That or Which?: The that’s that of which is which


By Soojin Lee

© 2006




That and which as relative pronouns


The relative pronouns that and which are used to connect a dependent clause to an independent clause and serve as subject or object in the dependent clause.  They do so by relating to the antecedent in the sentence (1).


   (1)   This is a flower that I have always had difficulty growing in my garden.

            Independent clause: This is a flower

                Dependent clause: that I have always had difficulty…

Antecedent: flower

Relative pronoun: that


Proper usage of that and which


The dependent clause introduced by the relative pronoun is either restrictive (defining) or non-restrictive (continuative).  A restrictive clause provides information that is critical to the understanding of the main clause (2)(3), whereas a non-restrictive clause provides information that is not critical to the understanding of the main clause (4)(5).  In the case of a restrictive clause, that is the appropriate relative pronoun; and in a non-restrictive clause, which is the proper choice.  The non-restrictive clause is set off with commas.


(2)       Daria wants to compose a song that will reflect her love for grammar.


   (3)   I want her to give me back the book  that I loaned her a year ago.


(4)   Kumar will never return his gold medal, which he believes he won fairly.


(5)   Mimi’s physics textbooks, which she bought from a used bookstore, contained               many typos.


Note that the “proper” usage is a top-down prescriptive rule and not a matter of descriptive linguistics.  In fact, that and which are widely used interchangeably by many people and contemporary usage guides are less likely to “prescribe” this strict usage (see “Usage Guides Today” below). 




The heavy functional load of that


Old English used the particle þe as a relative pronoun, and to a lesser degree, þæt was also used as a relative.  The relative particle þe was still actively in use in early Middle English alongside þat (that as it would have been written then), but well into the Middle English period, þat ‘that’ completely replaced þe and became the most frequently used relative pronoun (Millward 148).  This move towards simplification was complicated by the introduction of wh- forms (interrogative pronouns which, who, whose, etc.) in a relative function (Romaine 102).  Consequently, that would be undermined as the uniform relative marker.  Immediately, þat began to be limited to restrictive clauses when wh- pronouns adopted a relative function (Romaine 102).   


During EModE, that continued to be the most widely used relative pronoun as it was used as both a human and nonhuman referent, and used to introduce both nonrestrictive and restrictive clauses.  Matti Rissanen believes there is little doubt that the rise in wh- pronouns was largely a result of the heavy functional load of that: “When the connection between the antecedent and the relative link was loose, the likelihood of ambiguity and misunderstanding of the meaning of that increased” (CHEL III, 295). 


The rise of wh- pronouns


The transition from interrogative to relative wh- pronoun began after the pronouns started losing their interrogative qualities in indirect questions (Mustanoja 192).  Eventually, the generalizing relative pronouns very obviously referred to an antecedent and adopted a permanent relative function.  At the end of the Middle English period, that was the most common adnominal relative link, although which was preferred for non-restrictive clauses.   


The function of the relative pronoun in the clause played an important role in the choice of its form.  That is most resistant to replacement by wh-pronouns in subject position (Dekeyser, 73).  This means that the spread of wh-pronouns was a top-down change from formal and literary circles of language.   Yet, text representing oral usage illustrate the robustness of that despite the rise of wh-pronouns in written language (Dekeyser 62).   


To compensate for the heavy functional load of that, wh-pronouns appeared more frequently.  By the fourteenth century, which was the most frequent interrogative used relatively, and was employed with both human and nonhuman antecedents.  Wh-pronouns are well established in all types of non-restrictive relative clauses by early sixteenth century, and even finding their way into restrictive relative clauses at approximately 25% by mid-century (CHEL III The Latin influence during the Renaissance is another possible factor with the prevalence of the Latin relative pronoun paradigm (qui, quae, quod, etc.), but which, whose, whom, who were already frequently used in Middle English period making any influence minimal.  The inflected forms whom, and whose were common with personal antecedents in non-restrictive clauses. Overall, the establishment of current usage is a result of many factors: that was used frequently in subject position, who allows a distinction between subjective and non-subjective forms, and who(m) and which can be preceded by a preposition (CHEL III, 295).




