Why does Spanish tend to swap letters in words?

Question

I can't remember the source, but I recall hearing that Spanish (my native tongue) tends to swap letters in words (accidentally). Examples

• Latin diabolo becomes diablo(o). (perhaps this is a non-example, and it is losing an 'o' instead.)
• French guirlande becomes guirnalda.

• Ukrainian Ukraina (Україна) becomes Ucrania.

  1. Does this phenomenon have a name?

  2. Why does it happen in Spanish?

  3. Is it common throughout (romance?) languages or specific to Spanish?

Answer 1

Metathesis is common across languages, including in the varieties of Romance that emerged from Vulgar Latin. However, Western Romance had it more than Eastern Romance; within Western Romance, Iberian Romance appears to have had it more than Gallo-Romance.

Consonant clusters involving a dental and /n/ and /l/

In Western Romance, many new clusters of /tn/ were formed from the loss of a vowel. For example: Classical Latin retineō > Vulgar Latin *retina underwent intervocalic voicing to form */ˈrɛdina/, preserved in Italian as redine. But the unstressed vowel in the middle was lost in Western Romance, giving */ˈrɛdna/ which led to different outcomes:

Metathesis: Spanish rienda, Galician renda

Lenition > Assimilation: French resne > rêne, from which English rein

Dissimilation: Catalan regna

Another good example: descendants of Latin spatula:

Assimilation /t/ to /l/: Italian spalla, Sicilian spadda; also Friulian spale, Walloon spale, Venetian spała, and French espalle > espaule > épaule

Dissimilation of the /l/: Catalan espatlla

Metathesis: Spanish espalda, Portuguese espalda

preserved: Romansch spatla, Occitan espatla

Metathesis between /r/ and /l/

These two sounds are traditionally classified as "liquids", and they commonly underwent metathesis in both directions.

The renowned example of Old French crocodile ~ cocodril actually dates from medieval Latin crocodīlus vs cocodrillus, and modern languages probably fell on one side or the other in the early modern period. These are probably "semi-learned" borrowings rather than being fully inherited.

Nonetheless, semi-learned but genuine examples such as Latin miraculum > Spanish milagro, Galician and Portuguese milagre; periculum > Spanish peligro, Asturian and Leonese peligru (metathesis didn't happen in Galician and Portuguese perigo) testify to how common it was.

Glide metathesis

This surfaces as changes in the vowel in the modern language, and is common to much of Western Romance. It is most obvious with Latin /sj/ and /rj/ glides. For example, for Latin caseus, resulted in Spanish and Aragonese queso, Galician queixo, Portuguese queijo, Leonese queisu, but Sardinian casu, Neapolitan caso, Sicilian caciu, Italian cacio, Romanian caș.

In this case, Portuguese goes further than Spanish, as metathesis seems to have happened with Latin words with /pj/ and /bj/, e.g. Latin apiu led to Spanish and Galician apio but Portuguese aipo, Latin rabia led to Spanish and Galician rabia but Portuguese raiva.

Stronger metathesis in Old Spanish

There were quite a few more examples in Old Spanish, especially in certain future tense conjugations (such as the regular tenré and after metathesis terné, as well as the version tendré with consonant insertion which is the modern standard form). In summary, from an optimality perspective, a rising sonority across the syllable boundary led to a lot of metathesis, in a tendency to preserve the usual "sonority cycle" that makes up a syllable.

This also included the second person plural imperatives with enclitic object pronouns, e.g. decid + lo making decildo as well as the regular decidlo. The article says that this became non-productive by the 17th century, as the ALIGN constraint moved above the markedness constraint SYLLABLE CONTRAST.

Answer 2

The modern language Spanish does not have a significant "tendency" to moving segments (a process known as metathesis), but there are some historical cases of metathesis going historically from Latin or other older language stages to some branch of Romance (branches leading to Spanish). Some examples are caseo > queso 'cheese', catēnātum > candado 'padlock', parabola > palabra 'word', spatula > espalda 'back', oblitare > olvidar 'forget'. In modern Spanish, the verb 'forget' is simply learned as /olvid(a)/ and there is no metathesis.

This is not a regular sound change (compare the change of earlier [ʒ] to [x] spelled j): it sporadically affects some words but not others. It is found in other Romance languages (berbex > brebis 'sheep', formaticus > fromage 'cheese', mosquito (Spanish) > moustique. Very many languages in the world have this as a sporadic historical change. A few have it as a actual rules in the grammar (Faroese, Klallam), and many more have been said to have it but investigation of the details indicates that most cases are two separate processes. The question of why this happens is vexing, but it generally seems to be related to changes in syllable structure. For instance, the earlier form espadla runs counter to the emerging tendency to not have obstruents in the coda.

Cases like caseo > kasʲo > kaiso > keso, also found historically in Ancient Greek, seem involve intermediate palatalized consonants and perceptual "leftward attraction" of the palalization to the onset of the consonant (this is a variable in palatalized consonants across languages: sometimes they are not evidence on the preceding vowel, sometimes they are).

Liquid metathesis (palabra) seems to have a different explanation. It seems that it is generally difficult to acoustically parse and localize two liquids (r,l) in a word, so what people can get is that there are two of them, but it's hard to perceive what order they appear in. Thus r...l and l...r "metathesis" is also common across languages. So there are a number of explanations, depending on the particular case.