Borrowing words along with the articles or other grammatical parts (like Spanish from Arabic)

Question

Disclaimer: I do not know Arabic.

Here is an example of Spanish words of Arabic origin: alacrán, albañil, alquimia...

I wonder why Spanish language borrowed so many Arabic words along with the Arabic definite articles (the prefix 'al' in the examples). If I take for example English borrowings into Polish language I find no words which contain the English articles like 'the' or 'a'.

My questions are:

Why is this the case?

Are there any other examples of borrowings with articles/other grammatical markers?

I am guessing that:

• Arabic and English definite articles might actually work in quite a different way so, as for instance Arabic definite articles are more inseparable from words than their English counterparts

• Borrowing Arabic in medieval Spain was different than today's borrowing of English words by Polish. As some of us actually understand English and can skip the article. My intuition is that People in the Iberian Peninsula understood Arabic much less than Polish understand English nowadays, so the words Spanish borrowed are in their most common form, that is the form with the article included.

Answer 1

This is a linguistic process called rebracketing, and more specifically juncture loss.

Rebracketing is when word or morpheme boundaries are re-analyzed, especially when a word is borrowed from one language into another. The standard example of this is hamburger: the original German word was hamburg-er (because it came from the city of Hamburg), but in English it's been re-analyzed as ham-burger; this leads to new forms like "cheeseburger" and "chickenburger".

Juncture loss is a specific subset of this: when the boundary between two words is lost, and they become one. For example, Arabic al-kuħl "antimony" (two morphemes) to English alcohol (one morpheme).

This phenomenon isn't limited to Arabic: consider English the alligator, from Spanish el lagarto "the lizard"; English the hoi polloi, from Ancient Greek for "the people". But it's far-and-away more common with Arabic words than others.

One reason for this is that the definite article in Arabic is much more widely-used than in English or Spanish, and definitely more common than in Latin (which had no articles at all until well into the Romance period). For example, none of these languages tend to use the definite article in front of proper names. So the name al-Khwārizmī was borrowed into Latin as Algorismi, hence algorithm. This happens in reverse as well: the Greek name Alexander was analyzed in Arabic as al-Iskandar.

In addition, the Arabic definite article is much more closely attached to the noun than in English. A noun and article are written as a single word, and the last letter of the article tends to assimilate to the first letter of the noun (look up "sun and moon letters" for more info on this).

Closely-joined morphemes are known to make this sort of rebracketing much more likely to happen: the same thing often happens in Swahili, where nouns have prefixes indicating their class/gender. One type of inanimate object is marked with ki- in the singular and vi- in the plural, so the Arabic word kitāb "book" became kitabu/vitabu, and the English word "video" became kideo/video.

And finally, as you surmised, the medieval Europeans who were borrowing many of these words were not particularly familiar with Arabic. The speakers of European languages tend to be more familiar with each other, and thus rebracketing is less likely in loanwords—though it can still happen, as in hamburger and alligator.

Answer 2

Cases like this don't even require borrowing.

Consider English "newt". This is a native English word, but it's got half an article stuck on it. Old English "efte" became Middle English "ewte", but later, "an ewte" was reanalyzed a "a newte".

The same thing also happens in the opposite direction: Old English "naedre" became Middle English "naddere", but later "a naddere" was rebracketed as "an addere", hence modern "adder".¹

---

Obviously, Middle English speakers weren't unfamiliar with Middle English.

So, does that demolish your guess, or Draconis's answer? Not at all. It's pretty clear that this kind of rebracketing is much more common in Arabic->Spanish borrowings than in language-internal evolution. There are a handful of cases like "newt" and "adder", but many dozens of cases like "alcohol" and "algorithm".

Sometimes people hear unfamiliar words even in their own language.² This is especially true for words with dialectal variation.³ It isn't that common that a rebracketing or similar error spreads through the language, but it happens occasionally. Now consider a Spanish trader with a little Arabic⁴—it's a lot easier for him to mishear something, and there's a lot less pressure from other Spaniards correcting the error, so it's much more likely that his reanalysis will spread.

---

For the reasons you suggest, English-Polish borrowings don't have most of the same factors as Arabic-Spanish borrowings that make this kind of reanalysis more likely. And Adam Bittlingmayer's answer suggests further differences.

Also, consider that many modern borrowings from English are triggered by exposure through TV, movies, music, games, etc. If you mishear something from a trader who comes to your village, you're probably the only one who heard it; if you mishear something from a TV show broadcast at prime time, half your friends saw the exact same TV show. And, even if you all mishear it, a good chunk of the larger Polish-speaking community also saw it, so it's less likely to propagate from your group.

