Here is an example of a diachronic process which has been adduced as instantiating degrammaticalization: Modern High German has a word zig used as in .

The word means ‘umpteen’ and is distributed like a numeral. The latter are free words in German. The ancestor of the word is a Pre-German noun *tigus ‘decade, unit of ten’, which is documented in Gothic.

The free pseudo-numeral of does not appear in Old and Middle High German texts and has first been documented for Modern High German (no information on its first documentation is available).

On the other hand, the same morpheme is a suffix which derives the numerals for the tens, from zwanzig ‘twenty’ up to neunzig ‘ninety’. This suffix has been documented continually since Old High German (~ 1000 a.d.). It thus appears that the suffix has been recently separated from all those numerals of tens to designate a quantity in the tens. This would be a relatively good example of degrammaticalization.

However, the proof presupposes that the free numeral zig did not exist in the language during the Old and Middle High German period. Now the non-existence of something in an infinite space – and language is an infinite space – is not provable. Specifically in this case, zig in the use of is a neatly colloquial form which has very little chance to occur in the limited corpus of literary Old and Middle High German texts. It is perfectly possible that it has existed in the colloquial language all the time. This is, thus, an example of degrammaticalization which fails for methodological reasons.

There is, however, an analogous Italian example which has better chances to stand up to historical check of the documentation ().

The Italian numerals that are multiples of ten are derived by the suffix -anta from fourty on; and this has been so (with minor phonological changes) since Latin. A free form anta is first documented in the 20th century. Colloquial Modern Italian is well documented in the modern era so that one may be relatively sure that free anta was detached from those numerals in the 20th century. This is, thus, a historical example which comes rather close to the idea of degrammaticalization, taking into account the circumstance that a derivational affix is less clearly grammatical than an inflectional affix.

Another example which at least shows one aspect of degrammaticalization is the use of gender in “gendered language” in those languages which have the grammatical category of gender. By the middle of the 20th century, this was completely grammaticalized. Subsequently, it was resemanticized to designate sex by speakers supporting feminism. Other components of typical degrammaticalization are missing, though.

Uncontroversial cases of degrammaticalization like these are rare. They are no more than perhaps a dozen, in the face of hundreds of uncontroversial historical cases of grammaticalization. Anyway, even if they are few, they would falsify the unidirectionality hypothesis if this were put forward as a universal law. However, by the above wording, it is an overwhelming tendency. We are coming to the methodological status of the exceptions in a later section.