till Oct 26, 2022 civi en email only valid for prefer language with en_US/ en_AU / en_CA / en_GB / en_ZA / en_NZ / en_IN (should make last 2 official), while db got below. and they are also added as option for civi, needs to clean them and them remove the option.
MariaDB [civicrm]> select preferred_language, count(*) FROM civicrm_contact WHERE preferred_language like "en_%" and preferred_language not in ('en_US', 'en_AU', 'en_ZA', 'en_GB', 'en_CA', 'en_NZ'. 'en_IN') group by preferred_language;

190 rows in set (2.117 sec)

MariaDB [civicrm]> select is_deleted, count(*) FROM civicrm_contact WHERE preferred_language like "en_%" and preferred_language not in ('en_US', 'en_AU', 'en_ZA', 'en_GB', 'en_CA') group by is_deleted;

2 rows in set (11.325 sec)
Start for this one, that we should clean 3291583 of contact and then clean the option table civicrm_option_value where value = 'en';
with the drush commend as sh -c "echo '{\"values\":{\"preferred_language\":\"en_US\"},\"where\":[[\"preferred_language\",\"NOT IN\",[\"en_US\", \"en_CA\",\"en_ZA\", \"en_AU\", \"en_GB\", \"en_NZ\", \"en_IN\"]], [\"preferred_language\",\"LIKE\",\"en_%\"]],\"limit\":5000, \"version\":4}' | drush @wmff cvapi Contact.update --in=json"

then run delete from civicrm_option_value where value = 'en' and name not in ('en_US', 'en_AU', 'en_ZA', 'en_GB', 'en_CA','en_NZ', 'en_IN'); to get rid of the invalid en option from civi dropdown

echo '{"where":[["name","NOT IN",["en_GB","en_US","en_CA","en_AU","en_ZA","en_NZ","en_IN"]],["value","=","en"]]}' | cv api4 OptionValue.delete --in=json;
actually use https://gerrit.wikimedia.org/r/c/wikimedia/fundraising/crm/+/856688/ this (drush @wmff cvapi WMFDataManagement.CleanInvalidLanguageOptions version=4) to clean the unused language options

still seeing some new contact been created with the invalid en_xx, id 30851434, 57584905 and 750707 from online donation. but switching to other language first

for spanish, civi supports es_ES, es_MX, and es_PR, while from civi db, we have
MariaDB [civicrm]> select preferred_language, count(*) FROM civicrm_contact WHERE preferred_language like "es_%" and preferred_language not in ('es_PR', 'es_MX', 'es_ES') group by preferred_language;

112 rows in set (0.298 sec)
MariaDB [civicrm]> select is_deleted, count(*) FROM civicrm_contact WHERE preferred_language like "es_%" and preferred_language not in ('es_PR', 'es_MX', 'es_ES') group by is_deleted;

2 rows in set (1.363 sec)

462021 got cleaned up

for french, civi supports fr_FR, fr_CA, while from civi db, we have
MariaDB [civicrm]> select preferred_language, count(*) FROM civicrm_contact WHERE preferred_language like "fr_%" and preferred_language not in ('fr_FR', 'fr_CA') group by preferred_language;

99 rows in set (0.060 sec)

MariaDB [civicrm]> select is_deleted, count(*) FROM civicrm_contact WHERE preferred_language like "fr_%" and preferred_language not in ('fr_FR', 'fr_CA') group by is_deleted;

2 rows in set (0.653 sec)

cleaned with option value cleaned

for chinese, civi supports zh_TW, zh_CN, while from civi db, we have
MariaDB [civicrm]> select preferred_language, count(*) FROM civicrm_contact WHERE preferred_language like "zh_%" and preferred_language not in ('zh_TW', 'zh_CN') group by preferred_language;

34 rows in set (0.036 sec)
MariaDB [civicrm]> select is_deleted, count(*) FROM civicrm_contact WHERE preferred_language like "zh_%" and preferred_language not in ('zh_TW', 'zh_CN') group by is_deleted;

2 rows in set (0.327 sec)
cleaned up

Change 857083 had a related patch set uploaded (by Wfan; author: Wfan):

[wikimedia/fundraising/crm@master] Filter out is_deleted = 1 for valid contact local options

https://gerrit.wikimedia.org/r/857083

Change 857083 merged by jenkins-bot:

[wikimedia/fundraising/crm@master] Filter out is_deleted = 1 for valid contact local options

https://gerrit.wikimedia.org/r/857083

based on this query: select label, value, count( * ) as count from civicrm_option_value where name like '%_%' and label like '%_%' and is_active = 1 and option_group_id=86 group by value having count > 1 order by count desc;

73 rows in set (0.008 sec)