Bound forms and welded forms: Two basic concepts of general grammar

Linguists often try to characterize affixes in terms of a notion of “boundness“, as in this passage of the Wikipedia article “affix“:

“Lexical affixes are bound elements that appear as affixes, but function as incorporated nouns within verbs”

But what exactly is meant by “bound”? Is it just a synonym of “attached” (as in Wikipedia’s definition of affix: “an affix is a morpheme that is attached to a word stem to form a new word or word form”), and are both terms just informal ways of saying that an affix occurs together with and adjacent to a stem?

But if so, what is the difference between an affix occurring together with a stem, and a verb occurring together with (and adjacent to) an object nominal? Linguists don’t normally say that objects are bound to their verbs, but in what sense is an affix bound but an object nominal is not bound?

This seems like a very basic question, but linguists do not have consistent answers to this question. The first answer they give often involves the notion of “word“, but linguists have not found a general way of identifying words across languages (as I noted in my 2011 article). If at all, words can perhaps be characterized in terms of a more basic concept such as “bound”, but not the other way round.

So here I would like to make a proposal how to use the term bound which also involves the introduction of a new term: welded. Basically, I propose that a bound form is one that cannot occur in isolation, while a welded form is one that shows segmental phonological interaction with its neighbour. (An earlier version of this blogpost was posted on the Lingtyp list.)

For example, the English preposition from, the possessive pronoun my, and the definite article the are bound forms. Consider the following contrasts, where (b) is possible but (d) is not possible.

(1) a. The dog went through the fence.

b. The dog went through.

c. The cat jumped from the table.

d. *The cat jumped from.

(2) a. I saw Kim’s bike.

b. I saw Kim’s.

c. She found my umbrella.

d. *She found my.

(3) a. We like those caps.

b. We like those.

c. He bought the cap.

d. *He bought the. (‘He bought it.’)

Of course, most forms that are written as affixes are also bound, but as the examples (1)-(3) show, not all forms that are written separately are free in the sense that they can occur on their own. (This is why one cannot say that a form needs to be free in order to be a word.)

The definition of bound in terms of non-occurrence in isolation goes back to Bloomfield (1933: 160) (who seemed to have coined the term), and it has become particularly well-known through Arnold Zwicky’s work on clitics. Zwicky’s famous (1977) paper distinguishes three classes of elements: simple clitics, special clitics, and bound words. The most widely cited paper that proposes criteria for distinguishing between clitics and affxes, Zwicky & Pullum (1983), begins as follows:

“Two types of bound morphemes are found attached to (free) words in many languages: clitics and affixes”

Affixes are always thought of as parts of words, while clitics are generally thought of as words. Thus, the notion of boundness cross-cuts the distinction between words and parts of words.

In the discussions about wordhood that I often have with fellow grammarians, they often mention phonological interaction: Some elements interact phonologically with their neighbours, while others don’t. Some contrasts are given in (4)-(6). I propose to say that the forms in (b) are welded, while those in (a) are not.

(4) a. my pear / my apple

b. a pear / an apple

(5) a. German film-te/ golf-te

b. English film-ed [-d] / golf-ed [-t]

(6) a. good / good-ness

b. mortal [-əl] / mortal-ity [-æl-iti]

In (4)-(5), we see that some (but not all) of the bound forms have different phonological variants depending on phonological properties of their host, and in (6b), we see that the host can have different phonological variants depending on whether it combines with a bound form or not. A bound form is welded to its host if it shows different variants depending on the shape of the host or if the host shows different variants depending on the shape of the bound form.

So clearly, boundness and weldedness are different properties of linguistic form: A form may be bound but not welded (e.g. English my, -ness, German -te), or a form may be both bound and welded (e.g. English a/an, -ed [-t/-d]).

Can a form be welded but not bound? No, this is excluded by definition: If a form can be used either on its own or in combination with a bound form, and if it has a different shape when combined with the bound form, then we do not say that the form as such is welded. For example, the English verb write can be used on its own (e.g. as an imperative), and when the agent noun suffix -er is added, it may have a different shape (with flapped r): wri[ɾ]-er. Thus, -er is a welded form, but we would not want to say that write is a welded form.

