Initial position

• Almost all words beginning with a dental fricative have /θ/.
• A small number of common function words (the Middle English anomalies mentioned below) begin with /ð/. The words in this group are:
  • 1 definite article: the
  • 4 demonstratives: this, that, these, those
  • 2 personal pronouns each with multiple forms: thou, thee, thy, thine, thyself; they, them, their, theirs, themselves, themself
  • 7 adverbs and conjunctions: there, then, than, thus, though, thence, thither (though in the United States thence and thither may be pronounced with initial /θ/^[1])
  • Various compound adverbs based on the above words: therefore, thereupon, thereby, thereafter, thenceforth, etc.
• A few words have an initial ⟨th⟩ for /t/ (e.g. Thomas): see below.

Medial position

• Most native words with a medial ⟨th⟩ have /ð/.
  • Between vowels (including r-colored vowels), followed by a weak vowel: heathen, farthing, fathom, Worthington; and the frequent combination -ther-: bother, brother, dither, either, farther, father, further, heather, lather, mother, northern, other, rather, smithereens, slither, southern, together, weather, whether, wither; Caruthers, Netherlands, Witherspoon.
  • Followed by /r/: brethren.
• A few native words have a medial /θ/:
  • The suffixes -y, -ly, -ing and -ed normally leave terminal /θ/ unchanged: earthy, healthy, pithy, stealthy, wealthy, bothy (from booth); fourthly, monthly; earthing; frothed; but worthy and swarthy have /ð/.
  • Some plurals have /θs/, as discussed in more detail below: cloths, baths etc.
  • Compound words in which the first element ends or the second element begins with ⟨th⟩ frequently have /θ/, as these elements would in isolation: bathroom, Southampton; anything, everything, nothing, something.
  • The only other native words with medial /θ/ would seem to be brothel (usually) and Ethel.
• Most loan words with a medial ⟨th⟩ have /θ/.
  • From Greek: Agatha, anthem, atheist, Athens, athlete, cathedral, Catherine, Cathy, enthusiasm, ether, ethics, ethnic, lethal, lithium, mathematics, method, methyl, mythical, panther, pathetic, sympathy
  • From Latin: author, authority (though in Latin these had /t/; see below). Also names borrowed from or via Latin: Bertha, Gothic, Hathaway, Othello, Parthian
  • From Celtic languages: Arthur (Welsh has /θ/ medially: /ærθɨr/); Abernathy, Abernethy, as an anglicization, though Gaelic has no /θ/.
  • From Hebrew: Ethan, Jonathan, Bethlehem, Bethany, Leviathan, Bethel
  • From German: Luther, as an anglicized spelling pronunciation (see below).
• Loanwords with medial /ð/:
  • Greek words with the combination -thm-: algorithm, logarithm, rhythm. Exception : arithmetic /əˈrɪθmətɪk/. The word asthma may be pronounced /ˈæzðmə/ or /ˈæsθmə/, though here the ⟨th⟩ is usually silent.
• A few words have a medial ⟨th⟩ for /t/ or /th/ (e.g. lighthouse): see below.

Final position

• Nouns and adjectives
  • Nouns and adjectives ending in a dental fricative usually have /θ/: bath, breath, cloth, froth, health, hearth, loath, mouth, sheath, sooth, tooth/teeth, width, wreath.
  • Exceptions are usually marked in the spelling with a silent ⟨e⟩: tithe, lathe, lithe with /ð/.
  • blithe can have either /ð/ or /θ/. booth has /ð/ in England but /θ/ in America.
• Verbs
  • Verbs ending in a dental fricative usually have /ð/, and are frequently spelled with a silent ⟨e⟩: bathe, breathe, clothe, loathe, scathe, scythe, seethe, sheathe, soothe, teethe, tithe, wreathe, writhe. Spelled without ⟨e⟩: mouth (verb) nevertheless has /ð/.
  • froth has /θ/ whether as a noun or as a verb.
  • The verb endings -s, -ing, -ed do not change the pronunciation of a ⟨th⟩ in the final position in the stem: bathe has /ð/, therefore so do bathed, bathing, bathes; frothing has /θ/. Likewise clothing used as a noun, scathing as an adjective etc.
  • The archaic verb inflection "-eth" has /θ/.
• Others
  • with has either /θ/ or /ð/ (see below), as do its compounds: within, without, outwith, withdraw, withhold, withstand, wherewithal, etc.

