Category Archives: From Latin to Spanish

The origins of Spanish -ar, -er, and -ir verbs

I never would have thought to look into the origins of the Spanish verb classes if I hadn’t studied Hebrew. While Spanish has three verb classes (-ar, -er, and -ir), Hebrew has seven, called binyanim. Each is conjugated differently, as in Spanish, but each binyan additionally imparts meaning. For example, the three-consonant Hebrew root k-t-v, which refers to writing, appears in all seven binyanim. As shown in the table below, based on this helpful summary, each binyan reflects a different aspect of writing.


There are no patterns like this in modern Spanish; that is, you can’t infer anything about a verb’s meaning from its conjugation class. Differences like reflexive and passive are expressed through pronouns and auxiliary verbs, i.e. se escriben (“they write to each other”) and está escrito (“It is written”). But I’ve often wondered whether, if one goes back far enough, one can find any semantic logic behind which verbs are in which class.

Cornell’s Michael Weiss discusses exactly this topic in chapter 36 of his terrifyingly authoritative Outline of the Historical and Comparative Grammar of Latin. It turns out that there were clear connections between certain semantic categories and the conjugation classes of Latin, although this patterning was in no way as tidy or as far-reaching as in Hebrew. Some highlights are below.

Latin’s -āre verb class, which evolved into the -ar class of Spanish, was used:

  1. to turn nouns and adjectives into verbs. Some examples are curare “to care” from cura “care”, navigare “sail” from navex “sailor”, and novare “to renew” from novus “new”.
  2. for repeated or frequent actions. Some examples are dictare “recite” from dicere “say” and factitare “to practice” from facere “to do, make”.
  3. for intensives (with a prefix). One example is ocupare  “to seize” from capere “to take”.

Latin’s -ēre class, which evolved into the -er class of Spanish, was used:

  1. for causatives, such as monere “warn” from men “think” (i.e. to cause someone to think) and docere “to teach” from dek “accept” (i.e. to cause someone to accept).
  2. for verbs that describe states, e.g. calere “to be hot”, frigere “to be cold”, pendere “to be hanging”.

Latin’s -ĕre class, which merged into the -er class of Spanish, included a group of change-of-state verbs, e.g. calescere “get hot” (from calere) and tacescere “become quiet” (from tacere “to be quiet”).

Latin’s -īre class, which evolved into the -ir class of Spanish, was used:

  1. to turn nouns into verbs, as in finire “to finish” from finis “end” and servire “to serve, be a slave” from servus “slave”. I don’t know what, if anything, distinguished these from the verbs-from-nouns in the -are class.
  2. for desires, e.g. esurire “to be hungry” from esse “to eat”, parturire “to be in labor” from parere “to give birth”.

This partial patterning reminds me of gender. There’s nothing inherently masculine about most masculine nouns in Spanish, nor inherently feminine about most feminines. Yet one can see in such common words as el padre and la madre the meaningful basis of the original category difference. It’s refreshing to find that there’s likewise some logic, however fragmentary and forgotten, to the seemingly arbitrary verb classes of Spanish.

Some surprising Spanish-English cognates

I’ve always been a big fan of cognates, i.e. genetically related words across languages, like Spanish insistir and English insist. As a language learner, I’ve found cognates enormously helpful in reading and in memorizing vocabulary. I really missed them when I studied Hebrew.

As a teacher, I point out cognates and also teach my students to use them just as I do: as an aid in reading and in memorization. An additional benefit is intellectual. Articles in the popular press with titles like “Latin comeback in the schools” invariably make the claim that studying Latin helps students learn the roots of English vocabulary. There’s no reason why students can’t have the same benefit from studying Spanish or other Romance languages.

As a linguist, I’m happiest when I learn a non-obvious cognate: the kind that gives you an “Aha!” or “Really?” moment. The table below lists my favorite “Aha!” Spanish/English verb cognates. For example, I’ve always explained the verb disfrutar to my students metaphorically, as enjoying the fruits of life (or whatever) — here I normally mime plucking fruit from a tree – but until I looked it up I didn’t realize that this was the verb’s actual etymology. Again, I’ve left out what I consider to be more obvious cognates, such as savvy, savory, or homo sapiens for sabercognitive or acquaintance for conocer, or dictate for decir. [There, I worked them in anyway!] My sources for the table are the etymologies on the Real Academia website and Douglas Harper’s very impressive Online Etymology Dictionary, which is great fun to browse.