Some linguists use a different term for what I am calling “welded” here: “phonologically bound“. This seems to occur fairly frequently, but unlike the use of bound in the sense ‘free-standing’, it does not seem to have a clear pedigree. I have not even found a clear statement in the literature concerning the meaning of the term. Textbooks or handbook aricles do not introduce and explain it, as far as I have been able to determine. It seems to have the same status as the term “word”: It is a widespread stereotype and many people think that it has a clear meaning, but in fact, it doesn’t.

Note that I used “general grammar” in the heading of this post, to make it clear that I am not talking about language-particular categories. There are many potential language-particular criteria that can (or must) be used to set up language-particular categories, but in order to compare phenomena across languages, we need general concepts (also called comparative concepts). These must be defined in terms of concepts that are equally applicable to all languages, which rules out stress or tone as criteria for welding (because these cannot be readily identified in any language, in contrast to segmental changes). In a Lingtyp comment, Adam Tallman observed that the term “phonologically bound” may also refer to elements which are included in a particular phonological domain, but again, these domains are always language-particular.

Finally, how does weldedness related to “bondedness”? This is a term that was introduced by Lehmann (1982 [2015: §4.3.2]), and has since been used on and off in the grammaticalization literature (e.g. in three of the chapters of Narrog & Heine 2018). Unfortunately, Lehmann did not defined bondedness very clearly:

“The syntagmatic cohesion or bondedness of a sign is the intimacy with which it is connected to another sign to which it bears a syntagmatic relation. The degree of bondedness varies from juxtaposition to merger…”

Lehmann then mentions the well-known scale “juxtaposition > cliticization > agglutination > merger/fusion”, but he seems to presuppose that we know what these are, and I have noted in earlier work that neither the term “clitic” nor the term “agglutination” are clearly defined or motivated (Haspelmath 2009; 2015). So even though nobody doubts that Lehmann is onto something, we cannot make it precise in these terms. The new term weldedness is clearly needed.


Bloomfield, Leonard. 1933. Language. New York: H. Holt and Company.

Haspelmath, Martin. 2009. An empirical test of the Agglutination Hypothesis. In Sergio Scalise, Elisabetta Magni & Antonietta Bisetto (eds.), Universals of language today, 13–29. Dordrecht: Springer. (27 May, 2016).

Haspelmath, Martin. 2011. The indeterminacy of word segmentation and the nature of morphology and syntax. Folia Linguistica 45(1). 31–80.

Haspelmath, Martin. 2015. Defining vs. diagnosing linguistic categories: A case study of clitic phenomena. In Joanna Błaszczak, Dorota Klimek-Jankowska & Krzysztof Migdalski (eds.), How categorical are categories? New approaches to the old questions of noun, verb, and adjective, 273–304. Berlin: De Gruyter Mouton.

Lehmann, Christian. 2015. Thoughts on grammaticalization. Berlin: Language Science Press.

Narrog, Heiko & Bernd Heine (eds.). 2018. Grammaticalization from a typological perspective. Oxford: Oxford University Press.

Zwicky, Arnold M. 1977. On clitics. Bloomington: Indiana University Linguistics Club.

Zwicky, Arnold M. & Geoffrey K. Pullum. 1983. Cliticization vs. inflection: English n’t. Language 502–513.


Cite this blog post
Martin Haspelmath (2019, May 17). Bound forms and welded forms: Two basic concepts of general grammar. Diversity Linguistics Comment. Retrieved May 30, 2024, from

2 thoughts on “Bound forms and welded forms: Two basic concepts of general grammar

  1. The description of English “from” as a bound form seems to require not counting preposition stranding as as use of the preposition in isolation. “From” can occur unattached to any adjacent element in sentences like “Where did you get that from?” Does it not count as isolated in contexts like this because it is in a syntactic relationship with the non-adjacent word “where”?

    • Yes, this is interesting question. I would say that “from” is not a free form, because it cannot occur in complete isolation (e.g. it cannot occur as an answer to a question). But most bound forms are “bound to something”, i.e. they occur in a position that can be specified with respect to a type of host. English “from” is different in this regard, in that it can occur in rather different positions. There is no traditional term for this sort of element. (Using the term “clitic” here would be confusing, because it does not correspond to a traditional clitic stereotype. But I don’t know of a definition of “clitic” that includes most stereotypical clitics but excludes “from”.)

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.