Also, many Poles who speak English are also literate in English, and in fact read quite a bit of English (internet forums, game on-screen prompts, etc., even if they don't sit around reading Dickens all day). It's a lot harder to mix up "an ewte" and "a newte" in writing than in speech. That's even more true given English's orthographic convention of clearly separating words. Especially since Polish has the exact same convention. And even more so when 99% of what you're reading is digital or typeset, where the spacing is perfectly regular. Compare all of that to a medieval Spanish monk reading Arabic calligraphy, and it seems obvious that your literacy should counter rebracketing a lot more than his did.

---

For extra fun, look at "orange".⁵ Going from Dravidian to Sanskrit to Persian to Arabic "naranj", there was surprisingly little change, so where did English lose the "n"? It looks like a case of Arabic-Spanish article rebracketing, and it also looks just like English "adder", but it's actually neither. While Italian and Occitan both rebracketed it, losing the "n", Spanish borrowed it as "naranje", and French borrowed it from Spanish, not its nearer neighbors. But then, French calqued Italian "mela arancia" as "pome narenge", and then reanalyzed the "n" away and ended up with "pome orenge". And "orenge" is what got borrowed into English.

---

_{1. Dutch did the exact same thing with the same word, but Frisian, even closer to English, did not, and ended up with "njirre".}

_{2. There are other ways it can happen. For example, one theory on "naunt" was that people jokingly pretended to misunderstand "mine aunt" as "my naunt", it spread as slang, and eventually it became a perfectly normal colloquial word (although a few centuries later it was lost, at least in standard English).}

_{3. See Caxton's famous case with "egg" for a good illustration.}

_{4. Again, this isn't the only possibility. Consider a Spanish monk reading Arabic and Greek treatises in parallel, who's rarely heard Arabic spoken aloud.}

_{5. Some steps of this are disputed, and I might be pushing a now-discredited theory at some point along the way… but whatever happened is interesting.}

Answer 3

Here is an example of Spanish words of Arabic origin: alacrán, albañil, alquimia...

Yes, it's curious. But it was not just Spanish - alembic, alcohol, algorithm and elixir had the article added by Greek or Mediaeval Latin.

Why is this the case?

In linguistic evolution, it is often really hard or impossible to say why. I think it's fair to say that the Iberian Romance speakers probably did not speak Arabic very well.

Are there any other example of borrowings with articles/other grammatical markers?

Yes, there are examples, even examples of whole sentences like vasistas, imam bayildi or je ne sais quoi being understood or used as a noun, even in the source language.

Arabic and English definite articles might actually work in quite a different way so, as for instance Arabic definite articles are more inseparable from words than their English counterparts.*

Actually, Arabic articles are not so so different than modern Romance articles, and el and al (or el in some dialects) are superficially quite similar.

The main difference compared to Iberian Romance is that the adjectives also take an article. (kabir in Guadalquivir is an adjective.)

If we consider that your example target language, Polish, has no articles at all, like many languages, the potential for this sort of mistake between Arabic and Spanish is lower. And many many other languages took massive contributions from Arabic, but few of them introduced this error so consistently.

The thing is, Latin does not have definite articles, and Latin was still used as a written language in Spain in mediaeval times. It could be that the words were borrowed before that shift was formalised.

My intuition is that People in the Iberian Peninsula understood Arabic much less than Polish understand English nowadays, so the words we incorporated in its most common form, that is the form with the article included.

Your intuition is logical, it seems like a necessary but not sufficient condition for this type of error. But there are some other potential factors:

1. Polish and English are written in the same alphabet, and most Poles are literate, so the word boundary is obvious to Polish speakers.

2. English is a prestige language in Poland and internationally, not unwelcome, and the elite will laugh at those who butcher it too much.

3. Many of the Arabic words in Iberia were place names, that is, proper nouns. Polish does use articles in cases like The Beatles.

...

It's notable that just as many of the Arabic words did not have this article added when they were borrowed into Spanish, Portuguese and Ladino. It could be worth analysing the characteristics of both groups, and perhaps a variable - the part of speech, the phonetics, the era, the domain... - will reveal itself.

Answer 4

The same thing occurs in Berber. The Arabic definite article is often borrowed with the noun. The Arabic word "al xadma", meaning "the work", is used in Riffian and is pronounced "lxedmet" while this language does not have articles in its linguistic set (see Kossmann 2013).

This phenomenon can be also found when the borrowing is made between French and Berber, thus the syntagma "la police" (=the police) will be used with its article in Riffian "lpulis".