Please write in with your own favorite cognates, including cognates for nouns and adjectives.

Some verb cognates

Spanish and and and

You can’t speak Spanish without the word y (“and”). It’s the fourth most frequent word in Spanish, after el/lade, and que. It’s indispensable for telling time (Son las dos y veinte) and counting (treinta y cinco). It shows up in everyday expressions like blanco y negromadre y padre, and ida y vuelta. And it has the magical property of both creating longer sentences (Comimos en la taquería y fuimos al cine) and shortening questions (¿Y tú?).

You probably already know that y changes to e before a word that begins with i or hi, as in Mi madre es bonita e inteligente or Tengo veinte sobrinos hijos. This is more than a spelling change. The pronunciation changes, too, to prevent the adjacent /i/ sounds from blending together.

I knew this. And I also knew, from studying Latin for a year in college, that the Latin word for “and” is et. But somehow I never put two and two together (that’s dos y dos) to realize that Spanish e is a lot closer to the original Latin than the normal y form is.

In fact, as Tom Lathrop explains, y is another case of a historical “flip” in the language. Earlier this summer, this blog considered verbs like conocer and hacer, whose yo forms conozco and hago are irregular by modern standards but historically conservative. All the other present-tense forms of these verbs, plus the infinitives, have diverged substantially from Latin. Likewise, e can be traced back directly to Latin, while y is a Spanish innovation. So although from a modern perspective y is normal and e exceptional, from a historical perspective it’s the other way around.

Seen from a modern perspective, “y” is normal and “e” is exceptional. Seen from a historical perspective, it’s the other way around.

Let’s see how this happened. As Lathrop explains, Spanish lost the final t of et, as it did final -t in general; compare Latin dicit and Spanish dice (Lathrop p. 129). When the resulting e form came before a word that began with a vowel, it was natural to turn the sequence e + vowel into y + vowel. Lathrop gives these examples (p. 200). If you say the words out loud you should be able to recreate this transformation for yourself.

  • e amigos > yamigos > y amigos
  • e obispos > yobispos > y obispos
  • e uno > yuno > y uno

This change happened before all vowels but /i/, for obvious reasons, and then took over before consonants as well. Poor e was now the odd man out.

The change of o “or” to u before o is a horse of a different color, by the way: a change that never quite got off the ground. O is the natural development of Latin aut. Penny explains that “The form u probably arose in pre-vocalic position [like y for e], and has only in the modern period come to be restricted to use before words beginning with /o/” (p. 199). It coulda been a contender!

Conexión and corrección

I love the Spanish spelling of conexión. The x is somehow very elegant. But I’ve occasionally wondered why the word isn’t spelled conección, with a cc as in corrección. I looked into this question recently and the answer is very simple. In fact, there are two simple answers.

First, the spelling difference reflects the proper Castilian Spanish pronunciation of x versus cc. The letter x between two vowels (conexión, examen, etc.) is pronounced ks. The letter sequence cc as in corrección is pronounced kth, since th is the Castilian pronunciation of c before i. Two spellings, two pronunciations, fair and square.

Second, the spelling difference respects etymology. Spanish words ending in –xión had an x in Latin, too: conexión comes from connexĭōn(is), reflexión from reflexĭōn(is), and so on. Spanish words ending in –cción had a ct in Latin: corrección from correctĭōn(is)acción from actĭōn(is)inyección from iniectĭōn(is), and so on.

The cc words, by the way, greatly outnumber the x words. The latter include only the following (based on various google searches — I need a reverse Spanish dictionary!):

  • anexión
  • complexión
  • conexión etc. (desconexión, inconexión, interconexión, reconexión)
  • crucifixión, transfixión
  • flexión etc. (reflexión, inflexión, irreflexión, genuflexión)
  • fluxión

It’s interesting that English — especially American English — has moved many of the x words into the cc group, which we spell with the original ct (e.g. connection). In fact, the British “look” of the x is probably what makes Spanish conexión appear elegant to my American eyes. We’ve likewise adapted the various words derived from flexion (reflection, genuflection, and inflection), though not, mysteriously, the word flexion itself. Complexion and crucifixion are probably the most commonly used xion holdouts in American English.

