3.617e+08 – 3.858e+08	1
3.858e+08 – 4.099e+08	0
4.099e+08 – 4.34e+08	0
4.34e+08 – 4.582e+08	1
4.582e+08 – 4.823e+08	0
4.823e+08 – 5.064e+08	0
5.064e+08 – 5.305e+08	0
5.305e+08 – 5.546e+08	0
5.546e+08 – 5.787e+08	0
5.787e+08 – 6.028e+08	0
6.028e+08 – 6.27e+08	0
6.27e+08 – 6.511e+08	0
6.511e+08 – 6.752e+08	0
6.752e+08 – 6.993e+08	0
6.993e+08 – 7.234e+08	1
7.234e+08 – 7.475e+08	0
7.475e+08 – 7.716e+08	0
7.716e+08 – 7.958e+08	0
7.958e+08 – 8.199e+08	0
8.199e+08 – 8.44e+08	0
8.44e+08 – 8.681e+08	0
8.681e+08 – 8.922e+08	0
8.922e+08 – 9.163e+08	0
9.163e+08 – 9.404e+08	0
9.404e+08 – 9.646e+08	1

Fig 3.

s · This numeric status score (0–5) shows how languages rank on a vitality scale — look for the concentration at the higher (safer) end versus the tail of critically low scores.

Show data table

Histogram bins for s (median: 4.0).
bin	count
0 – 0.125	250
0.125 – 0.25	0
0.25 – 0.375	0
0.375 – 0.5	0
0.5 – 0.625	0
0.625 – 0.75	0
0.75 – 0.875	0
0.875 – 1	0
1 – 1.125	328
1.125 – 1.25	0
1.25 – 1.375	0
1.375 – 1.5	0
1.5 – 1.625	0
1.625 – 1.75	0
1.75 – 1.875	0
1.875 – 2	0
2 – 2.125	383
2.125 – 2.25	0
2.25 – 2.375	0
2.375 – 2.5	0
2.5 – 2.625	0
2.625 – 2.75	0
2.75 – 2.875	0
2.875 – 3	0
3 – 3.125	1768
3.125 – 3.25	0
3.25 – 3.375	0
3.375 – 3.5	0
3.5 – 3.625	0
3.625 – 3.75	0
3.75 – 3.875	0
3.875 – 4	0
4 – 4.125	1176
4.125 – 4.25	0
4.25 – 4.375	0
4.375 – 4.5	0
4.5 – 4.625	0
4.625 – 4.75	0
4.75 – 4.875	0
4.875 – 5	3093

Fig 4.

lat · The distribution of latitudes skews toward tropical and equatorial regions, revealing where the world's linguistic diversity is geographically concentrated.

Show data table

Histogram bins for lat (median: 6.37).
bin	count
-55.27 – -52.06	1
-52.06 – -48.85	1
-48.85 – -45.64	1
-45.64 – -42.43	0
-42.43 – -39.22	2
-39.22 – -36.01	5
-36.01 – -32.8	9
-32.8 – -29.59	13
-29.59 – -26.38	23
-26.38 – -23.17	48
-23.17 – -19.96	100
-19.96 – -16.75	101
-16.75 – -13.54	217
-13.54 – -10.33	202
-10.33 – -7.116	461
-7.116 – -3.906	749
-3.906 – -0.6958	640
-0.6958 – 2.514	344
2.514 – 5.725	434
5.725 – 8.935	609
8.935 – 12.15	676
12.15 – 15.36	264
15.36 – 18.57	368
18.57 – 21.78	210
21.78 – 24.99	282
24.99 – 28.2	320
28.2 – 31.41	131
31.41 – 34.62	124
34.62 – 37.83	148
37.83 – 41.04	81
41.04 – 44.25	107
44.25 – 47.46	65
47.46 – 50.67	68
50.67 – 53.88	68
53.88 – 57.09	35
57.09 – 60.3	20
60.3 – 63.51	34
63.51 – 66.72	18
66.72 – 69.93	13
69.93 – 73.14	6

Fig 5.

n · Most language names are short single words, but a long tail of compound names (up to 43 characters) reflects complex regional dialect naming conventions.