Plurals

• Plural ⟨s⟩ after ⟨th⟩ may be realized as either /ðz/ or /θs/:
  • Some plural nouns ending in ⟨ths⟩, with a preceding vowel, have /ðz/, although the singulars always have /θ/; however, a variant in /θs/ will be found for many of these: baths, mouths, oaths, paths, sheaths, truths, wreaths, youths exist in both varieties; clothes always has /ðz/ (if not pronounced /kloʊz/^[2]).
  • Others have only /θs/: azimuths, breaths, cloths, deaths, faiths, Goths, growths, mammoths, moths, myths, smiths, sloths, zeniths, etc. This includes all words in 'th' preceded by a consonant (earths, hearths, lengths, months, widths, etc.) and all numeric words, whether preceded by vowel or consonant (fourths, fifths, sixths, sevenths, eighths /eɪtθs/, twelfths, fifteenths, twentieths, hundredths /hʌndrədθs/, thousandths).
  • Booth has /ð/ in the singular and hence /ðz/ in the plural for most speakers in England. In American English, it has /θ/ in the singular and /θs/ or /ðz/ in the plural. This pronunciation also prevails in Scotland.

Grammatical alternation

In pairs of related words, an alternation between /θ/ and /ð/ is possible, which may be thought of as a kind of consonant mutation. Typically [θ] appears in the singular of a noun, [ð] in the plural and in the related verb: cloth /θ/, clothes /ð/, to clothe /ð/. This is directly comparable to the /s/-/z/ or /f/-/v/ alternation in house, houses or wolf, wolves. It goes back to the allophonic variation in Old English (see below), where it was possible for ⟨þ⟩ to be in final position and thus voiceless in the basic form of a word, but in medial position and voiced in a related form. The loss of inflections then brought the voiced medial consonant to the end of the word. Often a remnant of the old inflection can be seen in the spelling in the form of a silent ⟨e⟩, which may be thought of synchronically as a marker of the voicing.

Regional differences in distribution

The above discussion follows Daniel Jones' English Pronouncing Dictionary, an authority on standard British English, and Webster's New World College Dictionary, an authority on American English. Usage appears much the same between the two. Regional variation within standard English includes the following:

• The final consonant in with is pronounced /θ/ (its original pronunciation) in northern Britain, but /ð/ in the south, though some speakers of Southern British English use /θ/ before a voiceless consonant and /ð/ before a voiced one. A 1993 postal poll of American English speakers showed that 84% use /θ/, while 16% have /ð/ (Shitara 1993). (The variant with /ð/ is presumably a sandhi development.)
• In Scottish English, /θ/ is found in many words which have /ð/ further south. The phenomenon of nouns terminating in /θ/ taking plurals in /ðz/ does not occur in the north. Thus the following have /θs/: baths, mouths (noun), truths. Scottish English does have the termination /ðz/ in verb forms, however, such as bathes, mouths (verb), loathes, and also in the noun clothes, which can be realized without /ð/. Scottish English also has /θ/ in with, booth, thence etc., and the Scottish pronunciation of thither, almost uniquely, has both /θ/ and /ð/ in the same word. Where there is an American-British difference, the North of Britain generally agrees with the United States on this phoneme pair.
• Some dialects of American English use /ð/ at the beginning of the word "thank".

Germanic origins

Proto-Indo-European (PIE) had no dental fricatives, but these evolved in the earliest stages of the Germanic languages. In Proto-Germanic, /ð/ and /θ/ were separate phonemes, usually represented in Germanic studies by the symbols *đ and *þ.

• *đ (/ð/) was derived by Grimm's law from PIE *dʰ or by Verner's law (i.e. when immediately following an unstressed syllable) from PIE *t.
• *þ (/θ/) was derived by Grimm's law from PIE *t.