Animacy of the antecedent


When which developed as a relative pronoun it was used with either human and nonhuman antecedents.  The interrogative which still uses both today.   The most famous example of which being used with a personal antecedent is the sixteenth- and seventeenth-century example of The Lord’s Prayer, which opens with:


   (6)   “Our father, which art in heauen…”.


In Early Modern English, which was freely used with personal antecedents while who was used occasionally with non-personal antecedents. At this time, the possessive whose could refer to inanimate antecedents mainly because neither which nor that had a possessive form.  One-third of which usage occurs with a human antecedent in the sixteenth century, but by the seventeenth century, it is down to one-tenth (CHEL III, 294).  The growing change is a reflection of the systematization of the use of various grammatical forms in the course of the Early Modern English period and with the polite expression of Tudor and Stuart society, which probably emphasized the observation of the ‘personality’ of the referent (CHEL III, 294). 


Increasingly there was a move to PDE usage as which became increasingly confined to non-personal antecedents.   Therefore, in the seventeenth century, an animacy parameter is established and very much resembled PDE usage by the eighteenth century.  We should note, however, that a reaction against the ‘dehumanizing’ use of that with human antecedent took place after the end of the eighteenth century (Dekeyser 71).   The use of that with a personal antecedent would be replaced by the who form to decrease the functional load of that yet again.  Also, which was still being used with a personal referent by the uneducated at the end of the eighteenth century (CHEL IV, 277).  Even in educated circles, which was used rather than who for personal referents when the relative has a predicative function:


   (7)   To become a popular playwright, which Shakespeare certainly was, a man must             adapt his treatment of human life to the requirements of the stage

                                                            (1907 W.A. Raleigh, Shakespeare)

                                                            (CHEL IV, 277)


Written language vs. Spoken language


In Early Tudor times, that was more frequent than who and which combined.  By the sixteenth century, that becomes less frequent in formal prose.  However, it remains the commonest relative pronoun in informal and colloquial styles.  This can be seen in Elizabethan and Jacobean plays that try to give the illusion of contemporary speech, and especially those that are remote from courtliness (Barber 209).  It seems likely that the growing dominance of which was typical of a formal literary style.  In informal style, and probably in speech, that remained the dominant relative pronoun.  The trend continues in the seventeenth century; and in the Restoration period, which is entirely almost entirely confined to non-personal antecedents, and that is largely confined to restrictive clauses (Barber 210).


From 1700-1900, that was increasingly used less in written data than which (and who)—the change occurred rapidly during the eighteenth century and more gradually in the nineteenth century.  In the twentieth century, which is used slightly less than before (Ball 248-51).  Today, North American English prefers that to which in restrictive relative clauses even in written contexts (Biber et al. 1999: 353, 487-488, 616; qtd. in Nevalainen and Raumolin-Brunberg 206).


That or nothing


A zero relative is a variant of the that marker in restrictive clauses.  The zero relative is when acceptable when the function of the relative marker is an object of some kind within its clause (8).  Although it is discouraged in formal writing and entirely avoided in the King James Bible, it has picked up acceptance in the twentieth century, and has always been accepted in colloquial language. 


   (8)   The dog * we chased was fast.


According to C. M. Millward, omission of the relative pronoun did not occur in Old English (149), but Rissanen claims that both zero and that existed since some of the earliest written texts (English Corpus Linguistics 287).  Rissanen points to the Helsinki Corpus for proof and reveals that the changing distribution in usage of the two variant links indicates that the growing popularity of zero in late Middle English may have been inaccurately labeled an ‘omission’ of the expressed pronoun (275, 288).  By the fourteenth century, the zero relative was fairly common, especially when the relative would have been the subject of the subordinate clause (9).  Nonexpression of a relative that would have been an object is less frequent in Middle English, but does occur (10).  Note that this is a reversal of PDE usage in which nonexpression of an object relative is common, but nonexpression of a subject relative is considered incorrect. 


