Strong vs. weak nouns and adjectives

Originally, adjectives in Proto-Indo-European followed the same declensional classes as nouns. The most common class (the o/ā class) used a combination of o-stem endings for masculine and neuter genders and ā-stems ending for feminine genders, but other common classes (e.g. the i class and u class) used endings from a single vowel-stem declension for all genders, and various other classes existed that were based on other declensions. A quite different set of "pronominal" endings was used for pronouns, determiners, and words with related semantics (e.g., "all", "only").

An important innovation in Proto-Germanic was the development of two separate sets of adjective endings, originally corresponding to a distinction between indefinite semantics ("a man") and definite semantics ("the man"). The endings of indefinite adjectives were derived from a combination of pronominal endings with one of the common vowel-stem adjective declensions - usually theo/ā class (often termed the a/ō class in the specific context of the Germanic languages) but sometimes the i or u classes. Definite adjectives, however, had endings based on n-stem nouns. Originally both types of adjectives could be used by themselves, but already by Proto-Germanic times a pattern evolved whereby definite adjectives had to be accompanied by a determiner with definite semantics (e.g., a definite article, demonstrative pronoun, possessive pronoun, or the like), while indefinite adjectives were used in other circumstances (either accompanied by a word with indefinite semantics such as "a", "one", or "some" or unaccompanied).

In the 19th century, the two types of adjectives - indefinite and definite - were respectively termed "strong" and "weak", names which are still commonly used. These names were based on the appearance of the two sets of endings in modern German. In German, the distinctive case endings formerly present on nouns have largely disappeared, with the result that the load of distinguishing one case from another is almost entirely carried by determiners and adjectives. Furthermore, due to regular sound change, the various definite (n-stem) adjective endings coalesced to the point where only two endings (-e and -en) remain in modern German to express the sixteen possible inflectional categories of the language (masculine/feminine/neuter/plural crossed with nominative/accusative/dative/genitive - modern German merges all genders in the plural). The indefinite (a/ō-stem) adjective endings were less affected by sound change, with six endings remaining (-, -e, -es, -er, -em, -en), cleverly distributed in a way that is capable of expressing the various inflectional categories without too much ambiguity. As a result, the definite endings were thought of as too "weak" to carry inflectional meaning and in need of "strengthening" by the presence of an accompanying determiner, while the indefinite endings were viewed as "strong" enough to indicate the inflectional categories even when standing alone. (This view is enhanced by the fact that modern German largely uses weak-ending adjectives when accompanying an indefinite article, and hence the indefinite/definite distinction no longer clearly applies.) By analogy, the terms "strong" and "weak" were extended to the corresponding noun classes, with a-stem and ō-stem nouns termed "strong" and n-stem nouns termed "weak".

However, in Proto-Germanic - and still in Gothic, the most conservative Germanic language - the terms "strong" and "weak" are not clearly appropriate. For one thing, there were a large number of noun declensions. The a-stem, ō-stem, and n-stem declensions were the most common and represented targets into which the other declensions were eventually absorbed, but this process occurred only gradually. Originally the n-stem declension was not a single declension but a set of separate declensions (e.g., -an, -ōn, -īn) with related endings, and these endings were in no way any "weaker" than the endings of any other declensions. (For example, among the eight possible inflectional categories of a noun — singular/plural crossed with nominative/accusative/dative/genitive — masculine an-stem nouns in Gothic include seven endings, and feminine ōn-stem nouns include six endings, meaning there is very little ambiguity of "weakness" in these endings and in fact much less than in the German "strong" endings.) Although it is possible to group the various noun declensions into three basic categories — vowel-stem,n-stem, and other-consonant-stem (aka "minor declensions") — the vowel-stem nouns do not display any sort of unity in their endings that supports grouping them together with each other but separate from the n-stem endings.

It is only in later languages that the binary distinction between "strong" and "weak" nouns become more relevant. In Old English, the n-stem nouns form a single, clear class, but the masculine a-stem and feminine ō-stem nouns have little in common with each other, and neither has much similarity to the small class of u-stem nouns. Similarly, in Old Norse, the masculine a-stem and feminine ō-stem nouns have little in common with each other, and the continuations of the masculine an-stem and feminine ōn/īn-stem nouns are also quite distinct. It is only in Middle Dutch and modern German that the various vowel-stem nouns have merged to the point that a binary strong/weak distinction clearly applies.

As a result, newer grammatical descriptions of the Germanic languages often avoid the terms "strong" and "weak" except in conjunction with German itself, preferring instead to use the terms "indefinite" and "definite" for adjectives and to distinguish nouns by their actual stem class.

In English, both two sets of adjective endings were lost entirely in the late Middle English period.

Linguistic developments

The subgroupings of the Germanic languages are defined by shared innovations. It is important to distinguish innovations from cases of linguistic conservatism. That is, if two languages in a family share a characteristic that is not observed in a third language, that is evidence of common ancestry of the two languages only if the characteristic is an innovation compared to the family's proto-language.