In West Germanic, the Proto-Germanic *đ shifted further to *d, leaving only one dental fricative phoneme. However, a new [ð] appeared as an allophone of /θ/ in medial positions by assimilation of the voicing of the surrounding vowels. [θ] remained in initial and presumably in final positions (though later terminal devoicing would in any case have eliminated the evidence of final [ð]). This West Germanic phoneme, complete with its distribution of allophones, survived into Old English. In German and Dutch, it shifted to a /d/, the allophonic distinction simply being lost. In German, West Germanic *d shifted to /t/ in what may be thought of as a chain shift, but in Dutch, *þ, *đ and *d merged into a single /d/.

The whole complex of Germanic dentals, and the place of the fricatives within it, can be summed up in this table:

Note that this table shows only the basic rules. The actual developments in all of the mentioned languages are more complicated (due to dialectal variation, peculiar developments in consonant clusters, etc.). For more on these phonemes from a comparative perspective, see Grammatischer Wechsel. For the developments in German and Dutch see High German consonant shift.

Old English

Thus English inherited a phoneme /θ/ in positions where other West Germanic languages have /d/ and most other Indo-European languages have /t/: English three, German drei, Latin tres.

In Old English, the phoneme /θ/, like all fricative phonemes in the language, had two allophones, one voiced and one voiceless, which were distributed regularly according to phonetic environment.

• [ð] (like [v] and [z]) was used between two voiced sounds (either vowels or voiced consonants).
• [θ] (like [f] and [s]) was spoken in initial and final position, and also medially if adjacent to another unvoiced consonant.

Although Old English had two graphemes to represent these sounds, ⟨þ⟩ (thorn) and ⟨ð⟩ (eth), it used them interchangeably, unlike Old Icelandic, which used ⟨þ⟩ for /θ/ and ⟨ð⟩ for /ð/.

Development up to Modern English

The most important development on the way to modern English was the investing of the existing distinction between [ð] and [θ] with phonemic value. Minimal pairs, and hence the phonological independence of the two phones, developed as a result of three main processes.

1. In early Middle English times, a group of very common function words beginning with /θ/ (the, they, there, etc.) came to be pronounced with /ð/ instead of /θ/. Possibly this was a sandhi development; as these words are frequently found in unstressed positions, they can sometimes appear to run on from the preceding word, which may have resulted in the dental fricative being treated as though it were word-internal. This allowed a word-initial minimal pair like thigh:thy.
2. English has borrowed many words from Greek, including a vast number of scientific terms. Where the original Greek had the letter ⟨θ⟩ (theta), English usually retained the Late Greek pronunciation regardless of phonetic environment, resulting in the presence of /θ/ in medial position (anthem, methyl, etc.). This allowed a medial minimal pair like ether:either.
3. English has lost its original verb inflections. When the stem of a verb ends with a dental fricative, this was usually followed by a vowel in Old English, and was therefore voiced. It is still voiced in modern English, even though the verb inflection has disappeared leaving the /ð/ at the end of the word. Examples are to bathe, to mouth, to breathe. Sometimes a remnant of the original vowel remained in the spelling (see: Silent e), but this was inconsistent. This allowed a minimal pair in final position like loath:loathe.

Other changes that affected these phonemes included a shift /d/ → /ð/ when followed by unstressed suffix -er. Thus Old English fæder became modern English father; likewise mother, gather, hither, together, weather (from mōdor, gaderian, hider, tōgædere, weder). In a reverse process, Old English byrþen and morþor or myþra become burden and murder (compare the obsolete variants burthen and murther).

Dialectally, the alternation between /d/ and /ð/ sometimes extends to other words, as bladder, ladder, solder with /ð/ (possibly being restricted elsewhere by the former two clashing with blather and lather). On the other hand, some dialects retain original d, and extend it to other words, as brother, further, rather. The Welsh name Llewelyn appears in older English texts as Thlewelyn (Rolls of Parliament (Rotuli parliamentorum) I. 463/1, King Edward I or II), and Fluellen (Shakespeare, Henry V). Th also occurs dialectally for wh, as in thirl, thortleberry, thorl, for whirl, whortleberry, whorl. Conversely, Scots has whaing, whang, white, whittle, for thwaing, thwang, thwite, thwittle.