   (9)   he sente after a cherle was in the town (Chaucer)

          he went for a fellow (who) was in town


   (10)   the sorowe I suffred


Rissanen poses the primary question regarding the variation between that and zero in noun clauses: which one is the ‘original’ object clause link?  Since the zero relative is so often described as an ‘omission of that,’ there is an assumption that the expressed that was the predecessor.  Work by Jespersen suggests that the zero link was due to Scandinavian influence, but others point to fairly early Old English instances of zero (Rissanen 275).  The zero relative reached its peak in the late seventeenth century and lasted until the codifying grammarians of the eighteenth century began to once again favour that.


Usage Guides Today


Although the proper use of that and which is described in most grammar guides, there continue to be many “authoritative” texts that do not differentiate between them.  Tom MacArthur’s The Oxford Companion to the English Language (1992) states, “That can be used as a relative pronoun in place of who, whom, or which, except as complement of a preposition: the women who/that I rely on, but only the women on whom I rely” (860).  Similarly, Sidney Greenbaum’s An Introduction to English Grammar (1991) allows for indistinct changeability between that and which (11).  And as might be expected for simplicity’s sake, many online ESL lessons make no distinction either.


   (11)   The house which (or that or zero) we bought. (89)


   (12)   which refers to things; and that to either people or things. (61)


To contrast, some guides make the opposite choice of being too prescriptive.  For example, Harbrace Handbook for Canadians 5th ed. (1999) makes no distinction between restrictive and non-restrictive clauses; rather, it is more concerned with animacy (12).  Meanwhile, most Canadian usage guides make no distinction whatsoever.  The intended audience of a usage guide seems to have a direct influence on the specificity of proper usage.  It appears that while there is a prescriptive rule for the proper usage of that and which, the dearth of strict application has been increasingly represented in a leniency toward improper use and an acceptance of interchangeability.



Works Cited


***The Rissanen and Denison chapters in CHEL III and IV respectively provide a wealth of information and bibliographical resources on this topic. 


Survey texts:


Millward, C.M. A Biography of the English Language. Fort Worth: Harcourt Brace       Jovanovich College Publishers, 1994.


Historical Development:


Ball, C.  “A diachronic study of relative markers in spoken and written English.”             Language Variation and Change 8.2 (1996): 227-258.


Denison, David. “Syntax.” The Cambridge History of the English Language Vol. 4. Ed. Richard Hogg. New York: Cambridge University Press, 1992.


Nevalainen, Terttu, and Helena Raumolin-Brunberg. Historical Sociolinguistics. London:            Pearson Education, 2003.


Rissanen, Matti. “On the history of that/zero as object clause in links in English,” English            Corpus Linguistics. Eds. Karin Aijmer and Bengt Altenberg. New York:             Longman, 1991. 272-89.


Rissanen, Matti. “Syntax.” The Cambridge History of the English Language Vol. 3. Ed. Richard Hogg. New York: Cambridge University Press, 1992.


The Relative System in Middle English:


Mustanoja, Tauno F. A Middle English Syntax. Helsinki: Société néophilologique, 1960.


Romaine, Suzanne.  “Some historical and social dimensions of syntactic change in          Middle Scots relative clauses.” English historical linguistics: Studies in    development. Eds. N.F. Blake and Charles Jones. Sheffield: University of          Sheffield, 1984. 101-22.


The Relative Pronoun in Early Modern English:


Barber, Charles. Early Modern English. Edinburgh: Edinburgh University Press, 1997.


Dekeyser, Xavier. “Relativizers in Early Modern English: A dynamic quantitative study.”             Historical Syntax. Ed. Jacek Fisiak. Berlin: Mouton Publishers, 1984. 61-88.


Present Day Usage:


Guy, Gregory R., and Robert Bayley. “On the Choice of Relative Pronouns in English.”             American Speech: A Quarterly of Linguistic Usage 70.2 (1995): 148-62.


Usage Guides:


Greenbaum, Sidney.  An Introduction to English Grammar. Harlow: Longman, 1991.


Harbrace Handbook for Canadians. 5th ed. Toronto: Harcourt Brace & Company, 1999.


McArthur, Tom.  The Oxford Companion to the English Language. New York: Oxford             University Press, 1992.