In his 2003 book, Liddell gives an introduction to the common understanding of "compounds" in ASL, and then makes this cryptic statement (on p16):

If even Liddell says that what we typically call "compounds" aren't really compounds, then I think this deserves some discussion. I also have argued that many of the examples of compounds that people talk about in ASL aren't compounds at all. (While we are at it, I also don't believe in "morphemes", but I know what he means.) 

Let's say that "compounds" are combinations of two (or more) existing words whose job is to name some target referent, like English pumpkin pie or ASL SIGN LANGUAGE. In a compound, the two words work together to identify and refer to a subset of some category. A pumpkin pie is a specific type of 'pie', and SIGN LANGUAGE is a specific kind of 'language'.

In my 2016 conference proceedings paper, I identified a handful of examples of compounds in ASL videos online. These all combine two signs to "name" something. Unfortunately, I don't have links to videos of these ready to share easily (hashtag glossgesang!):

Talking about compounds also leads us to other topics in morphology and phonology. This is because individual compounds may change over time to give us new affixes, or they may undergo phonological change, as they become shared and used over and over:

"Bound forms" are parts of words that do not show up on their own as independent words. We know from historical research that bound forms often come from independent words, which get put next to another word, and then attached to that word, and then eroded down, as people use them a lot over time. Often the independent word is kept around, as well! An example in English is how we got "-ful" as in beautiful or powerful from the independent word full. An example in ASL is how we got the agent/person marker as in TEACHER and ACCOUNTANT from an independent sign meaning 'body/silhouette'. We do say full and PERSON on their own, but their "cousins" have become bound forms that don't appear on their own and always attach to something: [X-ful] and [X-agent/person].

Compounds also may undergo some reduction and erosion as people use them a lot, over time. An example in English is cupboard, in which we don't really pronounce the cup anymore, but we instead say "cubberd". An example in ASL is DEAF SCHOOL, where the sign DEAF is recognizable but a little reduced. Sometimes this reduction is quite advanced, and the original words/signs are almost completely obscured (as in English lady or ASL BROTHER). For many people, this is the most/only interesting property of compounds, for reasons I don't quite understand.

But other types of words that are not compounds also undergo degrees of erosion. A greeting like goodbye is eroded from the longer historical phrase "god be with you" and a contraction like could've is reduced from "could have". Similarly, the number 25 in ASL is reduced from the signs 20 and 5. But to me, it feels very odd to call greetings, contractions, and numbers like this, "compounds". They aren't used to name specific referents like compounds are.

Anyway, in other cases, we have some contrast between related words, because part of an existing word has been changed. This happens often in ASL morphology. We regularly change the movement or handshape of a word to create a related sign. But those handshapes and movements do not live on their own as independent signs (very strictly speaking). Similarly, all the "leftover" parts that do not change, also do not exist on their own as independent signs. This is for a very simple reason, that an independent sign requires its own full set of phonological information. But the parts that we identify in a pair of related signs like SIT and CHAIR or a pair of related signs like WEEK and THREE-WEEKS cannot really be split into their own full sets of phonological information.

I think that students are justified in wondering why we don't count the sign THREE as a "free" form, that combines with the sign WEEK to give us THREE-WEEKS. But really, what we observe is a similar handshape between the related signs THREE and THREE-WEEKS, and a similar "everything else" in WEEK and THREE-WEEKS. Like [X-ful] and [X-agent/person], these parts that we identify are "bound" elements because they aren't full signs on their own.

So! I think we can distinguish compounds as one kind of word-formation process (compounds are combinations of words whose job is to name a target referent), from other types of two-sign units (contractions, complex numbers), from reduction processes that affect many types of multiple-sign units (erosion), and from other word-formation processes such as changing part of a sign to make a new sign (nouns/verbs, number incorporation)!

posted Sept 22 2021, updated Mar 4, 2023