The old verb inflection -eth (Old English -eþ) was replaced by -s (he singeth → he sings), not a sound shift but a completely new inflection, the origin of which is still being debated. Possibilities include alveolarization (since s is easier to pronounce there than th), or displacement by a nonstandard English dialect.

th-fronting

In some areas, such as London, and certain dialects, including African American Vernacular English and less commonly New Zealand, many people realize the phonemes /θ/ and /ð/ as [f] and [v], respectively. Although traditionally stigmatized as typical of a Cockney accent, this pronunciation is fairly widespread, especially when immediately surrounded by other fricatives for ease of pronunciation, and has, in the early 20th century, become an increasingly noticeable feature of the Estuary English accent of South East England. It has in at least one case been transferred into standard English as a neologism: a bovver boy is a thug, a "boy" who likes "bother" (fights). Joe Brown and his Bruvvers was a Pop group of the 1960s. The song "Fings ain't wot they used t'be" was the title song of a 1959 Cockney comedy. Similarly, a New Zealander from the northernmost parts of the country might state that he or she is from "Norfland".

Note that, at least in Cockney, a word beginning with /ð/ (as opposed to its voiceless counterpart /θ/) can never be labiodental. Instead, it is realized as any of [ð, ð̞, d, l, ʔ], or is dropped altogether.^[3][4]

th-stopping

Many speakers of African American Vernacular English, Caribbean English, Liberian English, Nigerian English, Philadelphia English, and Philippine English (along with other Asian English varieties) pronounce the fricatives /θ, ð/ as alveolar stops [t, d]. Similarly, but still distinctly, many speakers of New York City English, Chicago English, Boston English, Indian English, Newfoundland English, and Hiberno-English use the dental stops [t̪, d̪] (typically distinct from alveolar [t, d]) instead of, or in free variation with, [θ, ð].

In Cockney, the th-stopping may occur when a word begins with /ð/ (but not its voiceless counterpart /θ/).^[3][4] This is also associated with the accent of the English city of Sheffield (such as the nickname dee-dahs for residents) but such pronunciations are now confined to the very oldest residents of Sheffield.^[5]

th-alveolarization

Th-alveolarization is a process that occurs in some African varieties of English where the dental fricatives /θ, ð/ merge with the alveolar fricatives /s/ and /z/. It is an example of assibilation.

In rarer or older varieties of African American Vernacular English, /θ/ may be pronounced [s] after a vowel and before another consonant, as in bathroom [ˈbæsɹum].^[6]

Th-alveolarization is often parodied as typical of French- and German-speaking learners of English, but it is widespread among many other foreign learners because the dental fricative "th" sounds are not very common among the world's languages. Due to the said ridicule, learners who are unable to realize these sounds sometimes opt for the less marked th-fronting or th-stopping instead of alveolarization.

th-debuccalization

In many varieties of Scottish English, /θ/ becomes [h] word initially and intervocalically.^[7]

Th-debuccalization occurs mainly in Glasgow and across the Central Belt. A common example is [hɪŋk] for think. This feature is becoming more common in these places over time, but is still variable. In word final position, [θ] is used, as in standard English.

The existence of local [h] for /θ/ in Glasgow complicates the process of th-fronting there, a process which gives [f] for historical /θ/. Unlike in the other dialects with th-fronting, where [f] solely varies with [θ], in Glasgow, the introduction of th-fronting there creates a three-way variant system of [h], [f] and [θ].

Use of [θ] marks the local educated norms (the regional standard), while use of [h] and [f] instead mark the local non-standard norms. [h] is well known in Glasgow as a vernacular variant of /θ/ when it occurs at the start of a word and intervocalically, while [f] has only recently risen above the level of social consciousness.

Given that th-fronting is a relatively recent innovation in Glasgow, it was expected that linguists might find evidence for lexical diffusion for [f] and the results found from Glaswegian speakers confirm this. The existing and particular lexical distribution of th-debuccalization imposes special constraints on the progress of th-fronting in Glasgow.