So, this post is one for your X-files!





The sadistic Spanish subjunctive

I can remember the exact moment when Spanish utterly and permanently captivated me. I was fifteen years old and in my fourth year as a Spanish student. Our class had wrapped up the basic tenses and the present subjunctive, and was ready to launch into the imperfect subjunctive. Our teacher explained to us that this tense was based on the pretérito and incorporated all of its irregulars.

This struck me as laugh-out-loud funny. We had already learned that the present subjunctive inherited all the idiosyncrasies of the normal present tense, including the ones that only show up in the yo form (the -zco and -go types). But the preterit is even thornier. It seemed bizarre beyond belief that the subjunctive should adopt the most problematic elements of both these tenses.

As a student, it amused me to imagine that a twisted “Spanish committee” (perhaps a branch of the Spanish Inquisition?) had designed the subjunctive. (My little PowerPoint below depicts this scenario.) As a teacher, I now like to tell my students that the present subjunctive is God’s way of getting them to review the irregular verbs that they’d studied weeks, months, or even years ago. I figure that teaching at a Jesuit university authorizes me to invoke God in the classroom.

In fact, the many irregulars of the subjunctive are neither a cosmic joke, an evil machination, nor an act of God. They’re simply a coincidence. The present and imperfect subjunctive happened to follow the same evolutionary paths as several distinct categories of irregular verbs in the present and pretérito indicative.

Consider the examples of irregular verbs shown in the table below, color-coded for your convenience.

irregular subjunctive

The “boot” verbs, in yellow, are irregular in the present tense because of a language-wide process that changed stressed /o/ to /ue/ and stressed /e/ to /ie/. The corresponding present subjunctives have the same vowels and the same stress pattern, and therefore the same irregularity.

The -ir “sole” verbs, in blue, are irregular in the present and the pretérito because of another general process: the raising of /e/ to /i/ and /o/ to /u/ before /j/ (the sound of English y). All Latin -ire present subjunctive endings, and the “sole” (3rd person) endings of the imperfect subjunctive, contained (or still contain) /j/, triggering the vowel change.

The -zco and -go irregulars of the present tense, in green, evolved because the o ending of the yo form insulated it from changes that affected the other present tense forms and the infinitive. The subjunctive endings for these verbs begin with a, which had the same insulating property.

Finally, the drastic stem-changing pretéritos, in magenta, descend from Latin’s “strong perfect” past tense forms. The imperfect subjunctive is based on Latin’s pluperfect tense, which had the same irregularities.

This leaves “only” the six additional irregulars of the present tense subjunctive. Their diverse origins are summarized below.

6 irregular present subjunctives

As always, if you want to learn more, the best source is Ralph Penny’s A History of the Spanish Language. But beware — nobody expects the Spanish Inquisition!

The Spanish pretérito is like Tolstoy

Leo Tolstoy began Anna Karenina with the wonderful observation that “Happy families are all alike; every unhappy family is unhappy in its own way.” This quote always makes me think of the Spanish pretérito (preterit) past tense. The pretérito has lots of irregulars, and they are irregular in many different ways.

Unhappy families are like the pretérito.

We’ve already seen that in the present tense, most irregulars — the boot verbs-zco and -go verbs, and ver — developed as Vulgar Latin transitioned to Spanish. The pretérito is different because most of its irregulars were already irregular in Latin. (Well, Spanish added a few for good measure…keep on reading.)

Basically, the Spanish pretérito tense derives from the Latin perfect, which was riddled with irregulars. Latin scholars refer to them as “strong” perfects, as opposed to the regular, or “weak” perfects. The strong perfects were characterized by stress on the root instead of the ending in some of the verb forms. You can still hear this difference in Spanish. For example, irregular dije and dijo are stressed on the di- root, while regular hablé and habló are stressed on the and endings. The strong perfects also lacked the normal /v/ ending of regular perfects, such as laudāvī “I praised” (from laudāre) or audīvit “he heard” (from audīre).