Show data table

Character-length distribution for n (mean: 8.971991997713632).
chars	count
1 – 2	25
2 – 3	193
3 – 4	748
4 – 5	1095
5 – 6	1082
6 – 7	833
7 – 8	521
8 – 9	333
9 – 10	271
10 – 12	231
12 – 13	208
13 – 14	213
14 – 15	188
15 – 16	199
16 – 17	144
17 – 18	91
18 – 19	75
19 – 20	68
20 – 21	63
21 – 22	69
22 – 23	150
23 – 24	48
24 – 25	26
25 – 26	30
26 – 27	21
27 – 28	8
28 – 29	7
29 – 30	10
30 – 31	14
31 – 32	6
32 – 34	8
34 – 35	2
35 – 36	10
36 – 37	1
37 – 38	4
38 – 39	0
39 – 40	1
40 – 41	0
41 – 42	1
42 – 43	1

Fig 6.

Per-column null rate across the corpus. Columns are ordered by input position.

Show data table

Per-column null rate across the corpus.
column	kind	null %
n	text	0.0%
lat	numeric	0.0%
lng	numeric	0.0%
p	numeric	0.0%
s	numeric	0.0%
ss	categorical	0.0%

Fig 7.

Pearson correlation across numeric columns (sampled, bounded).

Show data table

Pearson correlation across 4 numeric columns (values clipped to 2 decimals).
	lat	lng	p	s
lat	+1.00	-0.35	+0.13	-0.09
lng	-0.35	+1.00	+0.00	+0.16
p	+0.13	+0.00	+1.00	+0.09
s	-0.09	+0.16	+0.09	+1.00

n text label

This column appears to be a name field for human languages or dialects — top words include 'language', 'sign', 'Zapotec', 'Mixtec', and directional qualifiers like 'southern', 'northern', 'western', 'eastern', 'central', consistent with a linguistic taxonomy dataset. All 6,998 rows are unique with zero duplicates and zero nulls, confirming it functions as a label or identifier. The majority of values (73%) are single words, but multi-word entries push the mean length to ~8.97 characters and mean word count to ~1.37, reflecting compound names like 'Southern Zapotec'. The vocabulary size (7,003) slightly exceeding unique row count (6,998) is unremarkable given tokenization.

Treatment: Use as a human-readable label; encode as a categorical ID or embed via a language-name lookup for modelling.

anthropic:default · confidence high

Out[13]:

saturn.columns["n"].stats

stat	value
n	6,998
nulls	0 (0.0%)
unique	6,998
len_min	1
len_max	43
len_mean	8.972
len_median	7
len_p95	21
word_mean	1.369
word_median	1
n_empty	0
n_duplicates	0
duplicate_rate	0
vocab_size	7,003
readability_flesch_mean	57.13
emoji_rate	0
url_rate	0
one_word_rate	0.7314
allcaps_rate	0
boilerplate_rate	0
alert: near_unique	100.0% of rows are unique strings
alert: one_word	73.1% rows are a single word

Fig 8.

Character-length distribution for n.

Show data table

Character-length distribution for n (mean: 8.971991997713632).
chars	count
1 – 2	25
2 – 3	193
3 – 4	748
4 – 5	1095
5 – 6	1082
6 – 7	833
7 – 8	521
8 – 9	333
9 – 10	271
10 – 12	231
12 – 13	208
13 – 14	213
14 – 15	188
15 – 16	199
16 – 17	144
17 – 18	91
18 – 19	75
19 – 20	68
20 – 21	63
21 – 22	69
22 – 23	150
23 – 24	48
24 – 25	26
25 – 26	30
26 – 27	21
27 – 28	8
28 – 29	7
29 – 30	10
30 – 31	14
31 – 32	6
32 – 34	8
34 – 35	2
35 – 36	10
36 – 37	1
37 – 38	4
38 – 39	0
39 – 40	1
40 – 41	0
41 – 42	1
42 – 43	1

