"The three great causes of change in language," says Sayce, "may be briefly described as (1) imitation or analogy, (2) a wish to be clear and emphatic, and (3) laziness. Indeed, if we choose to go deep enough we might reduce all three causes to the general one of laziness, since it is easier to imitate than to say [Pg199] something new."[42] This tendency to take well-worn paths, paradoxically enough, is responsible both for the transfer of verbs from the strong to the weak declension, and for the transfer of certain others from the weak to the strong. A verb in everyday use tends almost inevitably to pull less familiar verbs with it, whether it be strong or weak. Thus /fed/ as the preterite of /to feed/ and /led/ as the preterite of /to lead/ paved the way for /pled/ as the preterite of /to plead/, and /rode/ as plainly performed the same office for /glode/, and /rung/ for /brung/, and /drove/ for /dove/ and /hove/, and /stole/ for /dole/, and /won/ for /skun/. Moreover, a familiar verb, itself acquiring a faulty inflection, may fasten a similar inflection upon another verb of like sound. Thus /het/, as the preterite of /to heat/, no doubt owes its existence to the example of /et/, the vulgar preterite of /to eat/. So far the irregular verbs. The same combination of laziness and imitativeness works toward the regularization of certain verbs that are historically irregular. In addition, of course, there is the fact that regularization is itself intrinsically simplification--that it makes the language easier. One sees the antagonistic pull of the two influences in the case of verbs ending in /-ow/. The analogy of /knew/ suggests /snew/ as the preterite of /to snow/, and it is sometimes encountered in the American vulgate. But the analogy of /snowed/ also suggests /knowed/, and the superior regularity of the form is enough to overcome the greater influence of /knew/ as a more familiar word than /snowed/. Thus /snew/ grows rare and is in decay, but /knowed/ shows vigor, and so do /growed/ and /throwed/. The substitution of /heerd/ for /heard/ also presents a case of logic and convenience supporting analogy. The form is suggested by /steered/, /feared/ and /cheered/, but its main advantage lies in the fact that it gets rid of a vowel change, always an impediment to easy speech. Here, as in the contrary direction, one barbarism breeds another. Thus /taken/, as the preterite of /to take/, has undoubtedly helped to make preterites of two other perfects, /shaken/ and /forsaken/.