Caeser loved those strong perfects!

Most of today’s irregular pretéritos can be traced to three subtypes of the Latin strong perfect:

Strong perfects

As always, analogy muddied the evolutionary trail. Several common verbs that were regular in Latin picked up the u pattern of hube, supe and the like, including tener/tuve, estar/estuve, and andar/anduve. The irregular pretérito of ver (vi, viste, vio, etc.) influenced that of dar (di, diste, dio, etc.), while the pretéritos of ser “to be” and ir “to go” merged. On the other hand, many verbs with strong perfects in Latin became regular in Spanish. Some examples are temer, which belonged to Latin’s –class (Latin timuī, timuístī, etc.), escribir “to write”, in the – class (scrīpsī, scrīpsístī, etc.), and leer “to read”, in the –ī class (lēgī, lēgístī). Their modern yo pretéritos are regular temí, escribí, and leí instead of something like tume, escrise, and lije.

Finally, to make matters worse, Spanish “invented” its very own irregular pattern: the so-called “sole” verbs. These are -ir verbs with a “boot” change in the present, like servir (sirvo, sirves etc.), mentir (mientomientes, etc.), and dormir (duermoduermes, etc.). In the pretérito, the e or o of the root changes to an i or u in the él and ellos forms. This happens for the same reason I described in my boot verb post: raising an /e/ to /i/ and an /o/ to /u/ anticipates the height of the y sound (transcribed properly as /j/) that you get when you pronounce the -ió and -ieron in fluent speech.


Suela verbs are only Irregular in the sole of the boot

That’s a lot of irregulars — and a lot of unhappy families, at least the night before a major test…

Twice the subjunctive, twice the fun

This post is a little denser on grammar than usual, so I’ve inset helpful expositions in red.

The American comedian W.C. Fields famously quipped that if first prize was a week in Philadelphia, second prize was two weeks. I suppose that most Spanish students feel the same way about the subjunctive. If first prize is the present subjunctive, second prize is the imperfect subjunctive. Third prize, then, must be the present subjunctive plus TWO imperfect subjunctives: the ones with -ra and the ones with -se.

(If you’re rusty on the imperfect subjunctive, a decent review is here.)

For a linguist, however, the many subjunctives of Spanish are pure candy. For one thing, it’s delightfully contrarian that neither imperfect subjunctive is directly related to the original Latin version. Instead, these two tenses (or moods, more precisely) arose from two Latin pluperfect conjugations: the pluperfect subjunctive (for -se) and the pluperfect indicative (for ra).

“Pluperfect” simply means a tense that is used to talk about actions completed before some past point in time. The Modern Spanish equivalents use compound structures, as in Si hubiera comido… “If I had eaten…” (pluperfect subjunctive) and Había comido “I had eaten” (pluperfect indicative).

The following table shows the original Latin conjugations and their outcome in Spanish.

imperfect subjunctive

Even more interesting is the mere fact that the two conjugations co-exist in modern Spanish. Although not as common as the -ra subjunctive, the -se subjunctive is certainly current, especially in the written language, and is understood throughout the Spanish-speaking world. This is remarkable because in grammar, a difference in form normally implies a difference in meaning. Comí, comía, and he comido are all different kinds of past tenses; voy a comer and comeré refer to  imminent vs. distant future events The Spanish imperfect subjunctive is thus a striking exception. In fact, it’s the only case I’m familiar with in any language where two parallel constructions co-exist in the core of the grammar. (If you know of any others, please share them!)

Essentially, what we are seeing here is a snapshot of a slowly moving change in the language. The -se subjunctive came first; Ralph Penny dates it to Old Spanish, roughly 1000 to 1400 CE. Its evolution was relatively simple, involving only a shift in time (from pluperfect to normal past tense) along with shortened verb endings as a result of normal sound change. The -ra subjunctive took longer to evolve because it also involved a switch from indicative to subjunctive: that is, from actuality to possibility. Penny dates its emergence to the Golden Age (1500 to 1700), and its overtaking the -se subjunctive to “more recent times”.