lat numeric feature

This column contains geographic latitude values, spanning from -55.27° (southern South America) to 73.14° (Arctic latitudes), covering a wide swath of the globe with concentration in tropical and subtropical regions (median 6.37°, Q1 -4.65°, Q3 18.29°). The 4,048 unique values out of 6,998 rows indicates coordinate granularity likely at 2 decimal places, with some location repetition. Mild positive skew (0.697) and near-mesokurtic kurtosis (0.477) confirm most records cluster in equatorial-to-subtropical bands, consistent with datasets heavy in African, South/Southeast Asian, or Latin American records. Only 149 outliers (~2.1%) were flagged, likely corresponding to high-latitude locations in Europe or North America.

Treatment: Pair with a longitude column for geospatial analysis; consider binning into regions or using as-is in spatial models — do not normalize without care for geographic interpretation.

anthropic:default · confidence high

Out[16]:

saturn.columns["lat"].stats

stat	value
n	6,998
nulls	0 (0.0%)
unique	4,048
min	-55.27
max	73.14
mean	8.437
median	6.37
std	18
q1	-4.65
q3	18.29
iqr	22.94
skew	0.6975
kurtosis	0.4773
n_outliers	149
outlier_rate	0.02129
zero_rate	0

Fig 9.

Distribution of lat. Vertical dash marks the median.

Show data table

Histogram bins for lat (median: 6.37).
bin	count
-55.27 – -52.06	1
-52.06 – -48.85	1
-48.85 – -45.64	1
-45.64 – -42.43	0
-42.43 – -39.22	2
-39.22 – -36.01	5
-36.01 – -32.8	9
-32.8 – -29.59	13
-29.59 – -26.38	23
-26.38 – -23.17	48
-23.17 – -19.96	100
-19.96 – -16.75	101
-16.75 – -13.54	217
-13.54 – -10.33	202
-10.33 – -7.116	461
-7.116 – -3.906	749
-3.906 – -0.6958	640
-0.6958 – 2.514	344
2.514 – 5.725	434
5.725 – 8.935	609
8.935 – 12.15	676
12.15 – 15.36	264
15.36 – 18.57	368
18.57 – 21.78	210
21.78 – 24.99	282
24.99 – 28.2	320
28.2 – 31.41	131
31.41 – 34.62	124
34.62 – 37.83	148
37.83 – 41.04	81
41.04 – 44.25	107
44.25 – 47.46	65
47.46 – 50.67	68
50.67 – 53.88	68
53.88 – 57.09	35
57.09 – 60.3	20
60.3 – 63.51	34
63.51 – 66.72	18
66.72 – 69.93	13
69.93 – 73.14	6

lng numeric feature

This column contains geographic longitude values, spanning from -178.78 to 179.31 — nearly the full valid range of -180 to 180 degrees. The mean (52.45) and median (47.65) both skew toward positive (eastern) longitudes, suggesting the dataset has a higher concentration of locations in Europe, Asia, or Africa than in the Americas. The IQR of 115.65 and low kurtosis (-0.67) indicate a broadly spread, relatively flat distribution across the globe with only 12 outliers flagged.

Treatment: Use as-is for geospatial modelling; consider pairing with latitude and projecting to a coordinate system, or binning into geographic regions as a categorical feature.

anthropic:default · confidence high

Out[19]:

saturn.columns["lng"].stats

stat	value
n	6,998
nulls	0 (0.0%)
unique	5,560
min	-178.8
max	179.3
mean	52.46
median	47.65
std	79.67
q1	8.282
q3	123.9
iqr	115.7
skew	-0.4982
kurtosis	-0.673
n_outliers	12
outlier_rate	0.001715
zero_rate	0

Fig 10.

Distribution of lng. Vertical dash marks the median.

