Why did two Spanish sibilants end up as /θ/ and /x/?

I’ve often wondered why standard European Spanish (Castilian) pronounces the letters written<c> and <j> as /θ/ and /x/. Penny (2002) and Alkire and Rosen (2010) explain that this comes from the history of sibilants in Spanish.

Old Spanish sibilants

In the 15^th century, Old Spanish had 7 sibilants (table 1).

Table 1. Old Spanish sibilants

4 of the sibilants are fricatives, formed by bringing the tongue close to (but not quite touching) top of the mouth (or the teeth). The other 3 are affricates, formed with a stop (/t/ or /d/) followed by one of the fricatives.     

The rows in table 1 are arranged in order (front to back) by the point at which the tongue nearly touches the top of the mouth.

• for dental-alveolar /t^s/ and /d^z/, the tongue initially touches the teeth (the position for /t/ and /d/). It then moves back into the position for /s/ and /z/.
• for alveolar /s/ and /z/, the tongue almost touches the area around the alveolar ridge just behind the teeth.
• for prepalatal /ʃ/ and /ʒ/ the tongue almost touches the roof of the mouth further back, just ahead of the hard palate.
• for alveolar-prepalatal /tʃ/, the tongue starts in (or maybe just behind) the teeth in the dental or alveolar area. It than moves backwards into position for /ʃ/.

6 of the 7 Old Spanish sibilants come in pairs (one member voiced, the other member unvoiced).

Table 2 shows how Old Spanish spelled its sibilants.

Table 2. Old Spanish spellings

Changes to the Old Spanish sibilants

The Old Spanish sibilants changed in the 16^th century. The 3 voiced sibilants lost their voicing, becoming unvoiced and undistinguishable from their unvoiced counterparts. This reduced the number of sibilants from 7 down to 4 (table 3).  

Table 3. Old Spanish sibilants after devoicing (16^th century)

Then, in the 17^th century, the affricate /t^s/ lost its affrication (the initial /t/) to become /s̪/, a sound similar to /s/ but pronounced with the tip of the mouth slightly further forward than for /s/ (table 4).

Table 4. Old Spanish sibilants after deaffrication of /t^s/ (17^th century)

Those 2 changes left Spanish with only 4 distinctive (phonemic) sibilants in place of the previous 7. Also, the remaining 4 phonemes were produced with the tongue in one of 4 places that were very close each other, making them difficult to distinguish.

To make the 4 sounds more easily distinguishable:

• /s̪/ moved forward to become /θ/, the interdental sound /θ/, produced with the tongue between the teeth (like <th> in English thin).
• /ʃ/ moved back to become /x/, with the tongue nearly touching the velum at the back on the mouth (like <ch> in Scottish loch).

Table 5 shows the resulting system in modern Castilian. As table 5 shows, the modern Castilian Spanish consonants are separated much more cleanly than in the 17^th century system of table 4.  

Table 5. Modern Spanish (Castilian) sibilants

Spelling

In Modern Spanish:

• the consonant pronounced /θ/ is spelled as ⟨c⟩ before ⟨e⟩ or ⟨i⟩ and elsewhere as ⟨z⟩.
• the consonant pronounced /x/ is spelled as ⟨g⟩ before ⟨e⟩ or ⟨i⟩ and elsewhere as ⟨j⟩.

When Cervantes wrote Don Quixote, the consonant written as <x> or as <j> was still pronounced /ʃ/. That pronunciation still exists in French (Don Quichotte) and Italian (Don Chisciotte) because the novel was translated into those languages during that period.

Andalusian and Latin-American Spanish

In the Spanish varieties of Andalusia (in southern Spain) and Latin America, /s̪/ did not end up in the same place as Castilian (as /θ/). Instead of moving /s̪/ forward to become /θ/, those varieties moved /s/ forward, merging it with /s̪/ and pronouncing the newly merged consonant more like /s̪/ than like /s/ (table 6). This practice is known as seseo. In a few (mainly inland) parts of Andalusia, the merged consonant sounds more like /θ/ (this is called ceceo).

Table 6. Modern Spanish (Andalusian and Latin American) sibilants

In many Spanish varieties, though not in Castilian Spanish:

• /s/ may be pronounced as [h] or omitted entirely, especially at the end of a syllable.
• velar [x] may be pronounced as uvular [χ] (in parts of Spain) or glottal [h] (in Andalusia, Canary Islands, and parts of the Americas).

Conclusion

Having read the explanations in Penny (2002) and Alkire and Rosen (2010), I feel I understand why Castilian pronounces the letters written<c> and <j> as /θ/ and /x/. The Spanish sibilant system was restructured in the 16^th and 17^th centuries.

That restructuring left 3 consonants so close together that they must have been hard to distinguish. To restore the distinctiveness of those consonants, one moved forward to the front of the mouth (/θ/) and another moved near the back of the mouth (/x/).

Sources

A History of the Spanish Language, Ralp Penny (2^nd edition, 2002)

The Romance Languages, Alkira and Rosen (2010)

Typologically Exceptional Phenomena in Romance Philology, Eulalia Bonet and Francesc Torres-Tamarit (section 9.2.4 Spanish ceceo) in chapter 9 of The Cambridge Handbook of Romance Linguistics, edited by Adam Ledgeway and Martin Maiden (2022).