In the Spanish imperfect subjunctive, then, the language’s past and future co-exist peacefully in the present. Although this is an unusual situation in grammar, it’s one that we’re used to in vocabulary, which changes much faster. A contemporary American English speaker, for example, understands the outdated swell and nifty, the timeless cool, and possibly the trendy swag and dope. The Spanish parallels would depend on dialect, but just consider the exuberant redundancy of maravillosofantásticoestupendofabulosoexcelentefenomenaltremendo, and buenísimo. The co-existence of the two Spanish subjunctives is therefore simultaneously exotic and familiar.

Spanish ver veered

Ver es creer (“Seeing is believing”).

In Old Spanish — the Spanish of El Cid, spoken into the 1400s — ver and creer had more in common than this refrán. The Old Spanish version of ver was veer, and its conjugation was identical to that of creer (and likewise leer).

veer creer leer

The three verbs were similar because they had similar Latin roots: viderecredere, and legere. In all three cases, the middle consonant (d or g) dropped out, creating a double ee in all but the yo form. (The change of i to e in videre was normal, see this earlier post.)

Although I didn’t bother to include it in the table above, poseer, from Latin possidēre, fits the same pattern.

Veer became an irregular verb when its history diverged from creer and leer. As Old Spanish gave way to the modern language, the double ee of veer simplified to a single e. (The expected yo form from this point onward was vo). Later, the edes ending for vosotros changed to eis for all -er verbs. These two changes together gave rise to the modern conjugations:

veer creer leer modernVer‘s history is both unique and familiar. Unique, because it created the only verb with this specific irregularity. (I’m not counting prever, which is of course based on ver.) Familiar, because as with the much more frequent -zco and -go verbs, the more conservative yo form, ironically, became irregular as a side-effect of changes that affected other forms of the verb, including the infinitive. Veo itself would have remained just as regular as creo and leo if the other verb forms hadn’t changed.

So, dear reader: Viste. Leíste. ¿Creíste?

Irregular yo in Spanish — NOT!

Because my high school friend Peter was on the yearbook committee, he managed to have his picture published upside-down (that’s NOT him in the picture below), with the senior quote “Maybe, just maybe, everyone else is wrong”.

"Maybe, just maybe, everyone else is wrong"

“Maybe, just maybe, everyone else is wrong” 

In our family, Peter’s quote has become an anti-trend mini-meme. No Facebook account? Don’t think Seth MacFarlane is funny? Every other “car” in the parking lot is an SUV? Maybe, just maybe…

Believe it or not, I thought of Peter immediately when I looked into the origin of the two biggest Spanish verb groups with irregular yo forms in the present tense: the so-called -zco and -go verbs. (This is particularly ironic because Peter didn’t even take Spanish.) It turns out that from a historical perspective, the yo forms are actually the most REGULAR — that is, most faithful to the original Latin. In a nutshell, their -o ending insulated them from sound changes that affected /k/ before /e/ and /i/; i.e. before front vowels.

So in this case, everyone else IS wrong, or at least linguistically radical.

For -zco verbs like florecer, from Latin florescere, the relevant change was the simplification of /sk/ to /s/ (in Andalucian and Latin American Spanish) or /Θ/ (in Castilian Spanish) before /e/ or /i/. [Note: this is a drastic abbreviation of a process that involved several intermediate steps; see Ralph Penny’s A History of the Spanish Language or another good history of Spanish for details.] So the infinitive changed from florescere to florecer, the  form from floresces to floreces, and so on. Only florezco kept the /k/ cluster of the original Latin.

For -go verbs like hacer, from Latin facere, the relevant change was the fronting and softening of /k/ to /s/ or /θ/ before front vowels. So the infinitive changed from facere (with a /k/ sound) to hacer, the  form to haces, and so on. Only haco remained in the present tense as a reflection of the original Latin /k/. Later, a separate change voiced the /k/ to /g/, giving us modern hago. (This /g/ is still a lot closer to /k/ than is /s/ or /θ/.) A similar sequence of events impacted Latin dicere as it evolved into decir, giving us the (yo) digo form.