Show data table

Histogram bins for lng (median: 47.65).
bin	count
-178.8 – -169.8	11
-169.8 – -160.9	4
-160.9 – -151.9	9
-151.9 – -143	11
-143 – -134	10
-134 – -125.1	15
-125.1 – -116.1	94
-116.1 – -107.2	37
-107.2 – -98.21	71
-98.21 – -89.26	257
-89.26 – -80.31	50
-80.31 – -71.35	176
-71.35 – -62.4	150
-62.4 – -53.45	112
-53.45 – -44.5	48
-44.5 – -35.54	21
-35.54 – -26.59	0
-26.59 – -17.64	3
-17.64 – -8.687	95
-8.687 – 0.265	261
0.265 – 9.217	420
9.217 – 18.17	687
18.17 – 27.12	297
27.12 – 36.07	402
36.07 – 45.03	211
45.03 – 53.98	113
53.98 – 62.93	27
62.93 – 71.88	73
71.88 – 80.84	212
80.84 – 89.79	191
89.79 – 98.74	242
98.74 – 107.7	386
107.7 – 116.6	202
116.6 – 125.6	437
125.6 – 134.5	266
134.5 – 143.5	512
143.5 – 152.5	597
152.5 – 161.4	106
161.4 – 170.4	170
170.4 – 179.3	12

p numeric feature

Column 'p' is a numeric field likely representing a price, population, or some monetary/count quantity with extreme positive skew (skew = 39.45, kurtosis = 1870.01). The median is 11,000 while the mean is 1,157,741 — a 100× gap — driven by a long upper tail that reaches 964,553,200, with 1,159 outliers (16.6% of rows). An additional 13.5% of values are exactly zero, suggesting a two-population distribution (absent/zero vs. non-zero values) that may require separate treatment.

Treatment: Separate zero and non-zero records, then log1p-transform the non-zero values before modelling; investigate whether zeros are true zeros or missing-data proxies.

anthropic:default · confidence medium

Out[22]:

saturn.columns["p"].stats

stat	value
n	6,998
nulls	0 (0.0%)
unique	1,627
min	0
max	9.646e+08
mean	1.158e+06
median	11,000
std	1.73e+07
q1	1,200
q3	83,000
iqr	81,800
skew	39.45
kurtosis	1870
n_outliers	1,159
outlier_rate	0.1656
zero_rate	0.1346
alert: high_skew	skew=+39.45
alert: outliers	16.6% rows beyond 1.5 IQR

Fig 11.

Distribution of p. Vertical dash marks the median.

Show data table

Histogram bins for p (median: 11000.0).
bin	count
0 – 2.411e+07	6943
2.411e+07 – 4.823e+07	28
4.823e+07 – 7.234e+07	5
7.234e+07 – 9.646e+07	12
9.646e+07 – 1.206e+08	0
1.206e+08 – 1.447e+08	3
1.447e+08 – 1.688e+08	1
1.688e+08 – 1.929e+08	0
1.929e+08 – 2.17e+08	0
2.17e+08 – 2.411e+08	2
2.411e+08 – 2.653e+08	0
2.653e+08 – 2.894e+08	0
2.894e+08 – 3.135e+08	0
3.135e+08 – 3.376e+08	0
3.376e+08 – 3.617e+08	0
3.617e+08 – 3.858e+08	1
3.858e+08 – 4.099e+08	0
4.099e+08 – 4.34e+08	0
4.34e+08 – 4.582e+08	1
4.582e+08 – 4.823e+08	0
4.823e+08 – 5.064e+08	0
5.064e+08 – 5.305e+08	0
5.305e+08 – 5.546e+08	0
5.546e+08 – 5.787e+08	0
5.787e+08 – 6.028e+08	0
6.028e+08 – 6.27e+08	0
6.27e+08 – 6.511e+08	0
6.511e+08 – 6.752e+08	0
6.752e+08 – 6.993e+08	0
6.993e+08 – 7.234e+08	1
7.234e+08 – 7.475e+08	0
7.475e+08 – 7.716e+08	0
7.716e+08 – 7.958e+08	0
7.958e+08 – 8.199e+08	0
8.199e+08 – 8.44e+08	0
8.44e+08 – 8.681e+08	0
8.681e+08 – 8.922e+08	0
8.922e+08 – 9.163e+08	0
9.163e+08 – 9.404e+08	0
9.404e+08 – 9.646e+08	1