The following innovations are common to the Northwest Germanic languages (all but Gothic):

· The lowering of /u/ to /o/ in initial syllables before /a/ in the following syllable ("a-Umlaut", traditionally called Brechung)

· "Labial umlaut" in unstressed medial syllables (the conversion of /a/ to /u/ and /ō/ to /ū/ before /m/, or /u/ in the following syllable)^[37]

· The conversion of /ē₁/ into /ā/ (vs. Gothic /ē/) in stressed syllables.^[38] In unstressed syllables, West Germanic also has this change, but North Germanic has shortened the vowel to /e/, then raised it to /i/. This suggests it was an areal change.

· The raising of final /ō/ to /u/ (Gothic lowers it to /a/). It is kept distinct from the nasal /ǭ/, which is not raised.

· The monophthongisation of /ai/ and /au/ to /ē/ and /ō/ in non-initial syllables (however, evidence for the development of /au/ in medial syllables is lacking).

· The development of an intensified demonstrative ending in /s/ (reflected in English "this" compared to "the")

· Introduction of a distinct ablaut grade in Class VII strong verbs, while Gothic uses reduplication (e.g. Gothic haihait; ON, OE hēt, preterite of the Gmc verb *haitan "to be called")^[39] as part of a comprehensive reformation of the Gmc Class VII from a reduplicating to a new ablaut pattern, which presumably started in verbs beginning with vowel or /h/^[40] (a development which continues the general trend of de-reduplication in Gmc^[41]); there are forms (such as OE dial. heht instead of hēt) which retain traces of reduplication even in West and North Germanic

The following innovations are also common to the Northwest Germanic languages but represent areal changes:

· Proto-Germanic /z/ > /r/ (e.g. Gothic dius; ON dȳr, OHG tior, OE dēor, "wild animal"); note that this is not present in Proto-Norse and must be ordered after West Germanic loss of final /z/

· Germanic umlaut

The following innovations are common to the West Germanic languages:

· Loss of final /z/. In single-syllable words, Old High German retains it (as /r/), while it disappears in the other West Germanic languages.

· Change of [ð] (fricative allophone of /d/) to stop [d] in all environments.

· Change of /lþ/ to stop /ld/ (except word-finally).^[42]

· West Germanic gemination of consonants, except r, before /j/. This only occurred in short-stemmed words due to Sievers' law. Gemination of /p/, /t/, /k/ and /h/ is also observed before liquids.

· Labiovelar consonants become plain velar when non-initial.

· A particular type of umlaut /e-u-i/ > /i-u-i/.

· The change of /b/ or /g/ to /w/ before nasal consonant.^[43]

· Changes to the 2nd person singular past-tense: Replacement of the past-singular stem vowel with the past-plural stem vowel, and substitution of the ending -t with -ī.

· Short forms (*stān, stēn, *gān, gēn) of the verbs for "stand" and "go"; but note that Crimean Gothic also has gēn.

· The development of a gerund.

The following innovations are common to the Ingvaeonic subgroup of the West Germanic languages languages, which includes English, Frisian, and to some extent Dutch and Low German, but not High German:

· The so-called Ingvaeonic nasal spirant law, with loss of /n/ before voiceless fricatives: e.g. *munþ, *gans > Old English mūþ, gōs > "mouth, goose", but German mund, gans.

· The loss of the Germanic reflexive pronoun.

· The reduction of the three Germanic verbal plural forms into one form ending in -þ.

· The development of Class III weak verbs into a relic class consisting of four verbs (*sagjan "to say", *hugjan "to think", *habjan "to have", *libjan "to live"; cf. the numerous Old High German verbs in -ēn).

· The split of the Class II weak verb ending *-ō- into *-ō-/-ōja- (cf. Old English -ian < -ōjan, but Old High Geman -ōn).

· Development of a plural ending *-ōs in a-stem nouns (note, Gothic also has -ōs, but this is an independent development, caused by terminal devoicing of *-ōz; Old Frisian has -ar, which is thought to be a late borrowing from Danish). Cf. modern English plural -(e)s, but German plural -e.

· Possibly, the monophthongization of Germanic *ai to ē/ā (this may represent independent changes in Old Saxon and Anglo-Frisian).

The following innovations are common to the Anglo-Frisian subgroup of the Ingvaeonic languages:

· Raising of nasalized a, ā into o, ō.

· Anglo-Frisian brightening: Fronting of non-nasal a, ā to æ,ǣ when not followed by n or m.

· Metathesis of CrV into CVr, where C represents any consonant and V any vowel.

· Monophthongization of ai into ā.

The peculiar Germanic system of word accentuation is one of the most important distinguishing features of the group; it arose in PG, was fully or partly retained in separate languages and served as one of the major causes for many linguistic changes.
It is known that in ancient IE, prior to the separation of Germanic, there existed two ways of word accentuation: musical pitch and force stress. The position of the stress was free and movable, which means that it could fall on any syllable of the word—a root-morpheme, an affix or an ending — and could be shifted both in form-building and word-building. Both these properties of the word accent were changed in PG.