In accents with th-debuccalization, the cluster /θr/ becomes [hr], giving these dialects a consonant cluster that does not occur in other dialects. The replacement of /θr/ with [hr] leads to pronunciations like:

• three – [hri]
• throw – [hro]
• through, threw – [hrʉ]
• thrash – [hraʃ]
• thresh – [hrɛʃ]
• thrown, throne – [hron]
• thread – [hrɛd]
• threat – [hrɛt]

⟨th⟩ for /θ/ and /ð/

Though English speakers take it for granted, the digraph ⟨th⟩ is in fact not an obvious combination for a dental fricative. The origins of this have to do with developments in Greek.

Proto-Indo-European had an aspirated /dʱ/ that came into Greek as /tʰ/, spelled with the letter theta. In the Greek of Homer and Plato, this was still pronounced /tʰ/, and therefore when Greek words were borrowed into Latin, theta was transcribed with ⟨th⟩. Since /tʰ/ sounds like /t/ with a following puff of air, ⟨th⟩ was the logical spelling in the Latin alphabet.

By the time of New Testament Greek (koiné), however, the aspirated stop had shifted to a fricative: /tʰ/→/θ/. Thus theta came to have the sound that it still has in Modern Greek, and which it represents in the IPA. From a Latin perspective, the established digraph ⟨th⟩ now represented the voiceless fricative /θ/, and was used thus for English by French-speaking scribes after the Norman Conquest, since they were unfamiliar with the Germanic graphemes ð (eth) and þ (thorn). Likewise, the spelling ⟨th⟩ was used for /θ/ in Old High German prior to the completion of the High German consonant shift, again by analogy with the way Latin represented the Greek sound. It also appeared in early modern Swedish before a final shift to /d/.

The history of the digraphs ⟨ph⟩ for /f/ and ⟨ch⟩ for Scots, Welsh or German /x/ is parallel.

⟨th⟩ for /t/

Since neither /tʰ/ nor /θ/ was a native sound in Latin, the tendency emerged at the latest in medieval Latin, to substitute /t/. Thus, in many modern languages, including French and German, the ⟨th⟩ digraph is used in Greek loan-words to represent an original /θ/, but is now pronounced /t/: examples are French théâtre, German Theater. In some cases, this etymological ⟨th⟩, which has no remaining significance for pronunciation, has been transferred to words in which there is no etymological justification for it. For example, German Tal ('valley', cognate with English dale) appears in many place-names with an archaic spelling Thal (contrast Neandertal and Neanderthal). The German spelling reform of 1901 largely reversed these, but they remain in some proper nouns. The name Rothschild is an example of this, being a compound of rot[h] ("red") and Schild ("shield").

Examples of this are also to be found in English, perhaps influenced immediately by French. In some Middle English manuscripts, ⟨th⟩ appears for ⟨t⟩ or ⟨d⟩: tho 'to' or 'do', thyll till, whythe white, thede deed. In Modern English we see it in Esther, Thomas, Thames, thyme, Witham (the town in Essex, not the river in Lincolnshire which is pronounced with /ð/) and the old spelling of Satan as Sathan. More recently, the name of the capital of Nepal was often written Katmandu down to the late 20th century, but is now usually spelt Kathmandu.

In a small number of cases, this spelling later influenced the pronunciation: amaranth, amianthus and author have spelling pronunciations with /θ/, and some English speakers use /θ/ in Neanderthal.

⟨th⟩ for /th/

A few English compound words, such as lightheaded or hothouse, have the letter combination ⟨th⟩ split between the parts, though this is not a digraph. Here, the ⟨t⟩ and ⟨h⟩ are pronounced separately (light-headed) as a cluster of two consonants. Other examples are anthill, goatherd, lighthouse, outhouse, pothead; also in words formed with the suffix -hood: knighthood, and the similarly formed Afrikaans loanword apartheid. In a few place names ending in t+ham, the t-h boundary has been lost and become a spelling pronunciation, for example Grantham.

Citations

Sources