All the sound changes mentioned above were general, occurring throughout Spanish vocabulary. For example, the /sk/ simplification gave us pez (from Latin pesce), /k/ fronting gave us cielo (from Latin caelu), and /k/ voicing gave us lugar (from Latin locale). The change of /f/ to /h/ in hacer (from facere) is also seen in words like hijo (from Latin filius).

As I described in an earlier post, analogy untidies the results of sweeping sound changes like these. This was certainly the case with the -zco and -go verbs. The verb lucir and related verbs like deslucir adopted the -zco pattern, as did several verbs ending in -ducir, such as producir, even though none of these had an /sk/ cluster in Latin. Likewise, the -go pattern spread to other common verbs including venir, tener, and salir, though at the same time, some verbs originally in the -go group became regular (cocer is one). After these back-and-forths, modern Spanish ended up with almost 100 –zco verbs, and around 10 –go verbs. You can check my Teaching page for a full list of the -zco verbs.

The other irregular yo types — ver, the -oy verbs, and the two -e verbs (haber and saber) –are another story entirely; maybe I’ll post about them later.

[Update: I have now posted about ver.]

Spanish boot verbs, sound change, and analogy

Lately I’ve been looking into the origins of the Spanish irregular verbs oh-so-affectionately called “boot” verbs. They are more properly called “stem-changing verbs” because their final stem vowel changes from e to ie (e.g. negar/niego), from o to ue (e.g. poder/puedo), or from e to i (e.g. medir/mido). It’s exciting to discover that this verb class is a perfect example of the two classic forces in language change: sound change and analogy. As any basic linguistics textbook will tell you, sound change affects all words with a given sound, and analogy then messes things up.

Two sound changes are responsible for the “boot” verbs: the change of Latin short  /ĕ/ and /ŏ/ to the diphthongs (vowel sequences) /ie/ and /ue/ in stressed syllables, and the raising of /e/ to /i/ before the sound /j/, which is pronounced like English y. The first change is more common because it affected -ar-er, and -ir verbs. It’s the change behind the form of much Spanish vocabulary, including such common words as fiesta and puerta, from Latin festa and porta (see this previous post). And because it’s confined to stressed syllables, it’s the source of the classic “boot” pattern, where the diphthong occurs in the singular and the 3rd person plural (pUEdo, pUEdes, pUEde, pUEden) but not in the nosotros and vosotros forms, where stress falls on the verb ending (podEmos, podÉIs).

Analogy messed up this tidy result by turning some regular verbs into boot verbs and some boot verbs into regulars. The former is akin to the emergence of dove as an alternative to dived, by analogy to drove and other irregular “strong” English verbs. The most common verb that “went boot” is pensar, which shouldn’t have a stem change because its /e/ comes from a Latin long vowel. Some examples of former boot verbs that are now regular are prestar (formerly priesto, priestas, etc.) and diezmar (the original infinitive was dezmar).

The raising of /e/ to /i/ before /j/ only happened in a few verb forms, but analogy took it the rest of the way. The /j/ that triggered this change occurred in Latin -īre verb endings; this is why all Spanish verbs in this boot class are -ir verbs. For example, the – ending of Latin mētiō “I measure” came to be pronounced /jo/ in Vulgar Latin, triggering the change of /met/ to /mit/. This vowel change then spread, by analogy, throughout the full “boot”, and the /j/ was eventually lost(The /t/ also turned into a /d/, obviously.)

Spanish was actually supposed to have four types of boot verbs, because /j/ affected /o/ as well as /e/, raising it to /u/ in a number of -ir verbs. However, in these cases analogy truly ran rampant and /u/ completely took over the verb, changing Latin mollire, for example, to Spanish mullir “to hoe”. No modern forms of this verb reflect the original /o/. The same thing happened to subir. Its Latin source was sŭbīre; without the /j/ the short /ŭ/ of the stem would have turned into an /o/, i.e. sobir.

For more, see pp. 156-161 In the 1991 edition of Ralph Penny’s marvelous A History of the Spanish Language (unfortunately, not the edition pictured below, which has eliminated some detail.)

I wrote again about boot verbs a few days later, in this post.