s numeric feature

This column is almost certainly an ordinal rating or severity score with exactly 6 discrete integer values ranging from 0 to 5. The distribution is notably left-skewed (skew = -1.04), meaning high scores (4–5) dominate — the median is 4.0 and Q3 is 5.0 — which would surprise an analyst expecting a balanced scale. Only 3.6% of rows are zero, suggesting the lowest value is rare rather than a default or sentinel.

Treatment: Treat as ordinal; consider whether to one-hot encode or keep as integer, and note the left-skew may bias tree splits toward high values.

anthropic:default · confidence high

Out[25]:

saturn.columns["s"].stats

stat	value
n	6,998
nulls	0 (0.0%)
unique	6
min	0
max	5
mean	3.796
median	4
std	1.366
q1	3
q3	5
iqr	2
skew	-1.041
kurtosis	0.4154
n_outliers	0
outlier_rate	0
zero_rate	0.03572

Fig 12.

Distribution of s. Vertical dash marks the median.

Show data table

Histogram bins for s (median: 4.0).
bin	count
0 – 0.125	250
0.125 – 0.25	0
0.25 – 0.375	0
0.375 – 0.5	0
0.5 – 0.625	0
0.625 – 0.75	0
0.75 – 0.875	0
0.875 – 1	0
1 – 1.125	328
1.125 – 1.25	0
1.25 – 1.375	0
1.375 – 1.5	0
1.5 – 1.625	0
1.625 – 1.75	0
1.75 – 1.875	0
1.875 – 2	0
2 – 2.125	383
2.125 – 2.25	0
2.25 – 2.375	0
2.375 – 2.5	0
2.5 – 2.625	0
2.625 – 2.75	0
2.75 – 2.875	0
2.875 – 3	0
3 – 3.125	1768
3.125 – 3.25	0
3.25 – 3.375	0
3.375 – 3.5	0
3.5 – 3.625	0
3.625 – 3.75	0
3.75 – 3.875	0
3.875 – 4	0
4 – 4.125	1176
4.125 – 4.25	0
4.25 – 4.375	0
4.375 – 4.5	0
4.5 – 4.625	0
4.625 – 4.75	0
4.75 – 4.875	0
4.875 – 5	3093

ss categorical label

This column encodes a conservation or endangerment status classification, with 7 distinct ordered categories ranging from 'safe' through 'extinct'. The dominant class is 'safe' at 43.9% of 6,998 rows, while 'extinct' accounts for 219 records — a meaningful but minority signal. Entropy ratio of 0.756 indicates reasonable spread across categories, though the distribution is notably right-skewed toward safer statuses. No nulls are present, and the label set is clean with no obvious noise.

Treatment: Ordinal-encode in conservation-severity order (safe < vulnerable < definitely endangered < severely endangered < critically endangered < extinct; treat 'unknown' as missing) before modelling.

anthropic:default · confidence high

Out[28]:

saturn.columns["ss"].stats

stat	value
n	6,998
nulls	0 (0.0%)
unique	7
top_value	safe
top_rate	0.4393
cardinality	7
entropy	2.122
entropy_ratio	0.7557

Fig 13.

Top values for ss.

Show data table

Top values for ss (7 unique shown, of 7 total).
value	count	share
safe	3074	43.9%
definitely endangered	1753	25.1%
vulnerable	1160	16.6%
severely endangered	374	5.3%
critically endangered	327	4.7%
extinct	219	3.1%
unknown	91	1.3%

data trove world languages endangerment silence

Overview

Summary confidence: high

n text label

lat numeric feature

lng numeric feature

p numeric feature

s numeric feature

ss categorical label

How to cite